Preoperative Evaluation and Surgical Simulation for Osteochondritis Dissecans of the Elbow Using Three-Dimensional MRI-CT Image Fusion Images

We used our novel three-dimensional magnetic resonance imaging-computed tomography fusion images (3D MRI-CT fusion images; MCFIs) for detailed preoperative lesion evaluation and surgical simulation in osteochondritis dissecans (OCD) of the elbow. Herein, we introduce our procedure and report the findings of the assessment of its utility. We enrolled 16 men (mean age: 14.0 years) and performed preoperative MRI using 7 kg axial traction with a 3-Tesla imager and CT. Three-dimensional-MRI models of the humerus and articular cartilage and a 3D-CT model of the humerus were constructed. We created MCFIs using both models. We validated the findings obtained from the MCFIs and intraoperative findings using the following items: articular cartilage fissures and defects, articular surface deformities, vertical and horizontal lesion diameters, the International Cartilage Repair Society (ICRS) classification, and surgical procedures. The MCFIs accurately reproduced the lesions and correctly matched the ICRS classification in 93.5% of cases. Surgery was performed as simulated in all cases. Preoperatively measured lesion diameters exhibited no significant differences compared to the intraoperative measurements. MCFIs were useful in the evaluation of OCD lesions and detailed preoperative surgical simulation through accurate reproduction of 3D structural details of the lesions.


Patient Selection
The institutional review board of the University of Tsukuba Hospital approved this study (Study Number: H29-58). Twenty-eight patients visited our facility and were diagnosed with OCD between July 2017 and March 2021. Among them, 16 patients whose lesions were evaluated using MCFIs were enrolled in this study. These patients subsequently underwent surgery. We obtained written informed consent from each patient. It was clearly stated that we would only use MCFIs for lesion evaluation and to decide the treatment strategy. All patients were boys (mean age: 14.0 ± 1.0 years, range: 12-16 years), and the right side was affected in fifteen patients, while the left side was affected in one patient. The average body weight of the patients was 56.9 (48.0-65.0) kg.

Obtaining the MR and CT Images
A 3-Tesla imager (MAGNETOM© Verio, Siemens, Munich, Germany) was used for MRI. We followed the procedures we published in a previous article on imaging sequence and settings, position of patients, and application of axial traction (7 kg) [28]. Axial traction widens the joint space and helps better visualize an outline of the articular cartilage of the humeral capitellum [28,30] (Figure 1). agnostics 2021, 11, x FOR PEER REVIEW Figure 1. Acquisition of magnetic resonance images with axial traction. Axial tract applied to the elbow to better visualize the articular cartilage of the humeral capitellu We used a 320-row scanner (Aquilion ONE TM , Toshiba, Tokyo, Japan) images were obtained with a 0.5 mm slice thickness. We did not apply axial ing CT because CT data of the whole joint were not necessary for the proced

Creation of 3D Models
Using the obtained data, we created 3D MRI models of the humerus cartilage and a 3D CT model of the humerus. The Materialise Mimics Inno version 20 (Materialise © , Leuven, Belgium) was used for this procedure. W the MR signal intensity of the articular cartilage and humerus in each case fo 3D MRI models. As the participants in this study were skeletally immature cartilage was present in some cases. First, we set a threshold for the MR sig for each target tissue. According to the set threshold, we selected the pixe sponded to the target tissue. The threshold was roughly set automatically manually while simultaneously referring to the monitor to ensure correct se target tissue. This procedure is called segmentation, which is crucial for cr images. While segmenting the articular cartilage, we defined the articular sures (ACFs) as the low-intensity lines within the articular cartilage, which are perpendicular to the articular surface [12,15,[31][32][33]. Articular surface defo was defined as irregularities in the outline of the articular cartilage [12,15] (F segmented structures, as well as the ACFs and ASDs were reconstructed and into 3D models. Similarly, we created 3D CT models of the humerus. Durin dure, the subchondral bone was considered segmented when the discontinui chondral bone to the floor was observed in all three planes: axial, sagittal, We manually created a separate 3D model of segmented subchondral bone ( played it in red color for better visualization (Figure 3). We used a 320-row scanner (Aquilion ONE TM , Toshiba, Tokyo, Japan) for CT, and images were obtained with a 0.5 mm slice thickness. We did not apply axial traction during CT because CT data of the whole joint were not necessary for the procedure.

Creation of 3D Models
Using the obtained data, we created 3D MRI models of the humerus and articular cartilage and a 3D CT model of the humerus. The Materialise Mimics Innovation Suite version 20 (Materialise©, Leuven, Belgium) was used for this procedure. We referred to the MR signal intensity of the articular cartilage and humerus in each case for creation of 3D MRI models. As the participants in this study were skeletally immature, the growth cartilage was present in some cases. First, we set a threshold for the MR signal intensity for each target tissue. According to the set threshold, we selected the pixels that corresponded to the target tissue. The threshold was roughly set automatically and adjusted manually while simultaneously referring to the monitor to ensure correct selection of the target tissue. This procedure is called segmentation, which is crucial for creating better images. While segmenting the articular cartilage, we defined the articular cartilage fissures (ACFs) as the low-intensity lines within the articular cartilage, which penetrate or are perpendicular to the articular surface [12,15,[31][32][33]. Articular surface deformity (ASD) was defined as irregularities in the outline of the articular cartilage [12,15] (Figure 2). The segmented structures, as well as the ACFs and ASDs were reconstructed and reproduced into 3D models. Similarly, we created 3D CT models of the humerus. During the procedure, the subchondral bone was considered segmented when the discontinuity of the subchondral bone to the floor was observed in all three planes: axial, sagittal, and coronal. We manually created a separate 3D model of segmented subchondral bone (SSB) and displayed it in red color for better visualization (Figure 3).  The same procedure as used with magnetic resonance imaging was used. The segm dral bone (SSB) was defined as the lesion whose continuity to the floor was lost in The separate SSB model was created manually and is displayed in red for bett Green, red and orange lines are the reference lines correspond to sagittal, axial and tively.

Fusion of Created 3D Models
We used the 3-matic software version 12 (Materialise © , Belgium) to fu The same procedure as used with magnetic resonance imaging was used. The segmented subchondral bone (SSB) was defined as the lesion whose continuity to the floor was lost in all three planes. The separate SSB model was created manually and is displayed in red for better visualization. Green, red and orange lines are the reference lines correspond to sagittal, axial and coronal, respectively.

Fusion of Created 3D Models
We used the 3-matic software version 12 (Materialise©, Belgium) to fuse the 3D models. First, we exported the created 3D models from the Materialise Mimics to the 3-matic. We then roughly fused both the 3D MRI and 3D CT models of the humerus using a function called N-point registration. This function enables two separate 3D models to be superimposed using an arbitrary number of corresponding points. In our procedure, we registered four corresponding points that are easy to recognize and belong to different planes, as reported in a previous study [28] (Figure 4a). Second, we used a function called global registration to fine-tune the position of the superimposed 3D models (Figure 4b).
Using this function, the positions of the aligned 3D models can be automatically corrected depending on their shapes. Throughout the procedures, the 3D MRI model of the articular cartilage was set to move together with the 3D MRI model of the humerus in order to maintain the positional relationship between the structures. Finally, we completed the fusion of the 3D CT model of the humerus and the 3D MRI model of the articular cartilage by hiding the 3D MRI model of the humerus (Figure 4c). We termed this fusion model MCFI. The first author, an elbow surgeon with 14 years of clinical experience, created all MCFIs. The average interval between the MCFI creation and surgery was 26 (5-70) days.

Lesion Evaluation Using MCFIs
The first author evaluated the OCD lesion immediately after image creation based on the MCFI. The evaluation was particularly focused on the findings of the articular cartilage and subchondral bone, such as the presence of ACFs, articular cartilage defect (ACD), or ASD (Figure 5a), and whether the subchondral bone was segmented or not. We recorded the findings of each case. In order to differentiate ASDs from ACDs, deformities were recorded as either protrusions or flattening of the articular surface. The positional relationship between the ACF and the SSB was evaluated by adjusting the transparency of the articular cartilage ( Figure 5b). The articular surface of the humeral capitellum was considered elliptical in the anteroposterior view of the MCFI and divided into four areas, areas 1 to 4, clockwise from the anteromedial area ( Figure 5c); we thereby recorded the location of each finding. Based on the findings reproduced in the MCFI, the vertical and horizontal diameters of the lesion were also measured ( Figure 5d). OCD lesions are defined as unstable when the lesion is displaced or dislocated, and thereby the articular surface is deformed [12,[14][15][16][32][33][34]. We defined unstable OCD lesions in MCFIs as those with ACFs and/or ASDs, along with the SSB underneath.
Based on the created MCFI, the ICRS classification [20] was predicted independently and comprehensively by two elbow surgeons (assessors 2 and 3 who had 13 and 12 years of clinical experience, respectively). The surgeons were requested to predict the ICRS classification at least 3 days preceding the operation. The surgeons were cognizant of the ICRS classification and were blinded to the patients' medical records. The two assessors participated as assistant surgeons in several surgeries performed in this study.

Surgical Simulation
We used the 3-matic software version 12 (Materialise © , Leuven, Belgium) for the surgical simulation. The first author performed all surgical simulations in this study. Articular surface reconstruction by costal osteochondral autograft transplantation [19] is our first choice when the OCD lesion is unstable and its maximum diameter exceeds 10 mm; thus, we simulated the surgical procedure. First, we decided on the resection area according to the reproduced ACFs and ASDs (Figure 5d). After selecting the resection area, we separated the area from the original 3D model of the articular cartilage and cre- OCD lesions are defined as unstable when the lesion is displaced or dislocated, and thereby the articular surface is deformed [12,[14][15][16][32][33][34]. We defined unstable OCD lesions in MCFIs as those with ACFs and/or ASDs, along with the SSB underneath.
Based on the created MCFI, the ICRS classification [20] was predicted independently and comprehensively by two elbow surgeons (assessors 2 and 3 who had 13 and 12 years of clinical experience, respectively). The surgeons were requested to predict the ICRS classification at least 3 days preceding the operation. The surgeons were cognizant of the ICRS classification and were blinded to the patients' medical records. The two assessors participated as assistant surgeons in several surgeries performed in this study.

Surgical Simulation
We used the 3-matic software version 12 (Materialise©, Leuven, Belgium) for the surgical simulation. The first author performed all surgical simulations in this study.
Articular surface reconstruction by costal osteochondral autograft transplantation [19] is our first choice when the OCD lesion is unstable and its maximum diameter exceeds 10 mm; thus, we simulated the surgical procedure. First, we decided on the resection area according to the reproduced ACFs and ASDs (Figure 5d). After selecting the resection area, we separated the area from the original 3D model of the articular cartilage and created an independent 3D model. This enabled us to freely hide the resection area, simulating the lesion resection ( Figure 5e). Second, we simulated autograft transplantation. A 3D model of the average-sized costal osteochondral autograft was created in advance ( Figure 6). We created this model based on the size of the actual autograft harvested from previously operated cases. This autograft model can be freely placed within the MCFI, and we simulated the position, direction, and depth of the transplantation (Figure 7).
ated an independent 3D model. This enabled us to freely ing the lesion resection ( Figure 5e). Second, we simulated model of the average-sized costal osteochondral autograf 6). We created this model based on the size of the actual ously operated cases. This autograft model can be freely simulated the position, direction, and depth of the transp    We simulated drilling when the OCD lesion was ICRS class IV and small, or whe the lesion was stable, articular surface reconstruction was not necessary, and the lesio did not respond to conservative therapy. Generally, subchondral OCD lesions with chronic history have a sclerotic component [1]. It is essential to penetrate the subchondra We simulated drilling when the OCD lesion was ICRS class IV and small, or when the lesion was stable, articular surface reconstruction was not necessary, and the lesion did not respond to conservative therapy. Generally, subchondral OCD lesions with a chronic history have a sclerotic component [1]. It is essential to penetrate the subchondral sclerosis to promote bone healing. Therefore, as the first step of the simulation of drilling, we manually selected the sclerotic region and highlighted it with a different color for better visualization. Second, we simulated the entry point, direction, and depth of drilling. When the lesion is stable and the articular surface is intact, it is important to avoid iatrogenic articular cartilage damage. Therefore, we simulated posteroanterior drilling for those cases (Figure 8).

Intraoperative Evaluation of ICRS Classification
We evaluated all OCD lesions intraoperatively either under direct observation or using arthroscopic examination to determine the ICRS classification [20]. According to the definition, class I and II lesions are stable, while class III and IV lesions are unstable.

Evaluation of MCFIs
We compared the predicted values with the actual intraoperative findings for the ACFs, ACDs, ASDs, vertical and horizontal lesion diameters, and ICRS classification. The corresponding rate between each finding determined from the MCFI and those determined intraoperatively were evaluated. We calculated the corresponding rate as follows: (number of cases in which the findings determined by the MCFI corresponded to the ac-

Intraoperative Evaluation of ICRS Classification
We evaluated all OCD lesions intraoperatively either under direct observation or using arthroscopic examination to determine the ICRS classification [20]. According to the definition, class I and II lesions are stable, while class III and IV lesions are unstable.

Evaluation of MCFIs
We compared the predicted values with the actual intraoperative findings for the ACFs, ACDs, ASDs, vertical and horizontal lesion diameters, and ICRS classification. The corresponding rate between each finding determined from the MCFI and those determined intraoperatively were evaluated. We calculated the corresponding rate as follows: (number of cases in which the findings determined by the MCFI corresponded to the actual intraoperative findings)/(total number of cases) [17].
Each finding and the corresponding ICRS classification were intraoperatively determined by both visual inspection and palpation. In class III lesions that underwent articular reconstruction, the diameter of the resected lesion was recorded. In class IV lesions, the diameters of the articular cartilage defect were recorded. In 9 of 16 cases, where the first author conducted the surgery, another elbow surgeon with 23 years of clinical experience evaluated the intraoperative findings. This surgeon did not evaluate the lesion using the MCFI. The primary surgeon assessed the intraoperative findings in the remaining seven cases. Among the seven cases, six cases were assessed by an elbow surgeon with 23 years of clinical experience and one case by an elbow surgeon with 34 years of clinical experience. We compared the predicted and intraoperatively measured vertical and horizontal lesion diameters using the Mann-Whitney U test, which was selected because of the small sample size. Statistical significance was set at p < 0.05.

Lesion Evaluation Using MCFIs
Tables 1-3 show the findings predicted by the MCFIs and intraoperatively determined findings for the ACFs, ACDs, and ASDs, respectively. The reproducibility of each finding was accurate in the MCFI, except for one case, in which the lesion was detached at the time of surgery, resulting in a corresponding rate of 93.8%. Table 4 depicts the measurements of the diameters of the lesions. The median values of the vertical diameter measured preoperatively and intraoperatively were 14.8 and 14.0 mm, respectively; there was no significant difference (p = 0.78). The median values of the horizontal diameter measured preoperatively and intraoperatively were 13.1 and 12.0 mm, respectively; there was no significant difference (p = 0.14). Figure 9 shows the intraoperative findings and corresponding MCFIs for representative cases. Table 5 presents the predicted and intraoperative ICRS classifications. The MCFI resulted in a corresponding rate of 93.8% for both examiners. This is because there was one case in which the lesion was detached at the time of surgery (case 15). We predicted this case as ICRS class III; however, the case was confirmed as class IV intraoperatively. Table 1. Validation of the three-dimensional magnetic resonance imaging-computed tomography fusion images (3D MRI-CT fusion images; MCFIs) against intraoperative findings: articular cartilage fissures (ACFs). The reproducibility of the ACFs by the MCFIs was accurate in all cases.  None None Table 3. Validation of the magnetic resonance imaging-computed tomography fusion images (MCFIs) against intraoperative findings: articular surface deformities (ASDs). The MCFIs precisely reproduced the ASDs in all cases except for case 15. The lesion appeared to be protruded on the MCFI but was detached from the floor at the time of surgery.

Surgical Simulation
We simulated drilling in four cases and osteochondral autograft transplantation in eleven cases. One case was indicated to perform free-body removal alone. All surgeries were conducted based on the simulations. We performed drilling for cases 1, 3, 11, and 15, free-body removal for cases 6 and 15, and costal osteochondral autograft transplantation for the remaining cases. Based on the MCFI, case 15 was indicated for articular surface reconstruction, but the patient did not consent to the surgical procedure and opted for the drilling procedure. We conducted all surgeries as simulated in all cases. Figure 10 shows the intraoperative findings and corresponding surgical simulation for representative cases.  The MCFI accurately reproduced the articular cartilage fissures (ACFs) in areas 1, 2, 3, and 4 (arrows). The lesion was predicted as unstable because the segmented subchondral bone (SSB) was present underneath the ACFs, and the articular surface was protruded. As predicted, the lesion was classified as unstable intraoperatively. (b) Case 8. The MCFI correctly reproduced the protrusion of the articular surface in areas 2 and 3 (arrows). The lesion was predicted as unstable owing to the presence of the articular surface deformity (ASD) and SSB underneath the ACF. The lesion was unstable on palpation, as predicted. (c) Case 16. The MCFI correctly reproduced the ACF and ASD in areas 2 and 3 (arrows). The lesion was predicted as unstable based on these findings and intraoperatively classified that the lesion was unstable. (d) Case 9. The MCFI correctly reproduced the ACFs in areas 1, 3, and 4 (arrows). The lesion was predicted as unstable based on the presence of the SSB underneath the ACF, as well as ASD. The lesion was unstable intraoperatively on palpation, as predicted. (e) Case 5. The MCFI correctly reproduced the ACF in areas 2 and 3 (arrows). The lesion was predicted as unstable because of the presence of ACF and the SSB underneath and ASD. The lesion was classified as unstable on palpation. Table 5 presents the predicted and intraoperative ICRS classifications. The MCFI resulted in a corresponding rate of 93.8% for both examiners. This is because there was one The MCFI accurately reproduced the articular cartilage fissures (ACFs) in areas 1, 2, 3, and 4 (arrows). The lesion was predicted as unstable because the segmented subchondral bone (SSB) was present underneath the ACFs, and the articular surface was protruded. As predicted, the lesion was classified as unstable intraoperatively. (b) Case 8. The MCFI correctly reproduced the protrusion of the articular surface in areas 2 and 3 (arrows). The lesion was predicted as unstable owing to the presence of the articular surface deformity (ASD) and SSB underneath the ACF. The lesion was unstable on palpation, as predicted. (c) Case 16. The MCFI correctly reproduced the ACF and ASD in areas 2 and 3 (arrows). The lesion was predicted as unstable based on these findings and intraoperatively classified that the lesion was unstable. (d) Case 9. The MCFI correctly reproduced the ACFs in areas 1, 3, and 4 (arrows). The lesion was predicted as unstable based on the presence of the SSB underneath the ACF, as well as ASD. The lesion was unstable intraoperatively on palpation, as predicted. (e) Case 5. The MCFI correctly reproduced the ACF in areas 2 and 3 (arrows). The lesion was predicted as unstable because of the presence of ACF and the SSB underneath and ASD. The lesion was classified as unstable on palpation. Table 5. Comparison of the International Cartilage Repair Society (ICRS) classification predicted by examiners and the actual intraoperative findings. The intraoperative ICRS classification accurately corresponded to the magnetic resonance imaging-computed tomography fusion image-based predictions in all cases except for case 15. The match rate was 93.8%.

MCFI
Intraoperative Findings Assessor 2 Assessor 3   1  II  II  II  2  III  III  III  3  IV  IV  IV  4  IV  IV  IV  5  III  III  III  6  IV  IV  IV  7  IV  IV  IV  8  III  III  III  9  III  III  III  10  IV  IV  IV  11  IV  IV  IV  12  III  III  III  13  III  III  III  14  III  III  III  15  III  III  IV  16  III  III  III   5  III  III  III  6  IV  IV  IV  7  IV  IV  IV  8  III  III  III  9  III  III  III  10  IV  IV  IV  11  IV  IV  IV  12  III  III  III  13  III  III  III  14  III  III  III  15  III  III  IV  16  III  III  III

Surgical Simulation
We simulated drilling in four cases and osteochondral autograft transplantation in eleven cases. One case was indicated to perform free-body removal alone. All surgeries were conducted based on the simulations. We performed drilling for cases 1, 3, 11, and 15, free-body removal for cases 6 and 15, and costal osteochondral autograft transplantation for the remaining cases. Based on the MCFI, case 15 was indicated for articular surface reconstruction, but the patient did not consent to the surgical procedure and opted for the drilling procedure. We conducted all surgeries as simulated in all cases. Figure 10 shows the intraoperative findings and corresponding surgical simulation for representative cases.

Discussion
OCD lesion evaluation using MCFIs is a minimally invasive technique that enables prediction of lesion severity with high accuracy [28]. In developing this technique, we intended to maximize the advantages and compensate for the shortcomings of MRI and CT. The application of axial traction widens the joint space of the radio-capitellar joint and improves visualization of the articular cartilage outline of the humeral capitellum with minimum pain and discomfort [30,31], making the creation of accurate 3D models of the articular cartilage possible, and enabling images with a slice thickness of 0.4 mm to be obtained using a 3D sequence. We used a 7 kg traction weight according to previous studies [30,31]. There are no studies clarifying the ideal traction weight for elbow MRI in a skeletally immature population. Ideally, the traction weight must be decided on the basis of the body weight, size, and muscle development of the patient. Therefore, in the future, we will attempt to determine the ideal traction weight to lower discomfort during application of traction during MRI as much as possible.
The assessor can evaluate the 3D structure of the lesion from arbitrary angles using the MCFI. Additionally, it is possible to precisely obtain a positional relationship between the articular cartilage and the subchondral bone by adjusting the transparency of the articular cartilage, which is not visible to surgeons even during surgery. We accurately predicted the lesion severity in 15 out of 16 cases (93.8% accuracy). Although we predicted the case 15 lesion as ICRS class III, the lesion was diagnosed as class IV intraoperatively. This may be due to the high instability of the lesion, which could have caused the lesion to be displaced from the floor sometime between image acquisition and surgery. However, regarding the detection of unstable lesions, our technique achieved 100% accuracy. Diagnosing the stability of the lesion is extremely important in determining treatment strategies for patients with OCD [1,6,20]. Therefore, we believe that our technique is highly effective.
We simulated the surgical procedure in all cases and conducted surgery per the simulations in each case, which is another advantage of the MCFI. Among the surgical procedures we performed, articular surface reconstruction using a costal osteochondral autograft is the most complex surgery. Poor reconstruction of the articular surface leads to osteoarthritis in the future. However, owing to the anatomical feature of the elbow joint [35], articular surface reconstruction referring only to intraoperative findings can be challenging in some cases. Detailed surgical planning using MCFIs has the potential to minimize the process of intraoperative decision making by surgeons.
We selected costal osteochondral autograft transplantation for articular surface reconstruction because we are accustomed to the procedure; however, some surgeons prefer to harvest the osteochondral autograft from the knee using the OATS technique. Our surgical simulation technique is also applicable to surgeries that incorporate the OATS technique. When simulating the procedure, it is necessary to prepare a 3D model of a cylindrical autograft resembling the autograft harvested using the OATS technique instead of the 3D model of the costal osteochondral autograft. By preparing cylindrical autografts of various diameters, surgeons can precisely simulate the location and direction of the transplant, as well as the number of autografts necessary for the procedure (Figure 11).
A limitation of this technique is that the segmentation of the articular cartilage is time-consuming and technically demanding. Precise segmentation of the articular cartilage is the most crucial part of the procedure, but it must be partly executed manually. No available software can automatically and completely segment the articular cartilage, and it takes approximately 2 h per case to complete the procedure. Because we rely on the created MCFIs to determine the treatment strategies, it takes several weeks from the image acquisition to the surgery. This delay may lead to the deterioration of the OCD lesion in some cases, such as case 15, in which we misdiagnosed the ICRS classification. Therefore, we need to simplify the procedure and minimize the waiting period to avoid such consequences. In this study, evaluation of intraoperative findings was subjectively performed by each surgeon. We hope to achieve objective superposition of the MCFI and the surgical field in the future. Another limitation is that the number of cases was limited.
The accumulation of cases takes time as OCD is a rare condition. We will continue to work on the current study to further evaluate the consistency between intraoperative findings and MCFIs. Increasing the number of cases could lead to the development of software that can automatically create the MCFI for OCD patients. We also aim to conduct more quantitative evaluations of OCD lesions, including assessment of mild cases to predict the progression of lesion severity, in the future. In addition, we hope to apply this technique to other joints, with the aim of assisting surgeons in various fields treating intra-articular osteochondral lesions. A limitation of this technique is that the segme time-consuming and technically demanding. Precise lage is the most crucial part of the procedure, but it mu available software can automatically and completely it takes approximately 2 h per case to complete the created MCFIs to determine the treatment strategies, it acquisition to the surgery. This delay may lead to the Figure 11. Sample simulation of articular surface reconstruction using the osteochondral autograft transplantation system technique with three-dimensional (3D) models of cylindrical autografts for case 16. We created two 3D models of cylindrical autografts with diameters of 5 mm and 7 mm. The 3D models were positioned to cover the lesion to simulate the location, direction, and depth of the transplant. Here, the 3D CT model of the humerus, the 3D MRI model of the articular cartilage and 3D models of cylindrical autografts are shown in gray, yellow and green, respectively.

Conclusions
Herein, we introduced OCD lesion evaluation and surgical simulation using MCFIs of OCD, a novel tool developed by us. This computer-aided technique made the accurate evaluation of 3D structural details of the articular cartilage and subchondral bone possible, as well as lesion stability, without surgical intervention. A detailed surgical simulation is potentially useful for minimizing intraoperative decision making.

Patents
A patent for the 3D MRI-CT fusion image construction technique described in this article was applied for in Japan; 3D model generation method, 3D model generation device, and 3D model-generation program. Application No. 2018-066054, pending approval.