Introduction. Experimental bone research often generates large amounts of histology and histomorphometry data, and the analysis of these data can be time-consuming and trivial. Machine learning offers a viable alternative to manual analysis for measuring e.g. bone volume versus total volume. The objective was to develop a neural network for image segmentation, and to assess the accuracy of this network when applied to ectopic bone formation samples compared to a ground truth. Method. Thirteen tissue slides totaling 114 megapixels of ectopic bone formation were selected for model building. Slides were split into training, validation, and test data, with the test data reserved and only used for the final model assessment. We developed a neural network resembling U-Net that takes 512×512 pixel tiles. To improve model robustness, images were augmented online during training. The network was trained for 3 days on a NVidia Tesla K80 provided by a free online learning platform against ground truth masks annotated by an experienced researcher. Result. During training, the validation accuracy improved and stabilised at approx. 95%. The test accuracy was 96.1 %. Conclusion. Most experiments using ectopic bone formation will yield an inter-observer or inter-method variance of far more than 5%, so the current approach may be a valid and feasible technique for automated image segmentation for large datasets. More data or a consensus-based ground truth may improve training stability and validation accuracy. The code and data of this project are available upon request and will be available online as part of our publication

Chronic rotator cuff tears are a major problem especially in the elderly population. Refixation is associated with high re-rupture rates. Therefore new implants or healing methods are needed. For a control of success biomechanical characteristics of native as well as treated tendons are of particular importance. Currently, tensile tests with static material testing machines are the most common technique for the biomechanical characterization of tendons. Resulting values are the maximum force (Fmax), stiffness and the Young´s modulus. However, no information is given about the allocation of strains over the tendon area. In addition, the determination of Fmax results in tissue destruction thus foreclosing further evaluation like histology. The Digital Image Correlation (DIC) is a contact-free non-destructive optical measuring method which gives information about distribution of strains by tracking the areal shift of an applied speckle pattern. The needed speckle pattern has to have a high contrast, a homogeneous distribution and a good adhesion to the surface. The method is established for the characterization of construction materials [1] to detect e.g. weak points. The present study examined if DIC is applicable for the complementary biomechanical evaluation of the sheep infraspinatus tendon. Fine ground powder extracted from a printer cartridge was chosen as a starting point. Preliminary to the in vitro experiments, the powder was applied on sheets with different methods: brushing, blowing, sieving and stamping. Stamping showed best results and was used for further in vitro tests on cadaveric native tendons (n=5). First, the toner powder was transferred to coarse-grained abrasive paper using a brush and stamped on the tendon surface. Afterwards DIC analysis was performed. For the in vivo tests, the left infraspinatus tendon of two German black-headed Mutton Sheep was detached and then refixed with bone anchors, the right tendon was used as native control (authorization: AZ 33.19-42502-04-17/2739). 12 weeks after surgery the animals were euthanized, the shoulders were explanted and DIC measurement performed. The speckle pattern could be applied adequately on the smooth tendon surfaces of native tendons. All specimens could be analyzed by DIC with sufficient correlation coefficients. The highest displacements were measured in the peripheral areas, whereas the central part of the tendon showed a low displacement. Repaired left tendons showed obvious differences already macroscopically. The tendons were thicker and showed inhomogeneous surfaces. Application of the toner powder by stamping was distinctly more complicated, DIC analysis could not produce sufficient correlation coefficients. In summary, transfer of DIC to native infraspinatus tendons of sheep was successful and can be further transferred to other animal and human tendons. However, irregular surfaces in tendon scar tissues affect the application of an adequate speckle pattern with a stamp technique. Therefore, further modifications are necessary. This research project has been supported by the German Research Foundation “Graded Implants FOR 2180 – tendon- and bone junctions” WE 4262/6-1

Objectives. This study aimed to evaluate the histological and mechanical features of tendon healing in a rabbit model with second-harmonic-generation (SHG) imaging and tensile testing. Materials and Methods. A total of eight male Japanese white rabbits were used for this study. The flexor digitorum tendons in their right leg were sharply transected, and then were repaired by intratendinous stitching. At four weeks post-operatively, the rabbits were killed and the flexor digitorum tendons in both right and left legs were excised and used as specimens for tendon healing (n = 8) and control (n = 8), respectively. Each specimen was examined by SHG imaging, followed by tensile testing, and the results of the two testing modalities were assessed for correlation. Results. While the SHG light intensity of the healing tendon samples was significantly lower than that of the uninjured tendon samples, 2D Fourier transform SHG images showed a clear difference in collagen fibre structure between the uninjured and the healing samples, and among the healing samples. The mean intensity of the SHG image showed a moderate correlation (R. 2. = 0.37) with Young’s modulus obtained from the tensile testing. Conclusion. Our results indicate that SHG microscopy may be a potential indicator of tendon healing. Cite this article: E. Hase, K. Sato, D. Yonekura, T. Minamikawa, M. Takahashi, T. Yasui. Evaluation of the histological and mechanical features of tendon healing in a rabbit model with the use of second-harmonic-generation imaging and tensile testing. Bone Joint Res 2016;5:577–585. DOI: 10.1302/2046-3758.511.BJR-2016-0162.R1

Osteoarthritis is one of the most common musculoskeletal diseases. It involves degeneration and loss of articular cartilage, leading to a painful bone on bone articulation during movement. Numerical FEA models exist to predict the mechanical behaviour of degenerated cartilage. One of the limitations of these models arises from the poor validation that can be attained with traditional experimental data. This typically relies on comparison with global mechanical quantities such as total tissue strain, which mask the individual contributions originating from the different layers. In order to improve on this, an experimental method was developed to visualise the through-thickness behaviour of articular cartilage. Four experiments were performed on hemi-cylindrical cartilage plugs, harvested from a porcine femoral head, and immersed in a fluid solution. An Indian ink speckle pattern was applied to the flat surface of each hemi-cylinder. The specimens were equilibrated in 2.5M NaCl solution, transferred to a custom designed testing rig, and a reference image of the tissue cross-section was taken. The solution concentration was then decreased to 0.15M and, predictably, the tissue thickness changed. Images of the tissue cross section were taken every 60s for the duration of the experiment (3600s). All images were analysed using a DIC algorithm (Ncorr open-source 2D digital image correlation matlab program), and documented the strain changes through the tissue thickness as a function of time. The measured total strain in the tissue was consistent with that reported by Lai et al. (1991). However the present technique allows to quantify the strain contribution from any of the tissue layers or sublayer. This poses a significant advantage over traditional methods as resulting information can further the understanding of the factors contributing to the mechanical behaviour of the tissue and provides an ideal platform for validating more and more refined models of tissue behaviour

Background. Following an anterior cruciate ligament (ACL) injury, the affected knee is known to experience bone loss and is at significant risk of becoming osteoporotic. Surgical reconstruction is performed to attempt to restore the function of the knee and theoretically restore this bone density loss. Cross-sectional analysis of the proximal tibia using peripheral quantitative computed tomography (pQCT) enables localised analysis of bone mineral density (BMD) changes. The aim of this study was to establish the pattern of bone density changes in the tibia pre- and post- ACL reconstruction using pQCT image analysis. Methods. Eight patients who underwent ACL reconstruction were included. A cross sectional analysis of the proximal tibia was performed using a pQCT scanner pre-operatively and one to two years post-operatively on both the injured and contralateral (control) knee. The proximal two and three percent slices [S2 and S3] along the tibia were acquired. These were exported to Matlab(tm) and automated segmentation was performed to remove the tibia from its surrounding structures. Cross correlation was applied to co-register pairs of images and patterns of change in BMD were mapped using a t-test (p<0.05). Connected components of pixels with significant change in BMD were created and used to assess the impact of ACL injury & reconstruction on the proximal tibial BMD. Results. Prior to surgical ACL reconstruction, the BMD in the injured leg was significantly reduced relative to the control leg [S2: p=0.002, S3: p=0.002]. Post surgery, the proximal tibial BMD did not change in either leg [Control S2: p=0.102, S3: p=0.181; Injured S2: p=0.093, S3: p=0.439]. The post surgical images demonstrated patterns of increasing BMD surrounding the tunnel in the form of compact bone. Discussion. A significant reduction in proximal tibial BMD was observed in the ACL injured legs relative to control legs. The pattern of pre-operative bone loss was generally observed to be global across the entire slice. No change in BMD was observed following ACL reconstruction, in either injured or control leg. These results indicate that proximal tibial BMD is reduced and does not change after ACL reconstruction

Introduction. Needle guided biopsy of a suspected musculoskeletal malignancy has become increasingly popular as an effective modality for diagnosis. Biopsy performed in a safe manner should be performed in a centre which is also capable of performing the definitive management of such conditions. Our aim was to determine accuracy and success rates of the image guided biopsies performed by our service. Methods. A retrospective review of the Bone and Soft Tissue Sarcoma service database was performed to identify all patients who underwent diagnostic biopsy and to identify the results of such investigations. A biopsy was deemed successful if a sample of the target lesion was sampled at the time of biopsy. The successful biopsies were then classified as diagnostic or non-diagnostic if the diagnosis could be reached from the sampled tissue. Results. 465 of the 1181 new referrals to the Bone and Soft Tissue Sarcoma service in a 4 year period underwent biopsy. 75% (350) were image guided biopsies – 60% (281) ultrasound and 15% (69) CT guided. The rate of successful ultrasound guided biopsy was 94.7% and the rate of a successful diagnostic biopsy was 93.6%. CT guided biopsies were successful in 95.7% and were both successful and diagnostic in 79.7%. Discussion. The rate of a successful diagnostic ultrasound biopsy within our institution reflects the reported rate within the literature. The rate of a successful diagnostic CT guided biopsy is lower however is also consistent with that reported within the literature. Lipomatous and cartilaginous lesions are associated with a more difficult histological diagnosis on biopsy alone which is consistent with our findings. For this reason our institution has stopped performing routine image guided biopsies on these lesions

Summary. At the clinical CT image resolution level, there is no influence of the image voxel size on the derived finite element human cancellous bone models. Introduction. Computed tomography (CT)-based finite element (FE) models have been proved to provide a better prediction of vertebral strength than dual-energy x-ray absorptiometry [1]. FE models based on µCTs are able to provide the golden standard results [2], but due to the sample size restriction of the µCT and the XtremeCT machines, the clinical CT-based FE models is still the most promising tool for the in vivo prediction of vertebrae's strength. It has been found [3] that FE predicted Young's modulus of human cancellous bone increases as the image voxel size increases at the µCT resolution level [3]. However, it is still not clear whether the image voxel size in the clinical range has an impact on the predicted mechanical behavior of cancellous bone. This study is designed to answer this question. Methods. For this study, 6 thoracolumbar vertebrae (Th12) obtained from the female donors were scanned in the non-dissected cadavers under 2 different resolutions – group A: 120 kVp, 100 mAs, with a resolution of 0.29×0.29×1.3 mm. 3. ; group B: 120 kVp, 360 mAs, with a resolution of 0.18×0.18×0.6 mm. 3. A solid calibration phantom (QRM-BDC) was placed beneath the cadavers during the scans. Cuboids with the size of 12.3×12.3×14.3 mm. 3. were cropped from the center of each vertebral body. The FE model was created by converting each image voxel into hexahedron (C3D8). Inhomogeneous material property was defined for the cuboid [4], i.e. the image greyscale value were firstly calibrated into the bone mineral density (BMD), then the Young's modulus and yield stress were calculated from the BMD [5] for each element. Statistical analysis was performed to compare the FE predicted mechanical properties between the groups and the significance level was set to 95% (α=0.05). Results. The trabecular structure is more clearly mimicked in the models from group B than those from group A. The modulus (mean ± SE) in group A is 5.9% higher than that in group B (193.33 ± 31.67 MPa vs. 182.50 ± 27.07 MPa). The yield strength (mean ± SE) in group A is 6.4% higher than that in group B (0.99 ± 0.21MPa vs. 0.93 ± 0.17MPa). However, the paired t-test shows there is no significant difference of the mechanical properties in the two groups (p=0.109 for the modulus and p=0.234 for the yield strength). Discussion. This study shows that there is no influence of the voxel size on the clinical CT derived FE cancellous bone models. This finding can help choose a better, less invasive CT protocol for the patient when creating a clinical CT image based FE model. Acknowledgements. This study is financially supported by the Federal Ministry of Education and Research and the state of Hamburg, Germany

Several specimen specific vertebral (VB) models have been proposed in the literature; these replicate the typical set-up of a vertebral body mounted in bone cement and subject to a compressive ramp. VB and cement geometries are obtained from micro-CT images, the cement is typically assigned properties obtained from the literature while VB properties are inferred from the Hounsfield units- where the conversion factor between grayscale data and Young's modulus is optimised using experimental load-displacement data. Typically this calibration is performed on VBs dissected from the same spines as the study group. This, alongside the use of non-specific cement properties, casts some doubts on the predictivity of the models thus obtained. The predictivity of specimen specific FE models was evaluated in this study. VBs obtained from three porcine cervical segments (C2-C6) were stripped of all soft tissues, potted in bone cement and subject to a compressive loading ramp. A speckle pattern was applied to the anterior part of the specimen for DIC imaging. Specimen specific FE models were constructed from these specimens and a conversion factor between grayscale and material properties was optimised. Cement properties were assigned based on literature data. VBs from a further cervical spine (C2-C7) were subject to the same experimental protocol. In this case, the models generated from microCT images the material properties of bone were assigned based on the average conversion factor obtained previously. The predicted load-displacement behaviour thus obtained was compared to experimental data. Generally, poor agreement was found between overall load-displacement. The use of generic cement properties in the models was found to be partly responsible for this. When the load displacement behaviour of the VB was studied in isolation, good agreement within one standard deviation was found with 4 out of 6 models showing good correlation between simulation and DIC data

Digital image correlation (DIC) is rapidly increasing in popularity in biomechanical studies of the musculoskeletal system. DIC allows the re-construction of full field displacement and/or strain maps of the surface of an object. DIC systems typically consist of two cameras focussing on the same region of interest. This constrains the angle between the cameras to be relatively narrow when studying specimens characterised by complex geometrical features, giving rise to concerns on the accuracy of the out of plane estimates of movement. The aim of this research was to compare the movement profiles of bony segments measured by DIC and by an optoelectronic motion capture system. Five porcine cervical spine segments (C2-C6) were obtained from the local butcher. These were stripped of all anterior soft tissues while the posterior structures were left intact. A speckle pattern was applied to the anterior aspect of the specimens, while custom made infrared clusters were rigidly attached to the 3 middle vertebral bodies (C3-C5). The specimens were mounted in a custom made impact rig which fully constrained C6 but allowed C2 to translate in the axial direction of the segment. Images were acquired at 4kHz, both for the DIC (Photron Europe Ltd, UK) and motion capture cameras (Qualisys Oqus 400, Sweden). The in-plane and out of plane displacements of each of the VBs were plotted as a function of time and the similarity between the curves thus obtained was analysed using the SPM1D technique which allowed a comparison to be made in terms of t-statistics. No statistical differences were found between the two techniques in all axis of movement, however the out of plane movements were characterised by higher variance which is attributed to the uncertainty arising from the near parallel positioning of the cameras in the experimental set-up

We implanted titanium and carbon fibre-reinforced plastic (CFRP) femoral prostheses of the same dimensions into five prosthetic femora. An abductor jig was attached and a 1 kN load applied. This was repeated with five control femora. Digital image correlation was used to give a detailed two-dimensional strain map of the medial cortex of the proximal femur. Both implants caused stress shielding around the calcar. Distally, the titanium implant showed stress shielding, whereas the CFRP prosthesis did not produce a strain pattern which was statistically different from the controls. There was a reduction in strain beyond the tip of both the implants. This investigation indicates that use of the CFRP stem should avoid stress shielding in total hip replacement

Introduction. Cartilage damage is a critical aspect of osteoarthritis progression, but effective imaging strategies remain limited. Consequently, multimodal imaging approaches are receiving increased attention. Gold nanomaterials, renowned for their therapeutic and imaging capabilities, hold promise in drug development. However, their potential for cartilage imaging is rarely discussed. Here, we developed a versatile nanomaterial, AuNC@BSA-Gd-I, for cartilage detection. By leveraging electrostatic interactions with sulfated glycosaminoglycans (sGAG), the AuNC@BSA-Gd-I can effectively penetrate damaged cartilage while accumulating minimally in healthy cartilage. This probe can be visualized or detected using CT, MRI, IVIS, and a gamma counter, providing a comprehensive approach to cartilage imaging. Additionally, we compared the imaging abilities, cartilage visualization capacities, and versatility of currently disclosed multimodal gold nanomaterials with those of AuNC@BSA-Gd-I. Method. The physicochemical properties of nanomaterials were measured. The potential for cartilage visualization of these nanomaterials was assessed using an in vitro porcine model. The sGAG content in cartilage was determined using the dimethylmethylene blue (DMMB) assay to establish the correlation between sGAG concentration and imaging intensity acquired at each modality. Results. The cartilage imaging abilities of AuNC@BSA-Gd-I for CT, MRI, and optical imaging were verified, with each imaging intensity demonstrating a strong correlation with the sGAG content (MRI; R2=0.93, CT; R2=0.83, IVIS; R2=0.79). Furthermore, AuNC@BSA-Gd-. 131. I effectively accumulated in defective cartilage tissue compared to healthy cartilage (23755.38 ± 5993.61 CPM/mg vs. 11699.97 ± 794.93 CPM/mg). Additionally, current gold nanomaterials excelled in individual imaging modalities but lacked effective multimodal imaging ability. Conclusion. Compared to current multimodal gold nanomaterials, AuNC@BSA-Gd-I demonstrates the potential to image cartilage across multiple medical instruments, providing investigators with a more powerful, visible, and convenient approach to detect cartilage defects. Acknowledgements. This work was financially supported by the National Health Research Institute, Taiwan (NHRI-EX112-11029SI), the National Science and Technology Council (NSTC 112-2314-B-182A-105-MY3), and Chang Gung Memorial Hospital, Taiwan (CMRPG8N0781 and CMRPG8M1281-3)

Imaging can provide valuable information about the function of tissues and organs. The capacity for detecting and measuring imaging biomarkers of biological activities, allows for a better understanding of the pathophysiology of any process in the human body, including the musculoskeletal system. This is of particular importance in oncologic, metabolic and rheumatologic diseases, but not limited to these. In the domain of the musculoskeletal system, functional imaging also means to be able to address biomechanical evaluations. Weight-bearing imaging and dynamic studies have a prominent role. All imaging techniques (X-rays, CT, MR, ultrasound) are in demand, and offer different applications, specific equipment and novel methods for addressing this. Functional imaging is also essential to drive minimally invasive treatments – i.e. interventional radiology, and new treatment approaches move together with the advances on imaging guidance methods. On both the diagnostic and the interventional side, the increasing availability of dedicated equipment and the development of specific imaging methods and protocols greatly helps the transition from research to clinical practice

Knee pain is common, representing a significant socioeconomic burden. Caused by a variety of pathologies, its evaluation in primary-care is challenging. Subsequently, an over-reliance on magnetic resonance imaging (MRI) exists. Prior to orthopaedic surgeon referral, many patients receive no, or incorrect, imaging. Electronic-triage (e-triage) tools represent an innovative solution to address this problem. The primary aim of this study was to ascertain whether an e-triage tool is capable of outperforming existing clinical pathways to determine the correct pre-hospital imaging based on knee pain diagnosis. Patients ≥18 years with a new presentation of knee pain were retrospectively identified. The timing and appropriateness of imaging was assessed. A symptom-based e-triage tool was developed, using the Amazon LEXbotplatform, and piloted to predict five common knee pathologies and suggest appropriate imaging. 1462 patients were identified. 17% of arthroplasty patients received an ‘unnecessary MRI’, whilst 28% of arthroscopy patients did not have a ‘necessary MRI’, thus requiring a follow-up appointment, with a mean delay of three months (SD 2.6, range 0.2-20.2). Using NHS tariffs, a wasted cost through unnecessary/necessary MRIs and subsequent follow-up appointments was estimated at £45,816. The e-triage pilot was trialled with 41 patients (mean age:58.4 years, 58.5% female). Preliminary diagnoses were available for 34 patients. Using the highest proportion of reported symptoms in the corresponding group, the e-triage tool correctly identified three of the four knee pathologies. The e-triage tool did not correctly identify anterior cruciate ligament injuries (n=3). 79.2% of participants would use the tool again. A significant number of knee pathology patients received incorrect imaging prior to their initial hospital appointment, incurring delays and unnecessary costs. A symptom-based e-triage tool was developed, with promising pilot data and user feedback. With refinement, this tool has the potential to improve wait-times and referral quality, whilst reducing costs

To detect early signs of infection infrared thermography has been suggested to provide quantitative information. Our vision is to invent a pin site infection thermographic surveillance tool for patients at home. A preliminary step to this goal is the aim of this study, to automate the process of locating the pin and detecting the pin sites in thermal images efficiently, exactly, and reliably for extracting pin site temperatures. A total of 1708 pin sites was investigated with Thermography and augmented by 9 different methods in to totally 10.409 images. The dataset was divided into a training set (n=8325), a validation set (n=1040), and a test set (n=1044) of images. The Pin Detection Model (PDM) was developed as follows: A You Only Look Once (YOLOv5) based object detection model with a Complete Detection Intersection over Union (CDIoU), it was pre-trained and finetuned by the through transfer learning. The basic performance of the YOLOv5 with CDIoU model was compared with other conventional models (FCOS and YOLOv4) for deep and transition learning to improve performance and precision. Maximum Temperature Extraction (MTE) Based on Region of Interest (ROI) for all pin sites was generated by the model. Inference of MTE using PDM with infected and un-infected datasets was investigated. An automatic tool that can identify and annotate pin sites on conventional images using bounding boxes was established. The bounding box was transferred to the infrared image. The PMD algorithm was built on YOLOv5 with CDIoU and has a precision of 0.976. The model offers the pin site detection in 1.8 milliseconds. The thermal data from ROI at the pin site was automatically extracted. These results enable automatic pin site annotation on thermography. The model tracks the correlation between temperature and infection from the detected pin sites and demonstrates it is a promising tool for automatic pin site detection and maximum temperature extraction for further infection studies. Our work for automatic pin site annotation on thermography paves the way for future research on infection assessment using thermography

Bone metastases radiographically appear as regions with high (i.e. blastic metastases) or low (i.e. lytic metastases) bone mineral density. The clinical assessment of metastatic features is based on computed tomography (CT) but it is still unclear if the actual size of the metastases can be accurately detected from the CT images and if the microstructure in regions surrounding the metastases is altered (Nägele et al., 2004, Calc Tiss Int). This study aims to evaluate (i) the capability of the CT in evaluating the metastases size and (ii) if metastases affect the bone microstructure around them. Ten spine segments consisted of a vertebra with lytic or mixed metastases and an adjacent control (radiologically healthy) were obtained through an ethically approved donation program. The specimens were scanned with a clinical CT (AquilionOne, Toshiba: slice thickness:1mm, in-plane resolution:0.45mm) to assess clinical metastatic features and a micro-CT (VivaCT80, Scanco, isotropic voxel size:0.039mm) to evaluate the detailed microstructure. The volume of the metastases was measured from both CT and micro-CT images (Palanca et al., 2021, Bone) and compared with a linear regression. The microstructural alteration around the metastases was evaluated in the volume of interest (VOI) defined in the micro-CT images as the volume of the vertebral body excluding the metastases. Three 3D microstructural parameters were calculated in the VOI (CTAn, Bruker SkyScan): Bone Volume Fraction (BV/TV), Trabecular Thickness (Tb.Th.), Trabecular Spacing (Tb.Sp.). Medians of each parameter were compared (Kruskal-Wallis, p=0.05). One specimen was excluded as it was not possible to define the size of the metastases in the CT scans. A strong correlation between the volume obtained from the CT and micro-CT images was found (R2=0.91, Slope=0.97, Intercept=2.55, RMSE=5.7%, MaxError=13.12%). The differences in BV/TV, Tb.Th. and Tb.Sp. among vertebrae with lytic and mixed metastases and control vertebrae were not statistically significant (p-value>0.6). Similar median values of BV/TV were found in vertebrae with lytic (13.2±2.4%) and mixed (12.8±9.8%) metastases, and in controls (13.0±10.1%). The median Tb.Th. was 176±18 ∓m, 179±43 ∓m and 167±91 ∓m in vertebrae with lytic and mixed metastases and control vertebrae, respectively. The median Tb. Sp. was 846±26 ∓m, 849±286 ∓m and 880±116 ∓m in vertebrae with lytic and mixed metastases and control vertebrae, respectively. In conclusion, the size of vertebral metastases can be accurately assess using CT images. The 3D microstructural parameters measured were comparable with those reported in the literature for healthy vertebrae (Nägele et al., 2004, Calc Tiss Int, Sone et al., 2004, Bone) and showed how the microstructure of the bone tissue surrounding the lesion is not altered by the metastases

The most important outcome predictor of Legg-Calvé-Perthes disease (LCPD) is the shape of the healed femoral head. However, the deformity of the femoral head is currently evaluated by non-reproducible, categorical, and qualitative classifications. In this regard, recent advances in computer vision might provide the opportunity to automatically detect and delineate the outlines of bone in radiographic images for calculating a continuous measure of femoral head deformity. This study aimed to construct a pipeline for accurately detecting and delineating the proximal femur in radiographs of LCPD patients employing existing algorithms. To detect the proximal femur, the pretrained stateof-the-art object detection model, YOLOv5, was trained on 1580 manually annotated radiographs, validated on 338 radiographs, and tested on 338 radiographs. Additionally, 200 radiographs of shoulders and chests were added to the dataset to make the model more robust to false positives and increase generalizability. The convolutional neural network architecture, U-Net, was then employed to segment the detected proximal femur. The network was trained on 80 manually annotated radiographs using real-time data augmentation to increase the number of training images and enhance the generalizability of the segmentation model. The network was validated on 60 radiographs and tested on 60 radiographs. The object detection model achieved a mean Average Precision (mAP) of 0.998 using an Intersection over Union (IoU) threshold of 0.5, and a mAP of 0.712 over IoU thresholds of 0.5 to 0.95 on the test set. The segmentation model achieved an accuracy score of 0.912, a Dice Coefficient of 0.937, and a binary IoU score of 0.854 on the test set. The proposed fully automatic proximal femur detection and segmentation system provides a promising method for accurately detecting and delineating the proximal femoral bone contour in radiographic images, which is necessary for further image analysis

Introduction. Anterior shoulder instability results in labral and osseous glenoid injuries. With a large osseous defect, there is a risk of recurrent dislocation of the joint, and therefore the patient must undergo surgical correction. An MRI evaluation of the patient helps to assess the soft tissue injury. Currently, the volumetric three-dimensional (3D) reconstructed CT image is the standard for measuring glenoid bone loss and the glenoid index. However, it has the disadvantage of exposing the patient to radiation and additional expenses. This study aims to compare the values of the glenoid index using MRI and CT. Method. The present study was a two-year cross-sectional study of patients with shoulder pain, trauma, and dislocation in a tertiary hospital in Karnataka. The sagittal proton density (PD) section of the glenoid and enface 3D reconstructed images of the scapula were used to calculate glenoid bone loss and the glenoid index. The baseline data were analyzed using descriptive statistics, and the Chi-square test was used to test the association of various complications with selected variables of interest. Result. The glenoid index calculated in the current study using 3D volumetric CT images and MR sagittal PD images was 0.95±0.01 and 0.95±0.01, respectively. The CT and MRI glenoid bone loss was 5.41±0.65% and 5.38±0.65%, respectively. When compared, the glenoid index and bone loss calculated by MRI and CT revealed a high correlation and significance with a p-value of <0.001. Conclusions. The study concluded that MRI is a reliable method for glenoid measurement. The sagittal PD sequence combined with an enface glenoid makes it possible to identify osseous defects linked to glenohumeral joint damage and dislocation. The values derived from 3D CT are identical to the glenoid index and bone loss determined using the sagittal PD sequence in MRI

Neck of femur fractures are a common trauma presentation and patients with a history of malignancy are sent for long leg femur views (LLF), to exclude a distal lesion which would alter the management plan (Intra-medullary nail/Long stem Hemiarthroplasty). The aim of this is to identify incidence of malignancy on LLF views, the length of time in between each xray (XR) and to identify demographics. Data was retrospectively collected from 01/01/2021 to 31/01/2021 from a single centre. All patients admitted to the Queen Elizabeth University Hospital had their electronic records (Bluespier, PACS, Clinical Portal) accessed. These confirmed if patients had a past medical history of malignancy, if they had LLF view and the time differences between diagnostic pelvis XR and LLF XR. A total of 784 patients were identified in the specified time period. Of these, 138 were identified with a malignancy and there were 85 LLF views completed. LLF views diagnosed 1 patient with known prostate cancer that had a new distal femoral metastasis (Incidence = 1.28 cases per 1000). This patient underwent further imaging (MRI Femur) and received a long stem hip hemiarthroplasty. The average length of wait between the images was 9 hours 27 minutes. LLF views can alter management of patients with malignancy and are therefore useful to perform. There can be a long delay between each image. Therefore we recommend imaging tumour with common bony metastasis (Renal, Thyroid, Breast, Prostrate, Lung) and other remaining tumours with known secondary metastasis. Imaging primary low risk (eg basal cell carcinoma) can lead to long delays in a frail patient cohort and consideration should be given to rationalise appropriate use of resources

Acute syndesmotic ankle injuries continue to impose a diagnostic dilemma and it remains unclear whether weighbearing or external rotation should be exerted rotation during the imaging process. Therefore, we aimed to implement both axial load (weightbearing) and external rotation in the assessment of a clinical cohort of patients with syndesmotic ankle injuries syndesmotic using weightbearing CT imaging. In this retrospective comparative cohort study, patients with an acute syndesmotic ankle injury were analyzed using a WBCT (N= 20; Mean age= 31,64 years; SD= 14,07. Inclusion criteria were an MRI confirmed syndesmotic ankle injury imaged by a bilateral WBCT of the ankle during weightbearing and combined weightbearing-external rotation. Exclusion criteria consisted of fracture associated syndesmotic ankle injuries. Three-dimensional (3D) models were generated from the CT slices. Tibiofibular displacement and Talar Rotation was quantified using automated3D measurements (Anterior TibioFibular Distance (ATFD), Alpha Angle, Posterior TibioFibular Distance (PTFD) and Talar Rotation (TR) Angle) in comparison to a cohort of non-injured ankles. Results. The difference in neutral-stressed Alpha° and ATFD showed a significant difference between patients with a syndesmotic ankle lesion and healthy ankles (P = 0.046 and P = 0.039, respectively) The difference in neutral-stressed PTFD and TR° did not show a significant difference between patients with a syndesmotic ankle lesion and healthy ankles (P = 0.492; P = 0.152, respectively). Conclusion. Application of combined weightbearing-external rotation reveals a dynamic anterior tibiofibular widening in patients with syndesmotic ankle injuries. This study provides the first insights based on 3D measurements to support the potential relevance of applying external rotation during WBCT imaging. However, to what extent certain displacement patterns are associated with syndesmotic instability and thus require operative treatment strategies has yet to be determined in future studies

Abstract. Objectives. Spinal disorders such as back pain incur a substantial societal and economic burden. Unfortunately, there is lack of understanding and treatment of these disorders are further impeded by the inability to assess spinal forces in vivo. The aim of this project is to address this challenge by developing and testing a novel image-driven approach that will assess the forces in an individual's spine in vivo by incorporating information acquired from multimodal imaging (magnetic resonance imaging (MRI) and biplane X-rays) in a subject-specific model. Methods. Magnetic resonance and biplane X-ray imaging are used to capture information about the anatomy, tissues, and motion of an individual's spine as they perform a range of everyday activities. This information is then utilised in a subject-specific computational model based on the finite element method to predict the forces in their spine. The project is also utilising novel machine learning algorithms and in vitro, six-axis mechanical testing on human, porcine and bovine samples to develop and test the modelling methods rigorously. Results & Discussion. MRI sequences have been identified that provide high-quality image data and information on different tissue types which will be used to predict subject-specific disc properties. In-vivo protocols to capture motion analysis, EMG muscle activity, and video X-rays of the spine have been designed with planned data collection of 15 healthy volunteers. Preliminary modelling work has evaluated potential machine learning approaches and quantified the sensitivity of the models developed to material properties. Conclusion. The development and testing of these image-driven subject-specific spine models will provide a new tool for determining forces in the spine. It will also provide new tools for measuring and modelling spine movement and quantifying the properties of the spinal tissues. Acknowledgments. Funding from the EPSRC: EP/V036602/1 (Meakin, Holsgrove & Javadi) and EP/V032275/1 (Holt & Williams). Declaration of Interest. (b) declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported:I declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research project