ABSTRACT. Introduction: Using navigation for TKR should enable the surgeon to always achieve repeatable and accurate joint alignment and component placement. However, even with navigation, there are still outliers when compared to the currently considered ideal ranges. Therefore the goal of this study was to assess the reasons for outliers after navigated knee replacements using intra-operative navigation data and post-operative CT scans.

Methods: 209 CTs taken six weeks after navigated TKR as part of an ethical approved clinical trial were reviewed. All operations had been carried out by a consultant orthopaedic surgeon who used navigation as part of their routine practice for all TKRs and had been doing this for at least two years. Seven measurements were taken. Five were from the protocol according to Ewald et al. with outliers being greater than ±3° from ideal: coronal femorotibial angle (CFTA), coronal femoral angle (CFA), coronal tibial angle (CTA), sagittal femoral angle (SFA), sagittal tibial angle (STA). Two were from Berger’s protocol with outliers according to Bell et al. with femoral rotation (FR), being >3.9° internal rotation and tibial rotation (TR) being >23.8° internal rotation. All TKRs identified as an outlier from any of these measurements were then investigated in order to find reason(s) for the deviation outside the expected range. Operation notes were reviewed and intra-operative computer data were analysed to understand whether outliers were related to navigation system or not. CT scans were used to assess cement mantle consistency. They were also further reviewed to assess quality of scan and any issues with metal flashing.

Results: Outliers were: CFTA 14 patients (7%); CFA 33 patients (16%), 21<4°, 8<5°, 3<7°, 1>7°; SFA 18 patients (9%), 11<4°, 5<5°, 1<7°, 1>7°; CTA 8 patients (4%), STA 3 patients (1%); FR 37 patients (18%); TR 3 patients (1%). Identified reasons for being outliers were: surgeon’s decision to leave the patient out of range; surgeon’s decision to cut the distal femur or proximal tibia with few degrees of varus or valgus; imperfect extension gap balancing affecting the coronal alignment of the lower limb; heterogeneous cement mantle giving a difference between actual bone cuts and actual implant position; ankle registration error impacting tibial orientation errors; hip centre error (theoretically possible in less than 1%) impacting distal femoral cuts; loose femoral tracker responsible for significant femoral cut inaccuracy and arguable CT measurements due to intense metal flashing.

Discussion: Although this series had few outliers, there is still room to improve accuracy and precision of navigated TKR to reach 100% perfect alignment by refining technical tools involved in the technology.

ABSTRACT. Introduction: Total knee arthroplasty in recurvatum deformity is challenging. The incidence is too low for studying. The technique for management is controversial.

Methods: Retrospectively study reviewed the records of 127 computer-assisted total knee arthroplasty (CAS-TKA) that had pre-op recurvatum deformity and analyzed the data about gap patterns and outcomes. TKA was performed by modified gap technique.

Results: 127 from the 790 (16.1%) CAS-TKA had pre-op recurvatum deformity. Grade 6-10o were 90 cases and >10o were 37 cases. The extension gap > flexion gap = 69 cases (54%) and the degree of recurvatum deformity has no significant correlation with the gap pattern. The mean pre-operative hyperextension deformity of 9.43 (6-25o), as measured by navigation of the start of the operation, improved to mean extension 1.63 (flexion 2o-extension7o), the mean knee flexion improved from 124.67o (86-139) pre-op. to 128.91 (94-150), none had mediolateral laxity > 2mm from the mechanical axis of 0o at the end of surgery. At the mean follow-up of 25.5 months (4-87) there was significant improvement in the mean Oxford knee and Western Ontario and McMaster Universities Osteoarthritis Index scores and no recurrent recurvatum.

Discussion: 54% from recurvatum knees had extension gap wider than flexion gap. Degrees of extension cannot predict pattern of gap. All cases of recurvatum deformity were recommended to measure gap.

ABSTRACT. Introduction: One of four normal people has mechanical alignment of 3 degrees varus and more that so-called “constitutional varus”. Parallel joint line to the floor can be found in both neutral and varus alignment. Therefore, joint line orientation may play an important role in clinical outcomes after TKA. Reconstituting joint line parallel to the floor advocated by 3° varus tibial cut introduced by Hungerford et al.

Methods: This study attempts to compare the clinical outcomes from difference radiographic parameter. The prospective study was conducted on 106 primary varus osteoarthritis knees undergone CAS TKA with either classical method (58 knees) or anatomical method (48 knees). Clinical outcomes, consisting of WOMAC scores, Oxford knee scores and ROM, were preoperatively and 1 year-postoperatively evaluated. Hip-knee-ankle with full leg standing was measured in terms of mechanical axis, tibial cut angle and tibial joint line angle after receiving surgery for 1 year.

Results: The results revealed that postoperative neutral alignment (mechanical axis 0± 3°), 4-5°varus and ≥6°varus showed no significant difference between WOMAC scores, Oxford scores and ROM. Moreover, the comparison of classical tibial cut and anatomical tibial cut there was also no significant difference in clinical outcomes between postoperative joint line parallel to the floor and oblique joint line. Nevertheless, joint line parallel to the floor had a significant WOMAC scores and ROM improvement than oblique joint line.

Discussion: In conclusion, the joint line parallel to the floor is one of the key success factors after TKA.

ABSTRACT. Introduction: The “correct” rotational aligment of femoral component and the “normal” rotational alignment of distal femur may not be the same position. Because the natural tibial plateau has average 3° varus. Whereas TKA with classical method make tibial cut perpendicularly to mechanical axis. Consequently, femoral component has to rotate externally 3° for compensation tibial cut. “Anatomical method” make tibial cut in 3° varus then the posterior femoral cut will be parallel to the posterior condylar axis and keep femoral component in natural position. This study aims to compare component rotaion between anatomical and classical method in CAS TKA.

Methods: The study conducted on 56 CAS TKAs with modified gap technique in varus osteoarthritis knees. Intraoperative femoral component rotation with reference to posterior condylar axis was recorded from navigation. Postoperative hip-knee-ankle scanography and CT scans were performed. The knees were defined as anatomical group (tibial cut = 0-1°) and classical group (tibial cut ≧ 2°). Posterior condylar angle and condylar twist angle were measured from CT.

Results: Classical group was 23 knees and anatomical group was 33 knees. Classical group had more intraoperative femoral component rotation externally than anatomical group (3.13° ± 1.05° vs 1.0° ± 1.11°, p =0.00). There were no difference between two group in terms of posterior condylar angle and condylar twist angle. Both methods had no differnce in percentage of knees to align within optimal range in Thai population.

Discussion: This study found that antomical method reduced femoral component rotation intraoperatively. However, postoperative femoral rotation had no difference between two methods.

ABSTRACT. Introduction: Restoration of the posterior femoral offset (PFO) is required after total knee arthroplasty (TKA). We hypothesized that the measurement by pre- and post-operative conventional radiographs of femoral offset is imprecise when compared to intraoperative measurement reference by a navigation system.

Methods: 150 TKAs have been analyzed. The pre- and postoperative radiological measurements were performed on lateral X-ray by measuring the distance between the tangent to the distal 10 cm of the anterior cortex of the femur and the tangent to the most posterior point of the native or prosthetic femoral condyles. Intraoperative measurements were performed using the navigation software by direct palpation of the same points. The actual change of the offset (resection thickness - thickness of the implant) was compared to the radiologic variation. Both measurements were compared using a paired Wilcoxon test and the calculation of the Spearman correlation coefficient with a significance level of 5%.

Results: The mean paired difference between the pre-operative actual and measured offsets was 3.9 mm (p <0.001). There was a moderate correlation between the pre-operative measurements (R² = 0.41, p <0.001). The mean paired difference between the post-operative actual and measured offsets was 6.3 mm (p <0.001). There was a weak correlation between the post-operative measurements (R² = 0.27, p <0.001).

Discussion: The conventional radiologic measurement technique of the PFO is unreliable and cannot be used either for pre-TKA planning of the posterior femoral resection and antero-posterior sizing of the femoral component, or for post-TKA quality control of the reconstruction of the PFO.

ABSTRACT. Introduction: Computer-assisted orthopaedic surgery (CAOS) improves accuracy and precision of the bony resections in total knee arthroplasty (TKA). However, reservation may exist among new CAOS users regarding the ability of achieving the same results as the experienced users can produce. This study investigated if there are immediate benefits in achieving surgical goals for the novice surgeons, as compared to the experienced surgeons, by using a contemporary CAOS system.

Methods: Two groups of surgeons using a contemporary CAOS system were evaluated, including: 1) “novice users”: 7 surgeons with no CAOS experience prior to using the system studied, and had only performed ≤20 CAOS TKAs; 2) “experienced users”: 6 surgeons, each had performed >150 TKAs using the CAOS system. The “novice group” was further divided into established surgeons with extensive experience in conventional TKA (novice-senior), and junior surgeons who were less experienced (novice-junior) to investigate intra-group variability. All the surgeries by the “novice users” (novice group: 86 cases) and the most recent 20 cases from each of the “experienced users” (experienced group, 120 cases) were collected. The intraoperative resection goal (planned resection) and the actual resection (checked resection) were compared within the novice group (novice-senior vs novice-junior), as well as between the novice and experience groups.

Results: All groups achieved accurate and precise TKA resections. No differences were found between the senior and the junior surgeons within the novice group, and between the experienced and novice groups, except for that the novice users resected slightly more but clinically irrelevant bone in the tibia. The experienced and novice groups had comparable, high percentages of the resections that were identified as optimal (<2°/mm deviation) and acceptable (<3°/mm deviation).

Discussions: This study demonstrated that regardless of the surgeon’s experience with TKA in general, new adoption of the CAOS system investigated can immediately benefit the accuracy and precision of the bony resections at a comparable level with experience CAOS users. The CAOS system offers substantial reduction of the outliers compared to TKAs performed with conventional instruments.

ABSTRACT. Femoroplasty is suggested as a therapeutic approach that aims to reducing the potential for hip fracture by injecting bone cement into osteoporotic hips (Beckmann 2007, Heini 2004, Sutter 2010). Previous work showed that successful and safe femoroplasty need to be rigorously planned to achieve sufficient strengthening without excessive injection of bone cement that may create other potential complications such as osteonecrosis and fracture in the junction of bone and cement (Basafa 2014, Basafa 2013, Basafa 2009, Basafa 2015). We have previously developed a computer-assisted femoroplasty workstation that utilized an optical tracker to help navigate the injection (Otake 2009, Otake 2010). Experimental testing, however, showed that the success of femoroplasty can be dependent on controlled and accurate intraoperative execution of the plan. This paper presents our attempts towards the development of a robot-assisted femoroplasty workstation to help achieve a controlled injection that closely follows the preoperative plan.

ABSTRACT. Introduction: Statistical shape model (SSM) has been used to represent the pathological variations and progression of the hip joint. In construction of SSM, shape correspondence between the healthy shapes (in other words, a group of shapes without pathological variation) and the diseased shapes (with pathological variation) is necessary. In this paper, we aim to improve shape correspondence between healthy shapes and severely diseased shapes of pelvis by manually identifying several corresponding anatomical landmarks based on the expert knowledge about the pathological progression.

Methods: In this paper, we used CT data of 24 preoperative (diseased) pelvises of the patients who underwent total hip arthoroplasty (THA). In the proposed shape correspondence method, the group-wise registration with additional manually identified anatomical landmarks is performed to embed expert knowledges about pathological progression for an improved accuracy.

Results: Three SSMs were constructed from the training dataset of 24 diseased pelvis and compared. In our proposed method, SSM correctly represented acetabular translation. In segmentation accuracy, our proposed method was significantly more accurate compared with the conventional method. These results show that the proposed group-wise registration with manually identified anatomical landmarks was effective.

Discussion: We have described a method to improve accuracy of the shape correspondence between healthy shapes and diseased shapes by adding manually identified anatomical landmarks to a conventional non-rigid registration. By initializing the registration with manually provided point correspondence at around the acetabular anatomical landmarks based on the knowledge about the pathological progression, the mode variation represented the pathological progression better than the conventional method potentially owing to the improved accuracy in shape correspondence. We also confirmed an improvement segmentation accuracy when the proposed method was employed compared with conventional method. Future works include (1) analysis, modeling and prediction of the pathological progression and (2) automatic classification of the disease categories.

ABSTRACT. Introduction: We previously reported about computer simulation (CS) system of elbow osteoarthritis (OA) to detect impinging osteophyte preoperatively by using three dimensional (3-D) bone models of patients. To apply a navigation system to arthroscopic surgery based on CS, we evaluated registration accuracy of humerus and ulna to use a resin bone model.

Methods: We use CT data of five normal elbows and five elbows of OA. Based on CT data, we created 3-D bone model. We marked 9 reference points on bone surface. 5 points are on humerus and 4 points are on ulna to evaluate registration accuracy of the navigation. Based on these marked 3-D bone data, we created a resin bone model. We imported the data of 3-D bone model to the navigation system and planed 6 anatomical registration points on humerus and 5 points on ulna. Point to point registration (PPR) and surface matching registration (SMR) was performed. We pointed 6 or 5 anatomical registration points by a pointer device and performed PPR. Subsequently we performed SMR to point 30 points on bone model as follows. For humerus, we pointed 6 points on lateral shaft of near from tracker, 4 points on medial and 4 points on lateral epicondyle. Another 16 points were divided anterior and posterior intra-articular area avoid cartilage for the set of AP, pointed on anterior for the set of A and pointed on posterior for the set of P. For ulna, we pointed 6 extra-articular points and 24 intra-articular points avoid cartilage. After registration, we point the reference points of ”marked” reference point on bone model and recorded the coordinate position. We evaluated target registration error (TRE) to measure the distance of the coordinate position between planned and navigated reference points.

Results: TRE of humerus (the set of points of AP/ A/ P) were (0.82/ 0.80 /0.94 mm) in humerus of OA and (0.97/ 1.06/ 0.99 mm) in normal humerus. TRE of ulna was 1.77mm in ulna of OA and 1.75mm in normal ulna.

Discussion: TRE of humerus were acceptable both OA and normal humerus to point only 30 points on intra-articular and distinct extra-articular surface. TRE of ulna was not acceptable because we could point only small intra-articular area of coronoid process and olecranon uncovered with cartilage. We will apply the navigation system to arthroscopic osteophyte resection of humerus with CS based on the result of this study. It helps to perform more accurate and efficient osteophyte resection easily.

ABSTRACT. Introduction: Alignment of the bones in standing position provides useful information in surgical planning. For example, in total hip arthroplasty (THA), pelvic sagittal inclination (PSI) in standing position is an important factor in planning of cup alignment. Some landmark-based manual methods to measure PSI was proposed. On the other hand, 2D-3D registration has been extensively explored to measure 3D alignment using a 2D x-ray and 3D CT images, however, the existing algorithms require manual input. We propose a system to conduct these measurements fully automatically by combining a robust bone segmentation and an intensity-based 2D-3D registration algorithms for elimination of inter- and intra-operator variability and for the studies on datasets of a large population.

Methods: In this study, we analysed a set of CT images taken in the supine position and radiographs taken in the standing position of 475 patients who underwent hip surgery. First, we segment pelvis and femora and identify anatomical landmarks in CT images by an automated segmentation method that we previously proposed. Then, an intensity-based 2D-3D registration of the pelvis and femora on the CT and the radiographs is performed.　The proposed registration method uses 100 “gold standard registration” dataset, which was computed with carefully selected manual initialization and its result was verified by an experienced surgeon. We compute an average pose as an initialization in the registration of an unknown subject. PSIs in the supine position and the standing position were measured in 475 cases by the proposed method. As an evaluation, the results of 100 cases were compared with the previously proposed manual method.

Results: The pelvis in the standing position was posteriorly inclined by 6.7±4.8 [deg] (minimum -8.5 [deg]. maximum 39.2 [deg]. ) compared to the supine position. And the average difference of PSI measurements between the proposed method and the manual method was 1.57 [deg] (correlation coefficient: 0.98).

Discussion: This result suggests that the proposed method achieved measurements comparable to the conventional manual method while eliminating inter- and intra-operator variability and allowing analysis on a large population dataset.

ABSTRACT. Introduction: The aim of this study is to evaluate a new image guided navigation system, Stryker ADAPT, compared to normal fluoroscopic guidance in gamma nail surgery. The unique feature of the system is that it analyzes the fluoroscopic images for implant position as well as lag screw length and trajectory. Potentially, this can help to place the lag screw more effectively below a Tip-Apex-Distance (TAD) of 25mm which has shown to reduce cut-out and implant failure.

Methods: At our level one trauma center, we included patients presenting with intertrochanteric fractures type 31A1 and 31A2. If exclusion criteria were not present, patient consent was obtained and randomization result was disclosed in the OR. The ADAPT group was operated with the navigation screen helping the surgeon to place the implants; in the control group, the screen was not visible. All images and times were recorded automatically and evaluated postoperatively.

Results: The Control group showed a significantly higher number of drilling attempts (mean 4.6 and STD 2.11 vs. mean 1.57 and STD 1.8; p=0.001) as well as a significantly higher failure rate to attain a TAD below 25mm (see Table 1, p=0.0244) while shorter TAD for the navigated group failed to reach significance (mean 20.56 and STD 4.56 vs. mean 25.88 and STD 7.13, p=0.062). Operation time showed a non-significant tendency towards shorter operation time (p=0.238).

Discussion: Our preliminary results show a better performance reaching TAD criteria with the help of the ADAPT system while not prolonging operation time and needing less drilling attempts, potentially reducing invasiveness and improving effectiveness of the procedure.

ABSTRACT. Introduction: Thornberry et al. proposed the hip center anterior pelvic plane (HCAPP) - formed by connecting the two anterior superior pelvic spines and the center of the involved hip – as the reference for cup alignment because HCAPP is more accurate for registration in the lateral position than anterior pelvic plane (APP). To apply this method widely, we studied the relationship between APP and HCAPP in Japanese with normal pelvis.

Methods: CT images of 86 normal hip joints from 43 healthy female volunteers with an average age of 71 years (range, 57–84 years) were obtained from our CT image database according to the following criteria: no hip symptoms, no signs of OA or acetabular dysplasia, and no previous hip surgery. We measured a sagittal angle between APP and HCAPP.

Results: The mean angle between APP and HCAPP was 40° ± 3.8° (range, 30° to 48°) in the right side and 40° ± 4.1° (range, 24° to 49°) in the left side. The mean difference of the angle between the right and left sides was 1.7° ± 1.3° (range, 0° to 5.8°). The mean angle in men was 40° ± 3.5° (range, 33° to 49°) and the mean angle in women was 40° ± 4.3° (range, 24° to 48°). There were no significant differences in the angle between right and left (P = 0.643) or between men and women (P = 0.635).

Discussion: In the study of Thornberry et al, the mean angle between APP and HCAPP was 37°± 3.5 ° and the mean difference between the right and left sides was 1.8°± 1.0°. Compared with our result, the mean angle between APP and HCAPP in Japanese was 3° larger than those in Caucasian and African American, but the standard deviation was comparable. The variability of the angle between APP and HCAPP in Japanese was also comparable with Caucasian and African American. These suggested that HCAPP can be reliably used as an anatomical coordinate system of pelvis for cup alignment.

The sagittal angle between APP and HCAPP in Japanese showed 3° larger than those previously reported in Caucasian and African American, but the variability of the angle was comparable.

ABSTRACT. Introduction: The purpose of this study was to examine relationship between gleno-humeral articulation and scapulo-thoracic articulation evaluated by scapular motion analyses during scapular plane abduction in patients with reverse shoulder arthroplasty (RSA).

Methods: We evaluated 5 shoulders in 5 patients (3 male, 2 female) who had received RSA with minimum six months follow up. The fluoroscopic images were obtained during scapular plane abduction at 0, 30, 60 and 90 degrees. CT derived bone images were matched semi-automatically with the silhouette of the bones at fluoroscopic images by GANESHA which was originally invented in our institute.

Results: During patients actively elevate their arm at 30, 60 and 90 degrees of abduction in the scapular plane, average scapular upward rotation angle was 14.7, 27.3 and 37 degrees respectively. At six months postoperative follow up, average abduction angle was 104 degrees. We divided into two groups: abduction angle more than 104 degrees (Good result group) and abduction angle less than 104 degrees (Poor result group). In good result group, average scapular upward rotation angle was 7.2, 14.4 and 31.5 degrees respectively. In poor result group, average scapular upward rotation angle was 19.8, 36.0 and 44.9 degrees respectively. At 30 and 60 degrees of abduction, average scapular upward rotation angle in good result group were significantly lower than in poor result group.

Discussion: Three dimensional motion analyses of the scapula after RSA revealed that relatively less scapula-thoracic articulation might be necessary for acquiring good shoulder abduction motion

ABSTRACT. Introduction: In Canada, X-ray devices are operated by medical radiation technologists (MRTs), but a substantial portion of their early training in using these devices is done without acquiring radiographs of human subjects because of well-founded concerns related to the health risks of radiation exposure. During MRT training, instructors evaluate patient positioning by visual inspection and coaching, but since no X-ray images are obtained, students do not get immediate feedback on how patient repositioning affects the resulting image. In this paper, we describe a prototype system that combines real-time tracking of a live patient stand-in with synthetic X-ray images generated on-the-fly as the patient stand-in is repositioned by the MRT. Our goal is for this system to be used by MRT instructors to facilitate instructional coaching, provide quantitative feedback regarding image composition and quality, and better prepare students for their initial clinical experiences.

Methods: To generate virtual radiographs, our system uses a Kinect V2 for real-time joint tracking data, we use to position two pre-segmented upper and lower arm CT subvolumes in 3D space to match the real-time limb segment coordinate systems. These two aligned subvolumes are then used to generate two digitally reconstructed radiographs (DRR), which are rendered onto a single plane as a DRR composite. Additionally, in a pilot study aimed at illustrating the system’s capabilities to measure joint angles, we compared the degree of discrepancy between by our system against a mechanical goniometer.

Results: Our initial results show that the discrepancies are modest - on the order of 4±5.5° - and that the composite simulated images provide a visually reasonable approximation of the changes in the joint’s appearance as the joint angle changes.

Discussion: In the future, we plan to present our system, in which students can simultaneously position a live patient and view a realistic radiograph, to MRT instructors for their continued input to determine how useful such a system would be in their training programs. We hypothesize that adding this integrative modality to the MRT training process will enable students to more quickly and reliably obtain clinically acceptable images in real life.

ABSTRACT. Introduction: In order to elucidate the accuracy of patient specific surgical guide (PSG) for pelvis osteotomy setting comparing between the preoperative planning and PSG setting, and to elucidate the accuracy of modified PSG comparing between the preoperative planning and final procedure in rotational acetabular osteotomy (RAO), we performed experimental study using fresh cadaver samples.

Methods: 20 hips in 10 fresh cadaveric samples were imaged from the whole pelvis to the femoral condyles using (first-time) helical CT. As the preoperative planning, the osteotomy line was decided and the acetabular fragment was rotated laterally to acquire more than 40 in lateral center-edge angle. PSG for RAO was designed based on the preoperative CT data using image-processing software. All PSGs were made from resins and produced by a machine. PSG consisted of osteotomy guide part, rotation distance guide part, and rotation angle guide part. RAO with PSG was performed in 10 hips (PSG group) while RAO without surgical guide was performed in other 10 hips (manual group). After PSG for RAO was fixed screws on the pelvis bone surface, second-time CT was imaged for validation assessment between the preoperative planning and PSG setting in 10 PSG group hips. After RAO procedure, which included osteotomy, rotational motion, and screw fixation of the acetabular fragment, third-time CT was imaged for validation assessment between the preoperative planning and final procedure in all 20 hips.

Results: The absolute error between the preoperative planning and PSG setting for RAO was 2.4±0.8 for the abduction angle, 3.6±1.7 for the flexion angle, and 1.6±0.9 mm for the center of the osteotomy sphere in PSG group. The absolute errors between the preoperative planning and final procedure were smallest in PSG-2 group (2 types of rotation guide use), followed by PSG-1 group (rotation distance guide part use) and manual group (PSG-2 group / PSG-1 group / manual group; 1.3±1.3/ 4.1±0.5 / 6.2±3.1 for the abduction angle, 1.8±1.0/ 1.9±1.7 / 7.5±6.4 for the flexion angle, and 1.9±1.9mm/ 4.1±0.9mm / 4.1±1.3mm for the center of the osteotomy sphere).

Discussion: PSG-2 for pelvic osteotomy showed good accuracy in PSG setting and final procedure.

ABSTRACT. An efficient registration named “12 visual points” is proposed for improving image projection accuracy of intra- operative orthopaedic surgeries by using ARCAGO system. The method is first to generate a virtual space with a 3D model of the patient from CT images while metal markers attached and using 12 points for camera calibration. The scanned 3D model is then used to fit the poses of the patient during surgery. The patient’s preoperative 3-D image model is registered by not only projecting it onto the patient’s torso but also synthetic 3-D model merges shown into wearable glasses with the real patient image, enabling the surgeon to see through the patient’s anatomy. Experimental results validate on the dummy model and clinical trials, and shown the effectiveness of the proposed method in the operating theater.

ABSTRACT. Introduction: Regarding correction of the knee alignments in Total knee arthroplasty TKA, surgeons intraoperatively check whether the patients can regain extension of 0 degrees by straight leg raising. However, since this checking procedure of knee extension is often misjudged due to surgical drape and/or is affected by anesthesia, even immediate sagittal alignment after TKA might not be able to show the extension angle which was measured intraoperatively. Although three-dimensional (3D) alignments of knee joint cannot be record conventionally, 3D scanner might solve this issue. The purpose of this study was to investigate whether intraoperatively measured extension angle differs from immediately postoperative extension angle using 3D models with 3D scanner.

Methods: Eleven patients with an average age of 71.1 years (range: 56–86 years) who underwent posterior cruciate ligament sacrificed TKA were enrolled in this study. After placement of total knee implants, operative lower limb was elevated by single examiner (one author), then was scanned with a low-cost handy 3D scanner (Kinect, Microsoft) from patient’s lateral side. 3D model of the operative lower limb was made by a combining process (KScan3D) with the scanned data. Intraoperative and immediate extension angles between femoral and tibial axes were measured.

Results: The intraoperative extension angle of the operative lower limb was -1.9 degrees. While, the immediate postoperative extension angle was -10.3 degrees. There is a statistically significant difference between the two extension angles of the operative lower limb (mean -8.4 degrees).

Discussion: We clarified that the knee extension of patients who have complaint about knee disease can be regained intraoperatively by regular TKA with the similar level of young healthy subjects. However, as the authors thought before starting this study, the intraoperative extension angle was slightly but significantly reduced immediately after TKA. Surgeons should be aware of this change between intraoperative and immediately postoperative knee status even if the knee extension might be improved postoperatively with rehabilitation.

ACKNOWLEDGMENTS This work was supported by JSPS KAKENHI Grant Number 26870724.

ABSTRACT. Introduction: We conduct the study using CAS to measure gap and angular change after sequentially release of posteromedial (PM) and superficial MCL (sMCL) for correction of varus OA knee in OA knee patients.

Methods: We prospectively collect data from patients whom underwent CAS TKA from 1 November 2015 to 28 February 2016. After arthrotomy and assemble of CAS system, we resect proximal tibia, inserted spring-loaded device and measured angular deformity and gap. We recorded data at initial position, 30, 60,90 and 120 degree respectively. Then we performed PM release and record data. Then if the residual varus deformity was more than 3 degrees, we performed sMCL release without detachment of pes anserinus and record data. We reported data using mean +/- standard deviation at each angle point.

Results: There were 22 patients (16 female 6 male) with mean age of 67.1+/-8.35 years. Mean pre-operative deformity was varus 8.11+/- 3.39 degrees with flexion contracture of 11.3 +/- 5.34 degrees. The mean correction of varus after PM release and sMCL release was 4.88+/- 2.82 and 3.39 +/- 1.7 degrees respectively. While the mean correction of flexion contracture after PM release and sMCL release was 5.57+/- 3.5 and 1.34+/-2.9 degrees respectively. The mean medial gap change on extension after PM and sMCL release was 1.83+/-1.39 and 1.67+/-1.04 mm. The mean medial gap change at 90 degree flexion after PM and sMCL release was 0.73+/-0.9 and 5.14+/-2.11 mm. There was no instability found in this case series.

Discussion: The data showed PM release has effect on varus correction on both gaps, with more effect on extension gap. The sMCL release also has effect on varus correction of both gaps, with more effect on flexion gap. Since there was no severe instability occurred after sMCL release as previously reported, we believe sMCL release with this technique could create stable TKA.

ABSTRACT. Introduction: The performance of implant placement methods for total knee arthroplasty are often evaluated using inaccurate measurement methods which may lead to erroneous conclusions. In this study we present and evaluate a method based on pre-operative and post-operative CT imaging that allows accurate measurement of the position and orientation of TKA components.

Methods: 3D CT models of bone and implants were generated using Mimics software (Materialise, Leuven, Belgium) then post-operative 3D CT models were registered on their respective pre-operative counterparts and implant models were moved along. We assessed the accuracy of the method using synthetic bones on which TKA were performed Optical scans and tantalum markers were used as ground truth to assess the reliability of the method. Three operators processed three sets of imaging. The proposed method can be described as a two step method where on the one hand implant models are registered on 3D CT models and on the other hand 3D CT models of long bone are registered on one another. Segmentation and registration errors can occur at each step. Implant registration error, bone to bone registration error and total error were measured in anatomical coordinate systems defined on the pre-operative bones.

Results & Discussion: Except for the proximal-distal direction, the mean total error was inferior to 0.3mm and 0.5°. Implant migration is suspected to have caused the higher error in the proximal-distal direction. We recommend our method to accurately assess TKA component position and orientation independently of the implant placement method or coordinate system used.

ABSTRACT. Introduction: Determination of hip center is usually done radiographically, which may not be practical during the surgery, and kinematic method is another way to determine the hip center in real time. During kinematic assessment, the thigh is moved passively in all direction, and the pivoting point is considered as the hip center. In this fresh frozen cadaver study, we attempted to find the optimal method for determining the hip center kinematically in terms of the hip motion.

Methods: The physical hip center was initially determined by painting the optical probe on the surface of femoral head, which was exposed via anterior arthrotomy, and this was considered as the real hip center. Optical infra-red skeletal markers were placed over the iliac crest and distal femur, and three types of passive motion were carried out over the hip joint, including flexion/extension (FE), abduction/adduction (AA), and circumduction (CIR). The hip centers were calculated through the sphere fitting method, and the results were then compared to the physical hip center.

Reults: Three sets of data were collected in each experiment, and the average value was obtained. The radius of the femoral head determined with the optical probe was 21.7mm, and the bone marker was placed at a distance of 265mm from the hip center. As the final result, the average distance error from the physical hip center to FE center was 1.3mm, to AA center was 8.2mm, and to CIR center was 1.8mm. The average motion arc was 57.1o for FE, 38o for AA, and 42.2o for CIR. When evaluating the relationship between the accuracy of data and the range of motion arc, higher accuracy was found with larger motion arc. The data reached to a consistent value if the motion arc is > 30o.

Discussion: In this study, we found that FE and CIR motion led to the most accurate results, and the larger the motion arc, the more precise the data could be. AA motion had the largest distance error from the physical hip center, which could be an interesting biomechanical phenomenon during such motion. A possible explanation could be that the femoral head might move laterally during abduction of hip joint. This is a fresh frozen cadaver study, and the true femoral head center was obtained as the reference point for comparison, therefore, we have the confidence to conclude that flexion/extension or circumduction motion of the hip joint is recommended when determining the center of femoral head via kinematic method.

ABSTRACT. Introduction: Femoral tunnel position during anterior cruciate ligament (ACL) reconstruction is known as a critical factor that affect clinical outcome.

Methods: For quantitative measurement, we developed new evaluation method for ACL femoral tunnel position based on approximate circle of femoral posterior condyle. The centre of approximate circle of posterior condyle (O) and posterior end of Blumensaat’s line (B) was defined on lateral view of femoral condyle. The tunnel position (A) was measured as an angle between OB and OA, and distance between OA was normalized at a radius of posterior condyle. 60 primary ACL reconstructed patients (average age; 22.8, male26, female34) were included. The guide wire for anteromedial (AM) tunnel was inserted using outside-in technique, and directed within anatomical femoral footprint. True lateral knee images were obtained by fluoroscopy, and tip of the guide wire was measured as AM femoral tunnel position.

Results: The angle between OB and OA was 44.9(±10.5) degrees. The distance between OA was 57.9(±10.5)%. The line connecting O and B was nearly parallel to the distal femoral shaft, and was appropriate for measurement baseline. B was also identifiable to the top of femoral condylar notch during arthroscopic surgery. The femoral tunnel position was clinically expressed as ‘High-Low’ and ‘Deep-Shallow’.

Discussion: Using new evaluation method, `High-Low’ position was quantitatively measured as an angle between OB and OA. ‘Deep-Shallow’ position was also measured as a distance between OA. Therefore, new evaluation method was scopic surgeon friendly, and was applicable during surgery.

ABSTRACT. Introduction: Open reduction and internal fixation is the standard treatment of dislocated calcaneus fractures. The visualization of the reduction and implant placement is limited in conventional intraoperative 2D-imaging. Therefore 3D-imaging is additionally used to detect insufficient reduction or implant malpositioning. To reduce the time adjusting the standard planes and detecting the screws would improve the surgical procedure. The aim of the study was to develop a software prototype that automatically detects screws and standard planes in 3D-imaging for the intraoperative treatment of calcaneus fractures.

Methods: The intraoperative 3D-images of 50 patients after reduction and internal plate fixation of calcaneal fractures were analyzed. An atlas-based feature registration was developed for the automatic standard plane detection. As standard planes the calcaneo-cuboidal and the posterior facet of the subtalar joint line were determined. Analyzed were the rates of correct detected automatic standard planes in comparision to the adjustment of the planes by three experienced surgeons. In addition an algorithm for an automatic screw detection was applied and validated by manually selected screws. Measured were the amount of correct detected screws and additionally the variance to the actual screw axis in degrees and to the screw tip in millimeter.

Results: The automatic detection of the standard planes of the calcaneus was achieved in 90% of the cases. The time for the automatic detection averaged 5 seconds in comparison to 49 seconds in average the three experienced surgeons needed. The 50 3D-images contained in total 309 screws. In 96,1% the screws were detected correctly. The variance of the screw axis averaged 2,1 ± 1,8 degrees and of the screw tip 1,0 ± 1,2 millimeters. The mean time for the detection was 4,57 ± 0,55 seconds.

Discussion: The detection of the standard planes and the screws intraoperatively is time consuming. An automatic detection of the standard planes and screws during intraoperative 3D-imaging of the calcaneus could significantly improve the procedure. This study shows promising results simplifying the intraoperative 3D visualization during challenging osteosyntheses by automatically detecting standard planes and positioned screws.

ABSTRACT. Introduction: Hemophilia A is caused by a deficiency of clotting factor VIII. The intra-articular changes in hemophilic arthropathy are characterised by synovial proliferation and destruction of articular cartilage caused by repeated episodes of hemarthrosis. With continued synovitis, joint-space narrowing occurs, followed by progressive joint-space loss, articular cartilage destruction, joint incongruity, and ankylosis of the joint. For hemophilic patients with angular extra-articular deformities of knee joint, intramedullary guiding may be impossible during conventional TKA. Navigation systems including image-based and image-free system have been used as an adjunct for TKA to improve alignment of the lesion site.

Methods: A 37-year-old man with hemophilia A, who had the history of right femur and tibia fractures with extra-articular deformity. He also had chronic osteomyelitis of right femur status post sequestrectomy with retained intramedullary cement beads. Due to the ankylosis and stiffness of the right knee joint, he underwent computer assisted TKA on 2013-09-03.

Results: The anatomical axis was improved from valgus 12 degrees to valgus 7 degrees. The mechanical axis was improved from valgus 6 degrees to valgus 1 degree. Range of motion was improved from 0 degrees to 90 degrees.

Discussion: In this case report, we demonstrated that navigation system is feasible and satisfactory for severe osteoarthritis of the knee with hemophilia and extra-articular deformity, which accurate bone cuts and rotational alignment cannot be accomplished using conventional TKA instrumentation. We suggested that navigation system without intramedullary reaming is a potent strategy in the TKA surgery for the knee with hemophilia and extra-articular deformity presumably with the advantages of avoiding excess blood loss, bone cut and proper soft tissue balancing.

ABSTRACT. Introduction: For the diagnosis of knee diseases radiographs are considered as gold standard for the assessment of joint gap and alignment. Although, the imaging is performed commonly under weight-bearing (WB) conditions it is limited to 2D. Alternatively, computed tomography (CT) and magnet resonance imaging (MRI) provide 3D information but they are acquired in the relaxed supine position and under non-weight-bearing (NWB) conditions. This means, if references, such as the mechanical leg axis, orientation of the patella or rotational knee deformities are derived for diagnostics and therapy in the supine NWB situation this might not adequately reflect the real WB situation. The goal of the study was to quantify the differences of knee joint alignment between the NWB and the WB situation using MRI.

Methods: The assessment relied on upright data, which allowed the assessment of knee kinematics under physiological loading conditions in single legged stance. For this purpose, supine and upright WB MRIs have been obtained from 10 healthy volunteers. In order to evaluate the changes of static NWB to WB a clinical expert identified anatomical landmarks as a basis for the subsequent alignment calculations. The calculations of the tibiofemoral and patellofemoral alignments in terms of overall 12 clinical parameters have been carried out for all 10 volunteers for the unloaded and the loaded case to quantify the changes.

Results: For the tibiofemoral joint alignment, the lowest changes from NWB to WB could be observed for anterior/posterior shifts (≤ 3 mm, 1 outlier) and medial/lateral (≤ 3 mm, 1 outlier). Varus/valgus alterations (≤ 3°, 1 outlier) were less pronounced, but lay within a range that is relevant for prostheses. The largest changes could be found for flexion/extension (50 % ≥ 6°) and for internal/external rotation (50 % ≥ 5°). For the patellofemoral joint alignment, the lowest changes from the unloaded to the loaded case could be observed for anterior/posterior shifts (80 % ≤ 2 mm), flexion/extension (80 % ≤ 4°) and medial/lateral tilt (≤ 3°, 1 outlier). The largest and most uncertain changes could be found for medial/lateral shifts (80 % ≥ 3 mm) and for internal/external rotations (large fluctuations).

Discussion: Thus, when taking medical images for clinical evaluation or pre-operative planning one must be aware of the fact that knee joint alignment differs substantially between unloaded and loaded situation.

ABSTRACT. Introduction: In total knee arthroplasty patient specific parameters such as implant design, alignment and soft-tissue management are major factors under consideration to improve longevity and patient-satisfaction. However, for an optimal implant design, especially in the case of deformities, a systematic description of morphological knee joint parameters and a study of their effects on the individual biomechanical situation are essential. The aim of this study was to identify morphological parameters and to develop and evaluate a full parametric model of the proximal tibia for comprehensive morphological analysis which also enables systematic variation of certain parameters for optimization of the knee shape.

Methods: Based on an extensive literature research and general functional considerations, a total number of 19 morphological parameters of the tibia have been identified, which were clustered into groups such as tibial slopes, curvatures, dimensions, etc. The full automatic computational framework was tested and evaluated on segmented pathological bone surfaces (n=436) of the proximal tibia. The program takes the 3D surface model and calculates an initial reference coordinate system located in the barycentre. Subsequently, the coordinate system is adjusted based on anatomical references. This is used to cut the condyles in the sagittal and frontal plane. Afterwards the program identifies automatically several reference points in the cutting contours for the parameterization. In the area between the reference points geometrical primitives (ellipses) are fitted to approximate the bony contour. At this stage, the user/program could manipulate the parameters of the ellipses in order to systematically change the approximated shape. Then, a 3D surface is generated by using a spline-interpolation method.

Results: The developed framework was able to process 408 of 436 datasets (93.6 %) full automatically without any need for interaction. Looking at the sagittal curvature, the mean medial curvature was positive whereas the lateral was negative, which means that the medial plateau is rather concave and the lateral plateau convex in shape. Comparing medial and lateral dimensions it was observed that the medial plateau is in average bigger than the lateral.

Discussion: This framework might offer the opportunity to study the effect of proximal tibial morphology on knee biomechanics and to provide new insights for patient-specific implant optimization.

ABSTRACT. Introduction: We compared the early clinical and radiological results of robot-assisted unicompartmental knee arthroplasty using a fixed bearing design versus a mobile type bearing design .

Materials and Methods: A data set of 50 cases of isolated compartmental degenerative disease that underwent robot-assisted UKA using a fixed bearing design were compared to a data set of 50 cases using a mobile bearing type design. The operations were performed by one-senior author with the same robot system. The clinical evaluations included the Knee Society Score (knee score, functional score) and postoperative complications. The radiological evaluations was assessed by 3-foot standing radiographs using the technique of Kennedy and White to determine the mechanical axis and femoro-tibial angle for knee alignment. Operative factors were evaluated including length of skin incision, operation time, blood loss, hospital stay and intraoperative complications. Results: There were no statistically significant differences in operation time, skin incision size, blood loss and hospital stay.( p > 0.05 ) There were no significant differences in Knee Society Scores at last follow up. An average preoperative femorotibial alignment was varus alignment of -1° in both groups. Postoperative patients with fixed-bearing implants had an average +2.1° valgus and the patients with mobile bearing implants had +5.4° valgus in femorotibial alignment, which was different.(p<0.05) There was one case of medial tibia plateau fracture in fixed bearing group in 3 months postoperatively. (See Figure 1.) And there were one case of liner dislocation with unstable knee in 6 weeks postoperatively (See Figure 2.) and one case of femoral component loosening in 1 year postoperatively in mobile bearing group. There was no intraoperative complication. The average preoperative knee score was 45.8, which improved to 89.5 in fixed bearing group and 46.5, which improved to 91.2 in mobile bearing group at last followup. The average preoperative function score was 62.4 which improved to 86.5 in fixed bearing group and 60.7 which improved to 88.2 in mobile bearing group at last followup. Conclusion: In our early experience, two types of bearing of robot-assisted UKA groups showed no statistical differences in clinical assessment but there was statistical difference in postoperative radiological corrected alignment. But in aspect of early complications, we think that mobile bearing seems to be requiring more attention in surgery.

ABSTRACT. Introduction: The reproduction of physiological knee kinematics following total knee arthroplasty (TKA) plays an important role in the factor of increasing patient's satisfaction. We hypothesized that using the navigation system, intra-operative knee kinematics after implantation measured may predict that post-operative kinematic in activities of daily living. Our aim was to compare intra-operative knee kinematics by a computed tomography (CT)-based navigation system and post-operative by the 2- to 3-dimensional registration techniques (2D3D).

Methods: Our study were performed for 8patients (1 male, 7 female, 10 knees) who underwent primary PS TKA using CT-based navigation system between Sept.2012 and Sept.2014. All patients were diagnosed with medial osteoarthritis. The median follow-up period from operation date to fluoroscopic surveillance date was 13months (range 5–37 months).

Intraoperative kinematics was measured using the navigation system after final implantation and closure of the retinaculum during passive full flexion and extension imposed by the surgeon.

Navigation and 2D3D had a common coordinate origin for components. AT/PT was the movement of the origin. Under fluoroscopic surveillance in the sagittal plane, each patient was asked to perform sequential deep knee flexion under both weight bearing (WB) and non-weight bearing (NWB) conditions from full extension to maximum flexion. Intra-operative and post-operative comparisons were made using a Wilcoxon signed-rank test.

Results: Mean ER (+)/IR (-) from 0°to 120° were WB conditions: 5.1±7.1°, Intra- operative(Intra):9.5±10.2°, NWB conditions：6.7±8.2, respectively. Mean AT (+)/PT (-) at 0° were WB; -15.4±1.8mm, Intra; -14.7±6.5mm, NWB; -14.1±2.8mm, respectively. Mean AT (+)/PT (-) at 90° were WB; -5.7±1.4mm, Intra; -4.4±4.3mm, NWB; -4.5±2.0mm, respectively. Mean AT (+)/PT (-) at 120° were WB; -17.0±1.4mm, Intra; -14.9±4.2mm, NWB; -17.5±2.3mm, respectively. Mean ADD (+)/ABD (-) at 0°were WB;-0.2±0.4, Intra;-0.4±2.9, NWB;-0.8±0.8, respectively. Mean ADD (+)/ABD (-) at 90°were WB;-0.9±0.6, Intra; 2.8±3.7, NWB;-0.5±2.1, respectively. Mean ADD (+)/ABD (-) at 120° were WB;-1.6±0.5, Intra; 1.1±3.4, NWB;-0.2±2.7, respectively.　No Statistically significant difference was found in Mean ER（+）/IR(-) of WB vs Intra and NWB vs Intra(p>0.05).

Discussion: Our study demonstrated that intra-operative kinematics could predict post-operative kinematics. There was a few similar previous studies. The study had three limitations as following;①Intra-operative measurement was under anesthesia and patient’s muscle was relaxed state.②Post-operative measurement was under the force of gravity③small number of the patients. Therefore, our study needs to increase the number of the patients.

ABSTRACT. Introduction: We have developed an automated algorithm of constructing patient-specific anatomical coordinate system (ACS) for the normal femur based on the bony geometry. However, by evaluating the algorithm at deformed joints, more than ten degrees errors were found especially in rotation comparing with a manually determined ISB recommended coordinate system. It was considered that the spherical approximation of posterior condyles was influenced by osteophytes and deformations around the articular surface. Therefore, in this study, the automated algorithm was improved for both normal and deformed femur.

Methods: We then compared automatically constructed ACS with ISB recommended coordinate system, to evaluate the accuracy of the algorithm. A minimum volume bounding box for point set of femoral model was calculated, the orthogonal coordinate system of the box was used as the temporary coordinate system. Outlines of two posterior condyles and femoral head were tracked; spherical approximation of each outline using the least squares method was performed; centers of the three spheres were used to construct the ACS of the femur. Only the middle third of outlines at both posterior condyles were used for spherical approximation to reduce the influence of osteophytes and deformations.

Results & Discussion: The direction of automatically constructed ACS was evaluated by comparing with ISB recommended coordinate system. Since there are no significant differences between two coordinate systems in 25 randomly chosen patients, it was considered that our algorithm had a high accuracy to take account of patient-specific bony geometry. Moreover, the influence of osteophytes and deformations were also reduced using the improved algorithm.

ABSTRACT. Introduction: Symptom and functional disability in knee osteoarthritis are mainly related with disorders in several principal structures such as articular cartilage, meniscus, ligament, and bone. Among those disorders, synovial inflammatory activities are strongly influential on severity of pain, disability, and prognosis of osteoarthritis. We hypothesized that T2 mapping of knee joints may be useful to discriminate synovial membrane and fluid without using contrast agent, due to remarkable variations of T2 values among those structures. We tried to assess the volume of synovial membrane and synovial fluid on non-enhanced T2 mapping and examine associations of synovial inflammation activity with clinical symptom and structural disorders of the articular cartilage.

Methods: MR imaging of the knee was obtained in 22 patients who showed knee mild or advanced osteoarthritis on plain radiographs and provided consent to this preliminary study. There were 5 males and 17 females, and the mean age of patients was 70.8 years (range; 53 to 80 years). 2D consecutive sagittal T2 map images (TR: 1500 ms; TE: 8 echoes between 10- 80 ms; slice thickness: 3 mm; FOV: 12 cm; acquiring time: 12min54sec.) were obtained using 3.0-T MRI system. Previous studies showed mean T2 value of synovial fluid as approximately 650ms ± 110 and that of articular cartilage or subcutaneous fat as approximately 50 ms ± 1 ms. On T2 mapping images, we defined synovial fluid as region with T2 value more than 430 ms (average T2 value -2SD according to the previous studies), and synovial membrane as region with T2 value between 100 ms and 430 ms. Manual segmentation of each synovial membrane and fluid was performed for joint cavities at the suprapatellar recess and femoro-tibial joints on a slice-by-slice basis using our custom-made software (Baum globe, Osaka Univ.), by one observer. Knee structural disorder in cartilage was assessed semiquantitatively using the Whole-organ Magnetic Resonance Imaging Score (WORMS). Severity of knee pain was assessed using the New Knee Society Score (NKSS). Relationships among volume of synovial membrane, volume of synovial fluid, M/F ratio (Ratio of synovial membrane volume to fluid volume), WORMS scores of cartilage and NKSS was analyzed.

Results & Discussion: In all cases, synovial membrane and fluid were easily differentiated on T2 mapping images. Volumes of synovial membrane, fluid and M/F ratio showed no significant correlations with WORMS score of cartilage. There was positive correlation between volume of synovial membrane and NKSS (r=-0.509, p=0.0298).

ABSTRACT. Introduction: Surgical treatment with hip replacement is recommended to deal with elderly femoral neck fracture. The risk of dislocation may be one major reason why orthopaedic surgeons hesitate to perform total hip replacement (THR). Imageless computer navigation can improve precision of acetabular cup placement that may minimise dislocation. We conducted a study to determine dislocation rate and functional outcome for elderly patients with femoral neck fractures.

Methods: Retrospective review study in elderly femoral neck fracture that underwent THR with imageless computer navigation in Maharat Nakhon Ratchasima hospital was done between January 2011 and June 2015. This study evaluated dislocation rate, acetabular component positioning, functional outcome and peri-operative complications.

Results: There is no dislocation in this study after follow-up at least 6 months. Means of acetabular cup abduction and anteversion angles were 37.6(32.5-42) and 11.1(8.9-19.2) degrees, respectively. Functional outcomes were evaluated with Barthel index scores at 6 months follow-up, 86% were excellent (mean 17 from 20). Eighty seven percent of the patients can return to pre-injury status. Four patients (8%) were dead after 6 months follow-up and half of them were dead within 30 days after surgery. There is no case of revision surgery at 6 months follow-up.

Discussion: Total hip replacement with imageless computer navigation is promising treatment options for elderly femoral neck fracture with low dislocation rate and excellent functional outcome.

ABSTRACT. In total knee arthroplasty hip centre is an important landmark for lower limb kinematics for CT-less approaches. This paper investigated the influence of different motions of femur during hip centre detection. The results show that the effects of pelvic motion and soft tissue on the non-sphericity of hip joint, the flexion/extension motion pattern may be the most reliable for locating the hip centre.

ABSTRACT. Maltracking of the patella can cause anterior knee pain and patellar instability. It is widely accepted that maltracking causes elevated patellofemoral contact pressures resulting in anterior knee pain and subsequent arthrosis. Although most patients with patellofemoral symptoms experience improvement with conservative treatment, a subset of patients fail to respond to such interventions and ultimately require surgery. Surgeons can perform a variety of soft tissue releases and bony procedures to address patellofemoral tracking pathology in conjunction with conservative treatments such as strengthening of the vastus medialis. Specifically, lateral retinacular release and tibial tuberosity osteotomies (TTO) are commonly used to correct abnormal patellar tracking. However, there is little experimental evidence regarding patellar kinematics and contact mechanics after these realignment procedures. As some TTO procedures are difficult, such as anteromedialisation, and are associated with elevated risk of postoperative fracture, experimental insights may benefit clinicians to help in understanding the potential outcomes of using a combination of TTO, soft tissue release and conservative intervention. The objective of this study was to use finite element analysis to evaluate the effects of soft tissue release, tibial tuberosity osteotomies, and quadriceps strength on patellar kinematics and patellofemoral joint (PFJ) contact pressure.

ABSTRACT. Introduction: Analysis of patient-specific fiber direction of skeletal muscles has been explored in various applications such as finite-element biomechanical modelling. We have been focusing on a clinical CT to perform the analysis of muscle fiber structure. The advantage of using clinical CT rather than other common modalities such as MRI is twofold: 1) it allows a study with a large number of population because CT is acquired for almost every patient in routine orthopaedic clinics, 2) it can cover a large muscle without being affected by motion artefact because of the fast acquisition time compared to MRI. In this paper, we present our preliminary analysis of estimating local orientation of the patient-specific muscle fiber direction using clinical CT using a cadaveric image dataset called Visible Korean Human dataset. This will form the basis for further extension of patient-specific biomechanical modelling and the statistical analysis of muscle fiber direction with a massive database.

Methods: Visible Korean Human dataset includes an optical image volume (i.e., a series of photographs of the cross section of a cadaveric specimen) and a CT volume. The optical volume was pre-processed with a low-pass filtering to eliminate artifact due to intensity inhomogeneity along the direction perpendicular to the cross section plane and a segmentation mask to extract the target muscle (in this study, the gluteus maximus muscle). The structure tensor analysis was performed to estimate local orientation of the muscle fiber at each voxel.

Results & Discussion: The angular error of the orientation estimated from CT was calculated using the optical volume as a ground truth, because the optical volume potentially yields a higher accuracy owing to the ten times higher resolution. We observed a reasonably good agreement between the muscle fiber orientations estimated from CT and optical volumes. The median error was 8.85 degrees. Our future work includes estimation of the global fiber geometry using the information of the fiber orientation estimated with the algorithm presented here.

ABSTRACT. Introduction: Removal of impinging osteophytes is one of the traditional surgical treatments for elbow arthritis. However, impingement region is resected by the surgeon according to his experience and the detailed method in determining the osteophytes was not demonstrated in literature. This study describes a reliable and repeatable method to locate the region of debridement using elbow motion simulation with only one CT scan of patient.

Methods: This study simulated the rotation axis of elbow by defining the line connecting centres of two spheres fit with the articular surface of capitellum and the lateral trochlea. Cadaveric study about the axis deviation between the true rotation axis and the simulated rotation axis of elbow was found to be 2.44°. Three patients suffering from primary elbow arthritis were included in this study. One CT scan with the region of interest covering the elbow joint was taken for each patient. The rotation axis of elbow joint was determined and the 3D model of the elbow was rotated around the defined simulated rotation axis in 3matic software® (Materialise, Leuven, Belgium) to simulate the full range elbow movement. The elbow movement was simulated from 0º to 150º. The overlapping regions of the bones throughout the elbow movement were regarded as the impinging osteophytes, which were then imported to the navigation system for intra-operative debridement.

Results & Discussion: The results show that the range of motion of elbows were improved obviously. Mean range of the flexion-extension motion was improved by 60°. The new approach described in this study can quantify and facilitate accurate resection of the region of impingement and osteophytes. Radiation dosage of patient can also be reduced since only one CT scan is required for simulating the elbow motion.

ABSTRACT. Introduction: We have been developing patient-specific musculoskeletal model from clinical CT images. The main challenge in application of the patient-specific biomechanical simulation is that the muscle attachment sites are not visible in CT or MRI. In order to obtain information about the attachment, specifically its inter-subject variability, we conducted cadaver studies. We accurately recorded the attachment site of eight hip muscles in eight cadaver specimens using an optical navigation system. In this study, we recorded attachments of twelve hip muscles in eight fresh cadavers. In this initial report, we present our preliminary analysis of variation of the muscle attachment sites based on the measurement in the cadavers.

Methods: Eight fresh cadaver specimens of the lower extremity including three male and five female with the age at death of 67.5 ±14 y.o. were used in the experiment. The orthopaedic surgeons dissected the target muscle and traced the attachment site using a pointer tool of the optical tracker. The attachment area was mapped on an average geometry using a non-rigid registration method and variation of the attachment areas was analyzed on the normalized geometry.

Results & Discussion: The evaluation of inter-subject variability in eight cadaver specimens suggested that the attachment sites of some muscles have a large variation. We believe this results serve as a valuable “training dataset” in statistically estimating muscle attachment sites from a clinical CT image. The goal of our project is to construct a statistical model for estimating the patient-specific muscle attachment sites based on the structure visible in the clinical CTs.

ABSTRACT. Introduction: The human hip is known to translate when imaged with CT or MRI scans and when tracked by surgical navigation systems. Hip translation has also been qualitatively observed from plain radiographs, which are appealing for their low cost and ubiquity. A first step in quantifying such kinematics is validation of a shape model. This work used best-fit ellipses of subchondral hip contours to model the femoral and acetabular shapes from AP pelvic radiographs. The first hypothesis of this study was that ellipses model arthritic hip shapes. The second hypothesis was that ellipses capture femoroacetabular contact and joint space.

Methods: Supine AP plain radiographs of the pelvis were obtained from 13 patients who underwent unilateral total hip arthroplasty. Ellipses were fit to the segmented femoral and acetabular contours. The RMS deviations of ellipse fits to the segmented contour of the femur were 0.7mm±0.2mm and RMS deviations for the acetabulum were 0.7mm±0.3mm. The fitting errors were sufficiently small that higher-order geometries were not considered. Regions of cotangency were calculated between the ellipse fit to the femur and that of the acetabulum, with a fit tolerance of 1 degree. Points of approximate or exact intersection of the two ellipses were calculated to find possible points of contact within the joint.

Results: The morphological analysis found contact in 12/13 of the operative side and 8/13 of the contralateral sides. These findings are unsurprising in an elderly arthritic population. Ellipse cotangency was found in 100% of the hips. In many, but not all, cases the cotangency regions were roughly aligned with the primary compressive group of the femoral head, suggesting an association with load bearing. This work complements the current, mainly qualitative, understanding of planar hip kinematics. Ellipse cotangency is also an appealing way of consistently measuring joint space, which currently uses rough guidelines such as “perpendicular to the femoral head” in the radiographic definition of hip osteoarthritis.

Discussion: Ellipses are promising as simple, numerically robust, and mathematically rigorous models of elementary joint shape. In a limited arthritic population, we found that ellipses modelled contact and joint space in useful and consistent ways.

ABSTRACT. INTRODUCTION Wolff’s Law (Wolff 1892) proposes that bone architecture adapts in response to mechanical loading. The current study investigates the relationships between the distribution of bone density in each of the proximal tibia, distal femur and patella with 3D measurements calculated following EOS imaging (EOS Imaging, Paris). It was hypothesized that the bone density distribution would reflect the pattern of loading through the knee. MATERIALS AND METHODS The study involved 49 patients (37F, 12M) with osteoarthritis of the knee. Each patient underwent both CT scans of the knee and EOS imaging. A 3D reconstruction model of the lower limb was produced using dedicated software (SterEOS, EOS Imaging, Paris), and the measurements were automatically computed. The CT images were analyzed using ImageJ (Rasband 1997-2015). The ratios of the average density of the tibia, femur and patella were calculated, along with medial/lateral (M/L) and anterior/posterior (A/P) density ratios for each. Correlational analysis was carried out to assess whether there were any relationships between these variables and the EOS parameters after controlling for age, sex, height and weight. RESULTS The results showed a number of correlations between the density ratios and EOS parameters. In particular the valgus/varus angle was negatively correlated with the tibia M/L and femur M/L density ratios and positively correlated for patella/tibia; the femoral mechanical angle was negatively correlated with the femur M/L density ratio; the tibial mechanical angle was negatively correlated with the tibia M/L and tibia/femur ratios and positively correlated with patella/tibia; and the femorotibial angle was significantly correlated with the tibial M/L ratio. Both the sagittal and lateral tilt of the pelvis were also found to be related to the density distribution in the knee. DISCUSSION The negative correlations between the varus/valgus angle and M/L ratio for both the femur and tibia are consistent with previous research and reflect the loading pattern through the knee. The correlations between the femoral mechanical angle and femoral M/L ratio, and the tibial mechanical angle and tibial M/L ratio are also consistent with this pattern. The results showed a number of other significant correlations between clinical parameters and density ratios that have not been previously reported. These may give additional insight into the loading effects of various biomechanical factors.

ABSTRACT. Introduction: Recent studies showed that glenoid bone density characterization revealed a uniform density for symmetric glenoid wear patterns. The purpose of this study is to compare the regional bone density in normal glenoids to different types of osteoarthritic glenoids. The preservation of the dense sclerotic remaining bone of the arthritic glenoid may be an important factor in decreasing implant migration and failure. So far, no intuitive and efficient tool was designed to guide the surgeon in the reaming process.

Methods: Ct-scan exams were semi-automatically segmented. 20 glenoids per group: Normal, A1, A2, B2. Each glenoid was separated into 4 quadrants (B2 glenoids were separated into neoglenoid and paleoglenoid). Volumes of Interest were defined at depths of 0-2.5 mm (Zone A), 2.5-5 mm (Zone B) and 5-7.5 mm (Zone C). Average bone density was measured in Hounsfield Units.

Results: Mean bone density between patients is not uniform in A2 and B2 glenoids. It confirms uniformity for Normal glenoids. It is also significantly higher in A2, B2 glenoids compared to Normal glenoids. In B2 Glenoid, bone density of the neoglenoid was significantly denser than the paleoglenoid. The postero-inferior and the postero-superior quadrant had the highest density in zones B and C. The results report that the glenoid mean density for symmetric/asymmetric erosion is higher than in normal glenoids.

Discussion: This study shows that A2 and B2 glenoids do not have a uniform wear pattern. It also confirms that Normal glenoids have a uniform erosion wear pattern. All osteoarthritic glenoids have a higher mean density compared to the controlled group. To know how to preserve the bone stock and to maximise the implant primary fixation, an automated segmentation tool and intuitive bone density display could be interesting to assist the surgeon in the surgical planning.

ABSTRACT. Introduction: Patient-specific instruments (PSIs) are becoming increasingly common in orthopaedic surgery. However, despite their ability to register to bony anatomy without clear landmark visualization, PSIs have only been used with traditional, highly invasive surgical approaches, which can result in significant postoperative functional complications. Therefore, to take full advantage of its strengths, we previously designed a novel PSI and surgical instrumentation system that enables total shoulder arthroplasty to be conducted minimally invasively. In this work, we investigate the accuracy of this system in creating the guide axes needed for humeral and glenoid bone preparation and implantation. 

Methods: Accuracy was investigated on three cadaveric shoulder specimens, each of which had a PSI custom designed and manufactured. The PSI was designed in two components that independently registered to the humerus and scapula before being rigidly connected together to place the bones into a pre-operatively planned pose – a pose in which their respective central guide axes were aligned collinearly. Once locked, a c-shaped drill guide was attached to the PSI, which allowed a hole to be drilled along the collinear guide axes from the lateral humerus, through the humeral head, and into the glenoid. The positional accuracy of the entry and exit holes on the humerus and scapula, as well as the direction of the drill hole, were assessed using an optical tracking camera and an iterative closest point algorithm based registration method to map measurements to the pre-operative plan. 

Results: Assessment of entry and exit drill holes produced a mapping accuracy of 3.7±2.6mm for the lateral humerus entry hole, 2.6±1.9mm for the humeral articulation exit point, 2.2±1.4mm for the glenoid entry hole, and 4.9±2.6° for the tunnel orientation. Comparison of these data to published accuracy results for unassisted (3.4mm & 7-12°) and PSI assisted open techniques (2.4mm & 4°) clearly demonstrated that this novel minimally invasive system produces similar accuracy. 

Discussion: With these results in mind, we believe that this minimally invasive PSI and instrumentation system can guide total shoulder arthroplasty with acceptable accuracy, while avoiding the complications inherent to traditional open approaches. In the future, this work will be extended to include bone preparation using novel, minimally invasive tools.

ABSTRACT. Introduction: Pedicle screw insertion is a common procedure in spinal surgeries that aim to stabilize vertebral segments. Free-hand procedures are the most common operational practice with high rates of screw malposition. A number of patients can experience neurological symptoms as a consequence of screw misplacement. Follow-up revision surgeries are seldom performed to correct for screw malplacement that are expensive and time consuming procedures and ideally must be avoided. The current screw position assessment methods (intraoperative X-ray, intraoperative O-arm and postoperative computed tomography) are associated with limitations such as: high radiation exposure, metal artifact, lack of 3D information and postoperative nature. Hence, there is a significant need for a reliable intraoperative technique for checking screw placement prior to closing the patients. This study describes a robust intraoperative (post-placement) pedicle screw position assessment system that can offer an intuitive 3D representation of the screw alignment relative to the vertebrae of interest.

Methods: Such a system enables surgeons to decide whether or not a re-insertion is required during the operation. Intraoperative fluoroscopic shots are first acquired and calibrated using Roentgen Stereophotogrammetric Analysis (RSA). 2D/3D registration methods are then utilized to analytically represent the vertebral/screw models within a common coordinate frame. Vertebral and screw models are then superimposed on one another to represent the actual anatomical configuration of the implant.

Results & Discussion: The proposed method is capable of localizing the implanted screws with average translational and angular errors on the order of 2 mm and 1.3 degrees respectively.

ABSTRACT. Introduction: The goal of this study is to report an original computer-assisted-surgery procedure for the development of a personalized implantation process in total ankle replacement (TAR), starting from image acquisition and original modelling, through manufacturing process, up to implantation in-vitro using a surgical navigation system (SNS).

Methods: Ten normal cadaver lower limbs were analysed. MRI/CT acquisitions were performed via for relevant ankle modelling. For each specimen, based on relevant models, a set of three custom-fit virtual articular surfaces were produced, one on them based on an original modelling approach. Corresponding prototypes in polylactide and Cobalt-Chromium were manufactured via Selective-Laser-Melting (SLM). Each specimen was tested using SNS, suitably adapted to track originally bones motion at the ankle, for kinematic evaluations before and after TAR implantation. The novel procedure was carried out with no problems.

Results: Repeatable motion patterns were generally observed together with a good restoration of the physiological motion after ankle replacement using originally modelled TAR device. The different joint modelling on which the three models were based resulted in kinematic differences. Tests on metal prototypes corroborated these results.

Discussion: An original and successful CAS procedure has been here reported for the development of a novel implantable personalised TAR device based on novel modelling. The feasibility of a manufacturing via SLM, also in metal powders, has been also established. The adaptability of SNS also to a small joint replacement, like TAR, is here confirmed. The in-vitro tests confirmed recent findings on natural joint motion restoration using the TAR device based on original modelling.

ABSTRACT. Introduction: The femoral anteversion angle is measured in the treatment such as total hip arthroplasty and femoral fracture. The femoral anteversion angle is defined as the anterior inclination of the neck axis relative to the femoral coordinate system. We propose a method that automatically determines a sub-region of the femoral surface corresponding to the neck by using the statistical shape model and robustly computes the neck axis by computing a center line of the sub-region.

Methods: First, we compute an average femur shape from a training dataset including 100 patients who underwent total hip arthroplasty. Then, we manually identify the neck region on the average shape. Given a CT of the target patient, we apply an automatic segmentation that we previously developed and register the average shape to the target femur to establish correspondence of the vertices, which provides the neck region of the target femur. Then, we compute a series of slices that naturally cut the neck region at approximately constant interval using a method called Harmonic function field. Finally, we define the neck axis as a center line of those contours. To evaluate the accuracy of the proposed method, we compared the automatic result with the gold standard measured by an expert surgeon. We tested feasibility of the proposed method in a large population study using CT image of 180 patients including femoral head necrosis and hip osteoarthritis. All the patients we analysed had operation on one side and no operation the other side, which allowed us comparative studies.

Results & Discussion: The accuracy evaluation experiment showed that the average error in the measured anteversion angle was 5.27 degrees.

ABSTRACT. Introduction: We have developed an automated algorithm of constructing patient-specific anatomical coordinate system (ACS) for the normal tibia based on the bony geometry, derived from computed tomography. Owing to the automated nature of the algorithm, all constructions were performed within 30 seconds; furthermore, between- and within- rater errors were zero. However, by evaluating the algorithm at deformed joints, more than ten degrees errors were found especially in rotation comparing with a manually determined Akagi' anteroposterior line. It was considered that the direction of anteroposterior axis was influenced by deformations and osteophytes around the articular surface. Therefore, in this study, the automated algorithm was improved for both normal and deformed tibia. We then compared automatically constructed ACS with several anatomical reference axes, to evaluate the accuracy of the algorithm.

Methods: The first principal component vector described the longitudinal direction of tibia was defined as the Z-axis. The middle of the posterior cruciate ligament and the medial border of the patellar tendon attachment were automatically determined in the improved algorithm to define the Akagi’s anteroposterior line. The direction of automatically constructed ACS was evaluated by comparing with manually digitized mechanical axis and Akagi’s line.

Discussion: Since there are no significant differences between the ACS and reference axes in 25 randomly chosen patients, it was considered that our algorithm had a high accuracy to take account of patient-specific bony geometry, and had better reliability and the advantage of not being influenced by any deformations.

ABSTRACT. Introduction: Patient’s position reportedly affects acetabular and femoral implant orientation. However, the effect of patient’s position on femoral rotation has been rarely discussed.

Methods: We evaluated the femoral rotation in standing position in 19 patients (mean age: 62 year-old, 15 females and 4 males) who suffered from hip osteoarthritis and was planed to undergo bilateral total hip arthroplasty. EOS examination, which is a biplane radiographic based novel 3D modeling system, was performed in standing position before and at 3 months after THA. 3D calculation of the anatomical or functional femoral anteversion was conducted by projecting femoral or implant head-neck line onto the plane that is perpendicular to table top plane or anterior pelvic plane. By subtracting anatomical femoral anteversion from functional femoral anteversion, the rotation angle of femoral plane to anterior pelvic plane in standing position was also calculated.

Results: In our population, despite significant increasing of anatomical femoral anteversion after THA (24.4°±12.4 vs 31.9°±14.7, p=0.04), functional femoral anteversion was less affected (19.3°±15.9 vs 23.8°±15.6). The rotation angle of femoral plane in standing position was decreased in 13 out of 19 patients (68.4%) after THA (-5.1°±10.7 vs -8.1°±14.0). The rotation angle of femoral plane tends to negatively correlate with combined anteversion (r=-0.39, p= 0.099).

Discussion: It is likely that postoperative femoral anteversion is automatically adjusted by changing rotation of femoral plane to obtain the adequate combined anteversion. Effects of rotation angle of femoral plane on functional femoral anteversion need to be considered for the better understanding of acetabular and femoral implant orientation after THA.

ABSTRACT. Introduction: This study evaluated the accuracy of computer-navigated arthroscopic surgery for cam-type femoroacetabular impingement (FAI) using a 3D-model comparison that allowed evaluation of bone resection during actual surgery. The accuracy of computer-navigated osteochondroplasty was reported in several previous studies in vitro. However, those studies were based on the results of simulation surgery, and mainly focused on alpha-angle correction, which could not represent detailed morphological features of the resection site.

Methods: Thirteen subjects who had received computer-navigated arthroscopic osteochondroplasty were included. A 3D surgical planning model was constructed based on the impingement simulation result, and imported as “guide” into the computer navigation system. The arthroscopic osteochondroplasty was performed under the guidance of a computer navigation system, which showed the planned resection area in real-time. Postoperative femur models were constructed from CT images obtained at 1 week postoperatively. The 3D model comparison between the planning model and postoperative femur model were performed using CloudCompare.

Results: The mean deviation from the planned depth in an under-resected lesion was 3.2 ± 1.5 mm, and 3.6 ± 1.7mm for the over-resected lesion. In most subjects, the under-resected area was seen on the lateral side of the femoral headneck junction, while the over-resected area was seen in the center of the resection site.

Discussion: The results of our method for comparing postoperative femur model and planning model indicated the accuracy of computer-navigated arthroscopic osteochondroplasty during actual surgery. We believe that this method is useful for performing postoperative evaluation of arthroscopic osteochondroplasty by computer navigation, which could also provide locational information of errors in resection depth.

ABSTRACT. Introduction: Malunited forearm diaphyseal fractures often limit forearm rotation; however, the biomechanical mechanism of the restriction in detail is unclear. This study assessed in vivo three-dimensional (3D) forearm motion with malunited forearm diaphyseal fractures to evaluate the role of deformity, bone impingement, and contractures of the interosseous membrane, especially central band (CB).

Methods: Seventeen malunited diaphyseal fractures with restricted forearm rotation were investigated (3 females, 14 males). Eleven cases had both the radius and the ulna deformities, 6 cases had the radius deformities alone. Computed tomography of the bilateral forearm was performed for three rotational positions (maximum pronation, supination, and neutral position), and 3D bone surface models of the radius and ulna were created. Kinematic variables were measured by superimposing the ulna in neutral and pronated positions on the images of supinated position with a surface-based registration technique using originally developed computer software (Bone Simulator; Orthree, Osaka, Japan). The path of the CB were created based on anatomical locations. As CB was wide membrane, proximal, middle, and distal longitudinal fibers and transverse fibers avoiding bone obstacles were calculated. Bone impingement was defined as the distance between radius and the ulna lower than 1.5mm during forearm rotation. The relations of 3D deformity to forearm motion range, bony impingement, and each CB fiber length were quantitatively analyzed.

Results: Extension deformity of the radius related to the restriction of pronation with coefficient correlation(R) 0.57 (p = 0.02) and valgus deformity of the ulna correlated with the limitation of supination (R = 0.71 (p = 0.01)). Sixteen cases had limited pronation and 13 cases had limited supination. Significance difference existed in impingement ratio between pronation (75%: twelve of 16 cases) and supination (8%: one of 13 cases, p < 0.01). Extension deformity of the radius increased risk of impingement during pronation with odds ratio 1.12 (95% confidence interval 1.012-1.319). As, the transverse fiber of CB was significantly elongated in supination (88.8±23.4%) compared with in pronation (61.5±15.2%, p < 0.01), the limitation of supination was caused by overstretching CB. When compared to normal side, the transverse fiber of CB in pronation significantly shortened (p < 0.01). The longitudinal fibers of CB ligament had no significant difference between supination and pronation.

Discussion: Limited pronation of malunited forearm diaphyseal fractures is principally caused by bone impingement. Limited supination is mainly caused by strained transverse fiber of CB.

ABSTRACT. Introduction: Initial fixation is a key factor in the success of cementless THA using a tapered wedge stem. The purpose of this study was to use three-dimensional templating software to examine the correlation between quantitative contact state and important clinical radiological outcomes, specifically stem subsidence, stress shielding, and cortical hypertrophy.

Methods: We conducted a retrospective consecutive review of 75 hips in 70 patients over a minimum three-year follow-up period. X-rays and CT scans were investigated to assess preoperative planning, quantify the contact state of implant and femur, and assess stem alignment, stem subsidence, stress shielding, and cortical hypertrophy. We evaluated the correlation between radiological outcomes and three-dimensional quantitative contact state according to Gruen Zone in each Dorr classification.

Results: Density mapping indicated that stem subsidence increased postoperatively if the stem had less cortical contact in the middle to distal portion of the implant in terms of initial fixation. Cases having too much cortical contact in the distal portion of the implant tended to have increased stress shielding. We found no correlation between cortical hypertrophy and the contact state of implant and femur.

Discussion: Density mapping with three-dimensional templating software can be useful in predicting stem subsidence and stress shielding following cementless THA with a tapered wedge stem. Further analysis is required to accurately depict the correlation between cortical hypertrophy and the contact state.