Development and Validation of a Multitask Deep Learning Model for Severity Grading of Hip Osteoarthritis Features on Radiographs

3D U-Net Neural Network Architecture-Assisted LDCT to Acquire Vertebral Morphology Parameters: A Vertebral Morphology Comprehensive Analysis in a Chinese Population

To evaluate the feasibility of acquiring vertebral height from chest low-dose computed tomography (LDCT) images using an artificial intelligence (AI) system based on 3D U-Net vertebral segmentation technology and the correlation and features of vertebral morphology with sex and age of the Chinese population. Patients who underwent chest LDCT between September 2020 and April 2023 were enrolled. The Altman and Pearson's correlation analyses were used to compare the correlation and consistency between the AI software and manual measurement of vertebral height. The anterior height (Ha), middle height (Hm), posterior height (Hp), and vertebral height ratios (VHRs) (Ha/Hp and Hm/Hp) were measured from T1 to L2 using an AI system. The VHR is the ratio of Ha to Hp or the ratio of Hm to Hp of the vertebrae, which can reflect the shape of the anterior wedge and biconcave vertebrae. Changes in these parameters, particularly the VHR, were analysed at different vertebral levels in different age and sex groups. The results of the AI methods were highly consistent and correlated with manual measurements. The Pearson's correlation coefficients were 0.855, 0.919, and 0.846, respectively. The trend of VHRs showed troughs at T7 and T11 and a peak at T9; however, Hm/Hp showed slight fluctuations. Regarding the VHR, significant sex differences were found at L1 and L2 in all age bands. This innovative study focuses on vertebral morphology for opportunistic analysis in the mainland Chinese population and the distribution tendency of vertebral morphology with ageing using a chest LDCT aided by an AI system based on 3D U-Net vertebral segmentation technology. The AI system demonstrates the potential to automatically perform opportunistic vertebral morphology analyses using LDCT scans obtained during lung cancer screening. We advocate the use of age-, sex-, and vertebral level-specific criteria for the morphometric evaluation of vertebral osteoporotic fractures for a more accurate diagnosis of vertebral fractures and spinal pathologies.

Hallux valgus and pes planus: Correlation analysis using deep learning-assisted radiographic angle measurements

BACKGROUND

The relationship between hallux valgus (HV) and pes planus remains unresolved. This study aims to determine the correlation between HV and pes planus using a deep learning (DL) model to measure radiographic angle parameters.

METHODS

In total, radiographs of 212 feet detectable by the DL model were analyzed. HV was evaluated using the hallux valgus and intermetatarsal angles, while pes planus was assessed using the lateral talo-first metatarsal (Meary's) and calcaneal pitch angles. Correlation analyses were performed for each DL model-measured angle parameter. We investigated whether pes planus worsened with increasing severity of HV and vice versa.

RESULTS

All parameters were significantly correlated with each other. Pes planus worsened with increasing severity of HV, and as the severity of pes planus increased, HV also worsened.

CONCLUSION

Utilizing the DL model-assisted radiographic angle measurements, this study established a significant correlation between HV and pes planus.

LEVEL OF EVIDENCE

III.

Artificial intelligence in musculoskeletal imaging: realistic clinical applications in the next decade

This article will provide a perspective review of the most extensively investigated deep learning (DL) applications for musculoskeletal disease detection that have the best potential to translate into routine clinical practice over the next decade. Deep learning methods for detecting fractures, estimating pediatric bone age, calculating bone measurements such as lower extremity alignment and Cobb angle, and grading osteoarthritis on radiographs have been shown to have high diagnostic performance with many of these applications now commercially available for use in clinical practice. Many studies have also documented the feasibility of using DL methods for detecting joint pathology and characterizing bone tumors on magnetic resonance imaging (MRI). However, musculoskeletal disease detection on MRI is difficult as it requires multi-task, multi-class detection of complex abnormalities on multiple image slices with different tissue contrasts. The generalizability of DL methods for musculoskeletal disease detection on MRI is also challenging due to fluctuations in image quality caused by the wide variety of scanners and pulse sequences used in routine MRI protocols. The diagnostic performance of current DL methods for musculoskeletal disease detection must be further evaluated in well-designed prospective studies using large image datasets acquired at different institutions with different imaging parameters and imaging hardware before they can be fully implemented in clinical practice. Future studies must also investigate the true clinical benefits of current DL methods and determine whether they could enhance quality, reduce error rates, improve workflow, and decrease radiologist fatigue and burnout with all of this weighed against the costs.

Automatic detection, segmentation, and classification of primary bone tumors and bone infections using an ensemble multi-task deep learning framework on multi-parametric MRIs: a multi-center study

OBJECTIVES

To develop an ensemble multi-task deep learning (DL) framework for automatic and simultaneous detection, segmentation, and classification of primary bone tumors (PBTs) and bone infections based on multi-parametric MRI from multi-center.

METHODS

This retrospective study divided 749 patients with PBTs or bone infections from two hospitals into a training set (N = 557), an internal validation set (N = 139), and an external validation set (N = 53). The ensemble framework was constructed using T1-weighted image (T1WI), T2-weighted image (T2WI), and clinical characteristics for binary (PBTs/bone infections) and three-category (benign/intermediate/malignant PBTs) classification. The detection and segmentation performances were evaluated using Intersection over Union (IoU) and Dice score. The classification performance was evaluated using the receiver operating characteristic (ROC) curve and compared with radiologist interpretations.

RESULT

On the external validation set, the single T1WI-based and T2WI-based multi-task models obtained IoUs of 0.71 ± 0.25/0.65 ± 0.30 for detection and Dice scores of 0.75 ± 0.26/0.70 ± 0.33 for segmentation. The framework achieved AUCs of 0.959 (95%CI, 0.955-1.000)/0.900 (95%CI, 0.773-0.100) and accuracies of 90.6% (95%CI, 79.7-95.9%)/78.3% (95%CI, 58.1-90.3%) for the binary/three-category classification. Meanwhile, for the three-category classification, the performance of the framework was superior to that of three junior radiologists (accuracy: 65.2%, 69.6%, and 69.6%, respectively) and comparable to that of two senior radiologists (accuracy: 78.3% and 78.3%).

CONCLUSION

The MRI-based ensemble multi-task framework shows promising performance in automatically and simultaneously detecting, segmenting, and classifying PBTs and bone infections, which was preferable to junior radiologists.

CLINICAL RELEVANCE STATEMENT

Compared with junior radiologists, the ensemble multi-task deep learning framework effectively improves differential diagnosis for patients with primary bone tumors or bone infections. This finding may help physicians make treatment decisions and enable timely treatment of patients.

KEY POINTS

• The ensemble framework fusing multi-parametric MRI and clinical characteristics effectively improves the classification ability of single-modality models. • The ensemble multi-task deep learning framework performed well in detecting, segmenting, and classifying primary bone tumors and bone infections. • The ensemble framework achieves an optimal classification performance superior to junior radiologists' interpretations, assisting the clinical differential diagnosis of primary bone tumors and bone infections.

Artificial intelligence can be used in the identification and classification of shoulder osteoarthritis and avascular necrosis on plain radiographs: a training study of 7,139 radiograph sets

BACKGROUND AND PURPOSE

Knowledge concerning the use AI models for the classification of glenohumeral osteoarthritis (GHOA) and avascular necrosis (AVN) of the humeral head is lacking. We aimed to analyze how a deep learning (DL) model trained to identify and grade GHOA on plain radiographs performs. Our secondary aim was to train a DL model to identify and grade AVN on plain radiographs.

PATIENTS AND METHODS

A modified ResNet-type network was trained on a dataset of radiographic shoulder examinations from a large tertiary hospital. A total of 7,139 radiographs were included. The dataset included various projections of the shoulder, and the network was trained using stochastic gradient descent. Performance evaluation metrics, area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were used to assess the network's performance for each outcome.

RESULTS

The network demonstrated AUC values ranging from 0.73 to 0.93 for GHOA classification and > 0.90 for all AVN classification classes. The network exhibited lower AUC for mild cases compared with definitive cases of GHOA. When none and mild grades were combined, the AUC increased, suggesting difficulties in distinguishing between these 2 grades.

CONCLUSION

We found that a DL model can be trained to identify and grade GHOA on plain radiographs. Furthermore, we show that a DL model can identify and grade AVN on plain radiographs. The network performed well, particularly for definitive cases of GHOA and any level of AVN. However, challenges remain in distinguishing between none and mild GHOA grades.

AI-assisted Analysis to Facilitate Detection of Humeral Lesions on Chest Radiographs

Purpose To develop an artificial intelligence (AI) system for humeral tumor detection on chest radiographs (CRs) and evaluate the impact on reader performance. Materials and Methods In this retrospective study, 14 709 CRs (January 2000 to December 2021) were collected from 13 468 patients, including CT-proven normal (n = 13 116) and humeral tumor (n = 1593) cases. The data were divided into training and test groups. A novel training method called false-positive activation area reduction (FPAR) was introduced to enhance the diagnostic performance by focusing on the humeral region. The AI program and 10 radiologists were assessed using holdout test set 1, wherein the radiologists were tested twice (with and without AI test results). The performance of the AI system was evaluated using holdout test set 2, comprising 10 497 normal images. Receiver operating characteristic analyses were conducted for evaluating model performance. Results FPAR application in the AI program improved its performance compared with a conventional model based on the area under the receiver operating characteristic curve (0.87 vs 0.82, P = .04). The proposed AI system also demonstrated improved tumor localization accuracy (80% vs 57%, P < .001). In holdout test set 2, the proposed AI system exhibited a false-positive rate of 2%. AI assistance improved the radiologists' sensitivity, specificity, and accuracy by 8.9%, 1.2%, and 3.5%, respectively (P < .05 for all). Conclusion The proposed AI tool incorporating FPAR improved humeral tumor detection on CRs and reduced false-positive results in tumor visualization. It may serve as a supportive diagnostic tool to alert radiologists about humeral abnormalities. Keywords: Artificial Intelligence, Conventional Radiography, Humerus, Machine Learning, Shoulder, Tumor Supplemental material is available for this article. © RSNA, 2024.

Cohort profile: Worldwide Collaboration on OsteoArthritis prediCtion for the Hip (World COACH) - an international consortium of prospective cohort studies with individual participant data on hip osteoarthritis

PURPOSE

Hip osteoarthritis (OA) is a major cause of pain and disability worldwide. Lack of effective therapies may reflect poor knowledge on its aetiology and risk factors, and result in the management of end-stage hip OA with costly joint replacement. The Worldwide Collaboration on OsteoArthritis prediCtion for the Hip (World COACH) consortium was established to pool and harmonise individual participant data from prospective cohort studies. The consortium aims to better understand determinants and risk factors for the development and progression of hip OA, to optimise and automate methods for (imaging) analysis, and to develop a personalised prediction model for hip OA.

PARTICIPANTS

World COACH aimed to include participants of prospective cohort studies with ≥200 participants, that have hip imaging data available from at least 2 time points at least 4 years apart. All individual participant data, including clinical data, imaging (data), biochemical markers, questionnaires and genetic data, were collected and pooled into a single, individual-level database.

FINDINGS TO DATE

World COACH currently consists of 9 cohorts, with 38 021 participants aged 18-80 years at baseline. Overall, 71% of the participants were women and mean baseline age was 65.3±8.6 years. Over 34 000 participants had baseline pelvic radiographs available, and over 22 000 had an additional pelvic radiograph after 8-12 years of follow-up. Even longer radiographic follow-up (15-25 years) is available for over 6000 of these participants.

FUTURE PLANS

The World COACH consortium offers unique opportunities for studies on the relationship between determinants/risk factors and the development or progression of hip OA, by using harmonised data on clinical findings, imaging, biomarkers, genetics and lifestyle. This provides a unique opportunity to develop a personalised hip OA risk prediction model and to optimise methods for imaging analysis of the hip.

Multicentric development and validation of a multi-scale and multi-task deep learning model for comprehensive lower extremity alignment analysis

Osteoarthritis of the knee, a widespread cause of knee disability, is commonly treated in orthopedics due to its rising prevalence. Lower extremity misalignment, pivotal in knee injury etiology and management, necessitates comprehensive mechanical alignment evaluation via frequently-requested weight-bearing long leg radiographs (LLR). Despite LLR's routine use, current analysis techniques are error-prone and time-consuming. To address this, we conducted a multicentric study to develop and validate a deep learning (DL) model for fully automated leg alignment assessment on anterior-posterior LLR, targeting enhanced reliability and efficiency. The DL model, developed using 594 patients' LLR and a 60%/10%/30% data split for training, validation, and testing, executed alignment analyses via a multi-step process, employing a detection network and nine specialized networks. It was designed to assess all vital anatomical and mechanical parameters for standard clinical leg deformity analysis and preoperative planning. Accuracy, reliability, and assessment duration were compared with three specialized orthopedic surgeons across two distinct institutional datasets (136 and 143 radiographs). The algorithm exhibited equivalent performance to the surgeons in terms of alignment accuracy (DL: 0.21 ± 0.18°to 1.06 ± 1.3°vs. OS: 0.21 ± 0.16°to 1.72 ± 1.96°), interrater reliability (ICC DL: 0.90 ± 0.05 to 1.0 ± 0.0 vs. ICC OS: 0.90 ± 0.03 to 1.0 ± 0.0), and clinically acceptable accuracy (DL: 53.9%-100% vs OS 30.8%-100%). Further, automated analysis significantly reduced analysis time compared to manual annotation (DL: 22 ± 0.6 s vs. OS; 101.7 ± 7 s, p ≤ 0.01). By demonstrating that our algorithm not only matches the precision of expert surgeons but also significantly outpaces them in both speed and consistency of measurements, our research underscores a pivotal advancement in harnessing AI to enhance clinical efficiency and decision-making in orthopaedics.

Three-Dimensional Reconstruction of Computed Tomography Imaging Is Not Reliable in Assessing Acetabular Rim Osteophytes or Acetabular Rim Pathology in Patients With Femoroacetabular Impingement

Purpose

To determine the reliability of 3-dimensional (3D) reconstruction of computed tomography (CT) imaging in evaluating acetabular rim morphology or acetabular rim osteophyte (ARO) existence and to group patients with femoroacetabular impingement (FAI) by ARO extent on coronal sections of CT and further compare clinical differences among groups.

Methods

Patients who underwent primary hip arthroscopy for FAI by the same surgeon between August 2016 and December 2018 with minimum 2-year follow-up were enrolled. The ARO was evaluated both on the acetabular gross anatomy (AGA) and coronal sections of CT, for its position, width (unit: mm), area (unit: mm2), and CT value (unit: HU). Patients were divided into 4 groups based on the extent of ARO on coronal CT: group A (ARO anterior to 12 o'clock), group P (ARO posterior to 12 o'clock), group AP (ARO across 12 o'clock), and group N (no ARO). Inter- and intraobserver correlation was analyzed. Demographic data, FAI deformity indicators on imaging, quantitative measurements of ARO, and pre- and postoperative patient-reported outcomes were compared among groups.

Results

There were 229 patients (229 hips) enrolled in total, 122 male (53.3%) and 107 female (46.7%), with a mean age of 37.2 ± 10.2 years. The correlation between 2 observers for grouping ARO using AGA was positive but poor (Kendall Tau-b coefficient = 0.157, P = .008). Moderate correlation was found between grouping based on AGA and coronal CT by the same observer (Kendall Tau-b coefficient = 0.482, P = .000). The patients were divided into 4 groups: 84 patients (36.7%) in group N, 2 patients (0.9%) in group A, 69 patients (30.1%) in group P, and 74 patients (32.3%) in group AP. Group N was younger in age (35.4 ± 10.7 years) than group P (39.6 ± 10.2 years) (P = 0.012) and had a larger proportion of women (57.1%) than group AP (36.5%) (χ2 = 6.869, P = .032). There was a greater proportion of positive posterior wall sign in group P (52.2%) than 48.6% for group AP and 33.3% for group N (χ2 = 6.397, P = .041). Group N had 61 (72.6%) Tönnis grade 0 hips compared with 37 (50%) in group AP (P = .014). No statistical significance was found among groups in pre- and postoperative α angle, lateral center-edge angle, and patient-reported outcomes. The widths of ARO in group AP for the 3 marked points from anterior to posterior were 3.88 ± 1.86, 4.84 ± 2.72, and 6.66 ± 3.18, separately (P＜.001); 15.73 ± 21.46, 19.22 ± 18.86, and 29.96 ± 17.05 for area (P＜.01); and 652.67 ± 214.12, 677.10 ± 274.81, and 728.84 ± 232.39 for CT value (P＜.05). For the ARO posterior to 12 o'clock, the group AP showed a larger width (6.66 ± 3.18), area (29.96 ± 17.05), and CT value (728.84 ± 232.39) than group P of (4.70 ± 2.25), (20.15 ± 12.91), and (641.84 ± 183.33) (P＜.001).

Conclusions

The evaluation of ARO on AGA is poor consistent with definite double-rim sign on coronal CT. There is a tendency of size-enlarging and density-increasing for ARO from anterior to posterior along the acetabular rim. Younger age, female gender, lower Tönnis grade, and negative posterior wall sign showed lower rate of ARO development.

Level of Evidence

Level IV, diagnostic case series.

Multitask deep learning on mammography to predict extensive intraductal component in invasive breast cancer

OBJECTIVES

To develop a multitask deep learning (DL) algorithm to automatically classify mammography imaging findings and predict the existence of extensive intraductal component (EIC) in invasive breast cancer.

METHODS

Mammograms with invasive breast cancers from 2010 to 2019 were downloaded for two radiologists performing image segmentation and imaging findings annotation. Images were randomly split into training, validation, and test datasets. A multitask approach was performed on the EfficientNet-B0 neural network mainly to predict EIC and classify imaging findings. Three more models were trained for comparison, including a single-task model (predicting EIC), a two-task model (predicting EIC and cell receptor status), and a three-task model (combining the abovementioned tasks). Additionally, these models were trained in a subgroup of invasive ductal carcinoma. The DeLong test was used to examine the difference in model performance.

RESULTS

This study enrolled 1459 breast cancers on 3076 images. The EIC-positive rate was 29.0%. The three-task model was the best DL model with an area under the curve (AUC) of EIC prediction of 0.758 and 0.775 at the image and breast (patient) levels, respectively. Mass was the most accurately classified imaging finding (AUC = 0.915), followed by calcifications and mass with calcifications (AUC = 0.878 and 0.824, respectively). Cell receptor status prediction was less accurate (AUC = 0.625-0.653). The multitask approach improves the model training compared to the single-task model, but without significant effects.

CONCLUSIONS

A mammography-based multitask DL model can perform simultaneous imaging finding classification and EIC prediction.

CLINICAL RELEVANCE STATEMENT

The study results demonstrated the potential of deep learning to extract more information from mammography for clinical decision-making.

KEY POINTS

• Extensive intraductal component (EIC) is an independent risk factor of local tumor recurrence after breast-conserving surgery. • A mammography-based deep learning model was trained to predict extensive intraductal component close to radiologists' reading. • The developed multitask deep learning model could perform simultaneous imaging finding classification and extensive intraductal component prediction.

Automatic hip osteoarthritis grading with uncertainty estimation from computed tomography using digitally-reconstructed radiographs

PURPOSE

Progression of hip osteoarthritis (hip OA) leads to pain and disability, likely leading to surgical treatment such as hip arthroplasty at the terminal stage. The severity of hip OA is often classified using the Crowe and Kellgren-Lawrence (KL) classifications. However, as the classification is subjective, we aimed to develop an automated approach to classify the disease severity based on the two grades using digitally-reconstructed radiographs from CT images.

METHODS

Automatic grading of the hip OA severity was performed using deep learning-based models. The models were trained to predict the disease grade using two grading schemes, i.e., predicting the Crowe and KL grades separately, and predicting a new ordinal label combining both grades and representing the disease progression of hip OA. The models were trained in classification and regression settings. In addition, the model uncertainty was estimated and validated as a predictor of classification accuracy. The models were trained and validated on a database of 197 hip OA patients, and externally validated on 52 patients. The model accuracy was evaluated using exact class accuracy (ECA), one-neighbor class accuracy (ONCA), and balanced accuracy.

RESULTS

The deep learning models produced a comparable accuracy of approximately 0.65 (ECA) and 0.95 (ONCA) in the classification and regression settings. The model uncertainty was significantly larger in cases with large classification errors ( P < 6 e - 3 ).

CONCLUSIONS

In this study, an automatic approach for grading hip OA severity from CT images was developed. The models have shown comparable performance with high ONCA, which facilitates automated grading in large-scale CT databases and indicates the potential for further disease progression analysis. Classification accuracy was correlated with the model uncertainty, which would allow for the prediction of classification errors. The code will be made publicly available at https://github.com/NAIST-ICB/HipOA-Grading .

Development and Validation of a Deep-Learning Model to Predict Total Hip Replacement on Radiographs: The Total Hip Replacement Prediction (THREP) Model

BACKGROUND

There are few methods for accurately assessing the risk of total hip arthroplasty (THA) in patients with osteoarthritis. A novel and reliable method that could play a substantial role in research and clinical routine should be investigated. The purpose of the present study was to develop a deep-learning model that can reliably predict the risk of THA with use of radiographic images and clinical symptom data.

METHODS

This retrospective, multicenter, case-control study assessed hip joints on weighted-bearing anteroposterior pelvic radiographs obtained from Osteoarthritis Initiative (OAI) participants. Participants who underwent THA were matched to controls according to age, sex, body mass index, and ethnicity. Cases and controls were uniformly split into training, validation, and testing data sets at proportions of 72% (n = 528), 14% (n = 104), and 14% (n = 104), respectively. Images and clinical symptom data were passed through a detection model and a deep convolutional neural network (DCNN) model to predict the probability of THA within 9 years as well as the most likely time period for THA (0 to 2 years, 3 to 5 years, 6 to 9 years). Model performance was assessed with use of the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity in the testing set.

RESULTS

A total of 736 participants were evaluated, including 184 cases and 552 controls. The prediction model achieved an overall accuracy, sensitivity, and specificity of 91.35%, 92.59% and 86.96%, respectively, with an AUC of 0.944, for THA within 9 years. The AUC of the DCNN model for assessing the most likely time period was 0.907 for 0 to 2 years, 0.916 for 3 to 5 years, and 0.841 for 6 to 9 years. Gradient-weighted class activation mapping closely corresponded to regions affecting the prediction of the DCNN model.

CONCLUSIONS

The proposed DCNN model is a reliable and valid method to predict the probability of THA-within limitations. It could assist clinicians in patient counseling and decision-making regarding the timing of the intervention. In the future, by increasing the size of the data set, enhancing the ethnic and socioeconomic diversity of the participants, and improving the follow-up rate, the quality of the conclusions can be further improved.

LEVEL OF EVIDENCE

Prognostic Level III . See Instructions for Authors for a complete description of levels of evidence.

Deep learning in rheumatological image interpretation

Artificial intelligence techniques, specifically deep learning, have already affected daily life in a wide range of areas. Likewise, initial applications have been explored in rheumatology. Deep learning might not easily surpass the accuracy of classic techniques when performing classification or regression on low-dimensional numerical data. With images as input, however, deep learning has become so successful that it has already outperformed the majority of conventional image-processing techniques developed during the past 50 years. As with any new imaging technology, rheumatologists and radiologists need to consider adapting their arsenal of diagnostic, prognostic and monitoring tools, and even their clinical role and collaborations. This adaptation requires a basic understanding of the technical background of deep learning, to efficiently utilize its benefits but also to recognize its drawbacks and pitfalls, as blindly relying on deep learning might be at odds with its capabilities. To facilitate such an understanding, it is necessary to provide an overview of deep-learning techniques for automatic image analysis in detecting, quantifying, predicting and monitoring rheumatic diseases, and of currently published deep-learning applications in radiological imaging for rheumatology, with critical assessment of possible limitations, errors and confounders, and conceivable consequences for rheumatologists and radiologists in clinical practice.

Anterior and posterior hip osteoarthritis: prevalence and potential value of CT compared to radiographs

PURPOSE

To determine the added value of computed tomography (CT) to identify severe hip osteoarthritis (OA).

MATERIALS AND METHODS

A retrospective query of all cases of hip or knee arthroplasty planning CTs between January 2018 and March 2022 was performed. Age, sex, and symptoms were collected from the medical record. CTs were evaluated for the degree of osteoarthritis and classified using an adapted Kellgren-Lawrence (KL) grading system in the anterior, posterior, superior, and superomedial hip. Frontal hip or pelvis radiographs within 1 year of the CT were also graded.

RESULTS

There were 265 eligible hips in 178 subjects, age 66 ± 11 (range 31-93) years, with 85/178 (48%) males and 93/178 (52%) females, and 127/265 (48%) right and 138/265 (52%) left hips. The posterior hip joint was the most common location for grade 2/3 OA (20%), followed by superior hip joint (14%). Anterior or posterior grade 2/3 OA occurred concurrently with superior or superomedial grade 2/3 OA in 32/68 (47%) of hips. Grade 2/3 OA was detected on CT more commonly than on XR both in the superior (14 vs 8.6%, P = 0.0016) and superomedial (8.7 vs 4.8%, P = 0.016) hip joint. Of the 71 symptomatic hips, 22 (31%) hips demonstrated either anterior and/or posterior grade 2/3 OA on CT, and 9 (9/22, 41%) of these hips had superior or superomedial grade 0/1 OA.

CONCLUSION

CT may be warranted when the patient has pain suggestive of osteoarthritis not detected on radiographs.

Immunotherapy and targeted therapy for cholangiocarcinoma: Artificial intelligence research in imaging

Cholangiocarcinoma (CCA) is a highly aggressive hepatobiliary malignancy, second only to hepatocellular carcinoma in prevalence. Despite surgical treatment being the recommended method to achieve a cure, it is not viable for patients with advanced CCA. Gene sequencing and artificial intelligence (AI) have recently opened up new possibilities in CCA diagnosis, treatment, and prognosis assessment. Basic research has furthered our understanding of the tumor-immunity microenvironment and revealed targeted molecular mechanisms, resulting in immunotherapy and targeted therapy being increasingly employed in the clinic. Yet, the application of these remedies in CCA is a challenging endeavor due to the varying pathological mechanisms of different CCA types and the lack of expressed immune proteins and molecular targets in some patients. AI in medical imaging has emerged as a powerful tool in this situation, as machine learning and deep learning are able to extract intricate data from CCA lesion images while assisting clinical decision making, and ultimately improving patient prognosis. This review summarized and discussed the current immunotherapy and targeted therapy related to CCA, and the research progress of AI in this field.

MRI radiomics-based decision support tool for a personalized classification of cervical disc degeneration: a two-center study

Objectives: To develop and validate an MRI radiomics-based decision support tool for the automated grading of cervical disc degeneration. Methods: The retrospective study included 2,610 cervical disc samples of 435 patients from two hospitals. The cervical magnetic resonance imaging (MRI) analysis of patients confirmed cervical disc degeneration grades using the Pfirrmann grading system. A training set (1,830 samples of 305 patients) and an independent test set (780 samples of 130 patients) were divided for the construction and validation of the machine learning model, respectively. We provided a fine-tuned MedSAM model for automated cervical disc segmentation. Then, we extracted 924 radiomic features from each segmented disc in T1 and T2 MRI modalities. All features were processed and selected using minimum redundancy maximum relevance (mRMR) and multiple machine learning algorithms. Meanwhile, the radiomics models of various machine learning algorithms and MRI images were constructed and compared. Finally, the combined radiomics model was constructed in the training set and validated in the test set. Radiomic feature mapping was provided for auxiliary diagnosis. Results: Of the 2,610 cervical disc samples, 794 (30.4%) were classified as low grade and 1,816 (69.6%) were classified as high grade. The fine-tuned MedSAM model achieved good segmentation performance, with the mean Dice coefficient of 0.93. Higher-order texture features contributed to the dominant force in the diagnostic task (80%). Among various machine learning models, random forest performed better than the other algorithms (p < 0.01), and the T2 MRI radiomics model showed better results than T1 MRI in the diagnostic performance (p < 0.05). The final combined radiomics model had an area under the receiver operating characteristic curve (AUC) of 0.95, an accuracy of 89.51%, a precision of 87.07%, a recall of 98.83%, and an F1 score of 0.93 in the test set, which were all better than those of other models (p < 0.05). Conclusion: The radiomics-based decision support tool using T1 and T2 MRI modalities can be used for cervical disc degeneration grading, facilitating individualized management.

Predicting the Risk of Total Hip Replacement by Using A Deep Learning Algorithm on Plain Pelvic Radiographs: Diagnostic Study

BACKGROUND

Total hip replacement (THR) is considered the gold standard of treatment for refractory degenerative hip disorders. Identifying patients who should receive THR in the short term is important. Some conservative treatments, such as intra-articular injection administered a few months before THR, may result in higher odds of arthroplasty infection. Delayed THR after functional deterioration may result in poorer outcomes and longer waiting times for those who have been flagged as needing THR. Deep learning (DL) in medical imaging applications has recently obtained significant breakthroughs. However, the use of DL in practical wayfinding, such as short-term THR prediction, is still lacking.

OBJECTIVE

In this study, we will propose a DL-based assistant system for patients with pelvic radiographs to identify the need for THR within 3 months.

METHODS

We developed a convolutional neural network-based DL algorithm to analyze pelvic radiographs, predict the hip region of interest (ROI), and determine whether or not THR is required. The data set was collected from August 2008 to December 2017. The images included 3013 surgical hip ROIs that had undergone THR and 1630 nonsurgical hip ROIs. The images were split, using split-sample validation, into training (n=3903, 80%), validation (n=476, 10%), and testing (n=475, 10%) sets to evaluate the algorithm performance.

RESULTS

The algorithm, called SurgHipNet, yielded an area under the receiver operating characteristic curve of 0.994 (95% CI 0.990-0.998). The accuracy, sensitivity, specificity, and F1-score of the model were 0.977, 0.920, 0932, and 0.944, respectively.

CONCLUSIONS

The proposed approach has demonstrated that SurgHipNet shows the ability and potential to provide efficient support in clinical decision-making; it can assist physicians in promptly determining the optimal timing for THR.

Deep Learning Approach for Diagnosing Early Osteonecrosis of the Femoral Head Based on Magnetic Resonance Imaging

BACKGROUND

The diagnosis of early osteonecrosis of the femoral head (ONFH) based on magnetic resonance imaging (MRI) is challenging due to variability in the surgeon's experience level. This study developed an MRI-based deep learning system to detect early ONFH and evaluated its feasibility in the clinic.

METHODS

We retrospectively evaluated clinical MRIs of the hips that were performed in our institution from January 2019 to June 2022 and collected all MRIs diagnosed with early ONFH. An advanced convolutional neural network (CNN) was trained and optimized; then, the diagnostic performance of the CNN was evaluated according to its accuracy, sensitivity, and specificity. We also further compared the CNN's performance with that of orthopaedic surgeons.

RESULTS

Overall, 11,061 images were retrospectively included in the present study and were divided into three datasets with ratio 7:2:1. The area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity of the CNN model for identifying early ONFH were 0.98, 98.4, 97.6, and 98.6%, respectively. In our review panel, the averaged accuracy, sensitivity, and specificity for identifying ONFH were 91.7, 87.0, and 94.1% for attending orthopaedic surgeons; 87.1, 84.0, and 89.3% for resident orthopaedic surgeons; and 97.1, 96.0, and 97.9% for deputy chief orthopaedic surgeons, respectively.

CONCLUSION

The deep learning system showed a comparable performance to that of deputy chief orthopaedic surgeons in identifying early ONFH. The success of deep learning diagnosis of ONFH might be conducive to assisting less-experienced surgeons, especially in large-scale medical imaging screening and community scenarios lacking consulting experts.

Risk stratification by nomogram of deep learning radiomics based on multiparametric magnetic resonance imaging in knee meniscus injury

PURPOSE

To construct and validate a nomogram model that integrated deep learning radiomic features based on multiparametric MRI and clinical features for risk stratification of meniscus injury.

METHODS

A total of 167 knee MR images were collected from two institutions. All patients were classified into two groups based on the MR diagnostic criteria proposed by Stoller et al. The automatic meniscus segmentation model was constructed through V-net. LASSO regression was performed to extract the optimal features correlated to risk stratification. A nomogram model was constructed by combining the Radscore and clinical features. The performance of the models was evaluated by ROC analysis and calibration curve. Subsequently, the model was simulated by junior doctors in order to test its practical application effect.

RESULTS

The Dice similarity coefficients of automatic meniscus segmentation models were all over 0.8. Eight optimal features, identified by LASSO regression, were employed to calculate the Radscore. The combined model showed a better performance in both the training cohort (AUC = 0.90, 95%CI: 0.84-0.95) and the validation cohort (AUC = 0.84, 95%CI: 0.72-0.93). The calibration curve indicated a better accuracy of the combined model than either the Radscore or clinical model alone. The simulation results showed that the diagnostic accuracy of junior doctors increased from 74.9 to 86.2% after using the model.

CONCLUSION

Deep learning V-net demonstrated great performance in automatic meniscus segmentation of the knee joint. It was reliable for stratifying the risk of meniscus injury of the knee by nomogram which integrated the Radscores and clinical features.

Convolutional neural networks on risk stratification of patients with suspected coronary artery disease undergoing coronary computed tomography angiography

To assess the prognostic value of convolutional neural networks (CNN) on coronary computed tomography angiography (CCTA) in comparison to conventional computed tomography (CT) reporting and clinical risk scores. 5468 patients who underwent CCTA with suspected coronary artery disease (CAD) were included. Primary endpoint was defined as a composite of all-cause death, myocardial infarction, unstable angina or late revascularization (> 90 days after CCTA). Early revascularization was additionally included as a training endpoint for the CNN algorithm. Cardiovascular risk stratification was based on Morise score and the extent of CAD (eoCAD) as assessed on CCTA. Semiautomatic post-processing was performed for vessel delineation and annotation of calcified and non-calcified plaque areas. Using a two-step training of a DenseNet-121 CNN the entire network was trained with the training endpoint, followed by training the feature layer with the primary endpoint. During a median follow-up of 7.2 years, the primary endpoint occurred in 334 patients. CNN showed an AUC of 0.631 ± 0.015 for prediction of the combined primary endpoint, while combining it with conventional CT and clinical risk scores showed an improvement of AUC from 0.646 ± 0.014 (based on eoCAD only) to 0.680 ± 0.015 (p < 0.0001) and from 0.619 ± 0.0149 (based on Morise Score only) to 0.6812 ± 0.0145 (p < 0.0001), respectively. In a stepwise model including all prediction methods, it was found an AUC of 0.680 ± 0.0148. CNN analysis showed to improve conventional CCTA-derived and clinical risk stratification when evaluating CCTA of patients with suspected CAD.

The Feasibility and Performance of Total Hip Replacement Prediction Deep Learning Algorithm with Real World Data

(1) Background: Hip degenerative disorder is a common geriatric disease is the main causes to lead to total hip replacement (THR). The surgical timing of THR is crucial for post-operative recovery. Deep learning (DL) algorithms can be used to detect anomalies in medical images and predict the need for THR. The real world data (RWD) were used to validate the artificial intelligence and DL algorithm in medicine but there was no previous study to prove its function in THR prediction. (2) Methods: We designed a sequential two-stage hip replacement prediction deep learning algorithm to identify the possibility of THR in three months of hip joints by plain pelvic radiography (PXR). We also collected RWD to validate the performance of this algorithm. (3) Results: The RWD totally included 3766 PXRs from 2018 to 2019. The overall accuracy of the algorithm was 0.9633; sensitivity was 0.9450; specificity was 1.000 and the precision was 1.000. The negative predictive value was 0.9009, the false negative rate was 0.0550, and the F1 score was 0.9717. The area under curve was 0.972 with 95% confidence interval from 0.953 to 0.987. (4) Conclusions: In summary, this DL algorithm can provide an accurate and reliable method for detecting hip degeneration and predicting the need for further THR. RWD offered an alternative support of the algorithm and validated its function to save time and cost.

Standardized Fixation Zones and Cone Assessments for Revision Total Knee Arthroplasty Using Deep Learning

BACKGROUND

Achieving adequate implant fixation is critical to optimize survivorship and postoperative outcomes after revision total knee arthroplasty (rTKA). Three anatomical zones (ie, epiphysis, metaphysis, and diaphysis) have been proposed to assess fixation, but are not well-defined. The purpose of the study was to develop a deep learning workflow capable of automatically delineating rTKA zones and cone placements in a standardized way on postoperative radiographs.

METHODS

A total of 235 patients who underwent rTKA were randomly partitioned (6:2:2 training, validation, and testing split), and a U-Net segmentation workflow was developed to delineate rTKA fixation zones and assess revision cone placement on anteroposterior radiographs. Algorithm performance for zone delineation and cone placement were compared against ground truths from a fellowship-trained arthroplasty surgeon using the dice segmentation coefficient and accuracy metrics.

RESULTS

On the testing cohort, the algorithm defined zones in 98% of images (8 seconds/image) using anatomical landmarks. The dice segmentation coefficient between the model and surgeon was 0.89 ± 0.08 (interquartile range [IQR]:0.88-0.94) for femoral zones, 0.91 ± 0.08 (IQR: 0.91-0.95) for tibial zones, and 0.90 ± 0.05 (IQR:0.88-0.94) for all zones. Cone identification and zonal cone placement accuracy were 98% and 96%, respectively, for the femur and 96% and 89%, respectively, for the tibia.

CONCLUSION

A deep learning algorithm was developed to automatically delineate revision zones and cone placements on postoperative rTKA radiographs in an objective, standardized manner. The performance of the algorithm was validated against a trained surgeon, suggesting that the algorithm demonstrated excellent predictive capabilities in accordance with relevant anatomical landmarks used by arthroplasty surgeons in practice.

Multi-task deep learning for medical image computing and analysis: A review

The renaissance of deep learning has provided promising solutions to various tasks. While conventional deep learning models are constructed for a single specific task, multi-task deep learning (MTDL) that is capable to simultaneously accomplish at least two tasks has attracted research attention. MTDL is a joint learning paradigm that harnesses the inherent correlation of multiple related tasks to achieve reciprocal benefits in improving performance, enhancing generalizability, and reducing the overall computational cost. This review focuses on the advanced applications of MTDL for medical image computing and analysis. We first summarize four popular MTDL network architectures (i.e., cascaded, parallel, interacted, and hybrid). Then, we review the representative MTDL-based networks for eight application areas, including the brain, eye, chest, cardiac, abdomen, musculoskeletal, pathology, and other human body regions. While MTDL-based medical image processing has been flourishing and demonstrating outstanding performance in many tasks, in the meanwhile, there are performance gaps in some tasks, and accordingly we perceive the open challenges and the perspective trends. For instance, in the 2018 Ischemic Stroke Lesion Segmentation challenge, the reported top dice score of 0.51 and top recall of 0.55 achieved by the cascaded MTDL model indicate further research efforts in high demand to escalate the performance of current models.

Systematic review of artificial intelligence tack in preventive orthopaedics: is the land coming soon?

PURPOSE

This study aims to describe and assess the current stage of the artificial intelligence (AI) technology integration in preventive orthopaedics of the knee and hip joints.

MATERIALS AND METHODS

The study was conducted in strict compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement. Literature databases were searched for articles describing the development and validation of AI models aimed at diagnosing knee or hip joint pathologies or predicting their development or course in patients. The quality of the included articles was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) and QUADAS-AI tools.

RESULTS

56 articles were found that meet all the inclusion criteria. We identified two problems that block the full integration of AI into the routine of an orthopaedic physician. The first of them is related to the insufficient amount, variety and quality of data for training, and validation and testing of AI models. The second problem is the rarity of rational evaluation of models, which is why their real quality cannot always be evaluated.

CONCLUSION

The vastness and relevance of the studied topic are beyond doubt. Qualitative and optimally validated models exist in all four scopes considered. Additional optimization and confirmation of the models' quality on various datasets are the last technical stumbling blocks for creating usable software and integrating them into the routine of an orthopaedic physician.

Automatic diagnosis and grading of patellofemoral osteoarthritis from the axial radiographic view: a deep learning-based approach

BACKGROUND

Patellofemoral osteoarthritis (PFOA) has a high prevalence and is assessed on axial radiography of the patellofemoral joint (PFJ). A deep learning (DL)-based approach could help radiologists automatically diagnose and grade PFOA via interpreting axial radiographs.

PURPOSE

To develop and assess the performance of a DL-based approach for diagnosing and grading PFOA on axial radiographs.

MATERIAL AND METHODS

A total of 1280 (dataset 1) axial radiographs were retrospectively collected and utilized to develop the high-resolution network (HRNet)-based classification models. The ground truth was the interpretation from two experienced radiologists in consensus according to the K-L grading system. A binary-class model was trained to diagnose the presence (K-L 2∼4) or absence (K-L 0∼1) of PFOA. A multi-class model was used to grade the stage of PFOA, i.e. from K-L 0 to K-L 4. Model performances were evaluated using the receiver operating characteristics (ROC), confusion matrix, and the corresponding evaluation metrics (positive predictive value [PPV], negative predictive value [NPV], F1 score, sensitivity, specificity, accuracy) of the internal test set (n = 129) from dataset 1 and an external validation set (dataset 2, n = 187).

RESULTS

For the binary-class model, the area under the curve (AUC) was 0.91 in the internal test set and 0.90 in the external validation set. For grading PFOA, moderate to severe stage of PFOA exhibited a good performance in these two datasets (AUC = 0.91-0.98, PPV = 0.69-0.90, NPV = 0.92-0.99, F1 score = 0.72-0.87, sensitivity = 0.75-0.87, specificity = 0.90-0.99, accuracy = 0.87-0.98).

CONCLUSION

The HRNet-based approach performed well in diagnosing and grading radiographic PFOA, especially for the moderate to severe cases.

Development and validation of a deep learning-based model to distinguish acetabular fractures on pelvic anteroposterior radiographs

Objective: To develop and test a deep learning (DL) model to distinguish acetabular fractures (AFs) on pelvic anteroposterior radiographs (PARs) and compare its performance to that of clinicians. Materials and methods: A total of 1,120 patients from a big level-I trauma center were enrolled and allocated at a 3:1 ratio for the DL model's development and internal test. Another 86 patients from two independent hospitals were collected for external validation. A DL model for identifying AFs was constructed based on DenseNet. AFs were classified into types A, B, and C according to the three-column classification theory. Ten clinicians were recruited for AF detection. A potential misdiagnosed case (PMC) was defined based on clinicians' detection results. The detection performance of the clinicians and DL model were evaluated and compared. The detection performance of different subtypes using DL was assessed using the area under the receiver operating characteristic curve (AUC). Results: The means of 10 clinicians' sensitivity, specificity, and accuracy to identify AFs were 0.750/0.735, 0.909/0.909, and 0.829/0.822, in the internal test/external validation set, respectively. The sensitivity, specificity, and accuracy of the DL detection model were 0.926/0.872, 0.978/0.988, and 0.952/0.930, respectively. The DL model identified type A fractures with an AUC of 0.963 [95% confidence interval (CI): 0.927-0.985]/0.950 (95% CI: 0.867-0.989); type B fractures with an AUC of 0.991 (95% CI: 0.967-0.999)/0.989 (95% CI: 0.930-1.000); and type C fractures with an AUC of 1.000 (95% CI: 0.975-1.000)/1.000 (95% CI: 0.897-1.000) in the test/validation set. The DL model correctly recognized 56.5% (26/46) of PMCs. Conclusion: A DL model for distinguishing AFs on PARs is feasible. In this study, the DL model achieved a diagnostic performance comparable to or even superior to that of clinicians.

Deep-learning based quantification model for hip bone marrow edema and synovitis in patients with spondyloarthritis based on magnetic resonance images

Objectives: Hip inflammation is one of the most common complications in patients with spondyloarthritis (SpA). Herein, we employed use of a deep learning-based magnetic resonance imaging (MRI) evaluation model to identify irregular and multiple inflammatory lesions of the hip. Methods: All of the SpA patients were enrolled at the Xijing Hospital. The erythrocyte sediment rate (ESR), C-reactive protein (CRP), hip function Harris score, and disease activity were evaluated by clinicians. Manual MRI annotations including bone marrow edema (BME) and effusion/synovitis, and a hip MRI scoring system (HIMRISS) assessment was performed by experienced musculoskeletal radiologists. The segmentation accuracies of four deep learning models, including U-Net, UNet++, Attention-Unet, and HRNet, were compared using five-fold cross-validation. The clinical agreement of U-Net was evaluated with clinical symptoms and HIMRISS results. Results: A total of 1945 MRI slices of STIR/T2WI sequences were obtained from 195 SpA patients with hip involvement. After the five-fold cross-validation, U-Net achieved an average segmentation accuracy of 88.48% for the femoral head and 69.36% for inflammatory lesions, which are higher than those obtained by the other three models. The UNet-score, which was calculated based on the same MRI slices as HIMRISS, was significantly correlated with the HIMRISS scores and disease activity indexes (p values <0.05). Conclusion: This deep-learning based automatic MRI evaluation model could achieve similar quantification performance as an expert radiologist, and it has the potential to improve the accuracy and efficiency of clinical diagnosis for SpA patients with hip involvement.

Associations Between Baseline and Longitudinal Semiautomated Quantitative Joint Space Width at the Hip and Incident Hip Osteoarthritis: Data From a Community-Based Cohort

OBJECTIVE

To evaluate quantitative joint space width (JSW) at 10-, 30-, and 50-degree locations in relation to incident radiographic and symptomatic hip osteoarthritis (HOA) in a community-based cohort.

METHODS

Data were from Johnston County OA Project participants with supine hip radiographs at each of 4 time points; all had Kellgren/Lawrence (K/L) grades and quantitative JSW. We assessed covariates (age, race, height, weight, body mass index [BMI]) associated with quantitative JSW and hip-level associations between quantitative JSW and HOA over time using sex-stratified and multivariable-adjusted linear mixed models. A cluster analysis with logistic regression estimated associations between quantitative JSW trajectory groups and incident radiographic HOA and symptomatic HOA.

RESULTS

At baseline, 397 participants (784 hips, 41% men, 24% Black, mean age 57 years) had a mean BMI of 29 kg/m2 . Over a mean of 18 years, 20% and 12% developed incident K/L grade-defined radiographic HOA or symptomatic HOA, respectively. Quantitative JSW was more sensitive to changes over time at 50 degrees. Values were stable among men but declined over time in women. Heavier women lost more quantitative JSW; changes in quantitative JSW were not significantly associated with race, education, or injury in women or men. In women only, loss of quantitative JSW over time was associated with 2-3 times higher odds of radiographic HOA and symptomatic HOA; among women and men, narrower baseline quantitative JSW was associated with these outcomes.

CONCLUSION

Hip quantitative JSW demonstrates marked differences in respect to sex, with significant loss over time only in women. Loss of quantitative JSW over time in women and narrower baseline quantitative JSW in men and women were associated with incident radiographic HOA and symptomatic HOA.

Hip osteoarthritis: A novel network analysis of subchondral trabecular bone structures

Hip osteoarthritis (HOA) is a degenerative joint disease that leads to the progressive destruction of subchondral bone and cartilage at the hip joint. Development of effective treatments for HOA remains an open problem, primarily due to the lack of knowledge of its pathogenesis and a typically late-stage diagnosis. We describe a novel network analysis methodology for microcomputed tomography (micro-CT) images of human trabecular bone. We explored differences between the trabecular bone microstructure of femoral heads with and without HOA. Large-scale automated extraction of the network formed by trabecular bone revealed significant network properties not previously reported for bone. Profound differences were discovered, particularly in the proximal third of the femoral head, where HOA networks demonstrated elevated numbers of edges, vertices, and graph components. When further differentiating healthy joint and HOA networks, the latter showed fewer small-world network properties, due to decreased clustering coefficient and increased characteristic path length. Furthermore, we found that HOA networks had reduced length of edges, indicating the formation of compressed trabecular structures. In order to assess our network approach, we developed a deep learning model for classifying HOA and control cases, and we fed it with two separate inputs: (i) micro-CT images of the trabecular bone, and (ii) the network extracted from them. The model with plain micro-CT images achieves 74.6% overall accuracy while the trained model with extracted networks attains 96.5% accuracy. We anticipate our findings to be a starting point for a novel description of bone microstructure in HOA, by considering the phenomenon from a graph theory viewpoint.

John Charnley Award: Deep Learning Prediction of Hip Joint Center on Standard Pelvis Radiographs

BACKGROUND

Accurate hip joint center (HJC) determination is critical for preoperative planning, intraoperative execution, clinical outcomes after total hip arthroplasty, and commonly used classification systems in primary and revision hip replacement. However, current methods of preoperative HJC estimation are prone to subjectivity and human error. The purpose of the study was to leverage deep learning (DL) to develop a rapid and objective HJC estimation tool on anteroposterior (AP) pelvis radiographs.

METHODS

Radiographs from 3,965 patients (7,930 hips) were included. A DL model workflow was created to detect bony landmarks and estimate HJC based on a pelvic height ratio method. The workflow was utilized to conduct a grid-search for optimal nonspecific, sex-specific, and patient-specific (using contralateral hip) pelvic height ratios on the training/validation cohort (6,344 hips). Algorithm performance was assessed on an independent testing cohort for HJC estimation comparison.

RESULTS

The algorithm estimated HJC for the testing cohort at a rate of 0.65 seconds/hip based on features in AP radiographs alone. The model predicted HJC within 5 mm of error for 80% of hips using nonspecific ratios, which increased to 83% with sex-specific and 91% with patient-specific pelvic height ratio models. Mean error decreased utilizing the patient-specific model (3.09 ± 1.69 mm, P < .001).

CONCLUSION

Using DL, we developed nonspecific, sex-specific, and patient-specific models capable of estimating native HJC on AP pelvis radiographs. This tool may provide clinical value when considering preoperative component position in patients planned to undergo THA and in reducing the subjective variability in HJC estimation.

LEVEL OF EVIDENCE

Diagnostic, level IV.

External Validation of Deep Learning Algorithms for Radiologic Diagnosis: A Systematic Review

Purpose

To assess generalizability of published deep learning (DL) algorithms for radiologic diagnosis.

Materials and Methods

In this systematic review, the PubMed database was searched for peer-reviewed studies of DL algorithms for image-based radiologic diagnosis that included external validation, published from January 1, 2015, through April 1, 2021. Studies using nonimaging features or incorporating non-DL methods for feature extraction or classification were excluded. Two reviewers independently evaluated studies for inclusion, and any discrepancies were resolved by consensus. Internal and external performance measures and pertinent study characteristics were extracted, and relationships among these data were examined using nonparametric statistics.

Results

Eighty-three studies reporting 86 algorithms were included. The vast majority (70 of 86, 81%) reported at least some decrease in external performance compared with internal performance, with nearly half (42 of 86, 49%) reporting at least a modest decrease (≥0.05 on the unit scale) and nearly a quarter (21 of 86, 24%) reporting a substantial decrease (≥0.10 on the unit scale). No study characteristics were found to be associated with the difference between internal and external performance.

Conclusion

Among published external validation studies of DL algorithms for image-based radiologic diagnosis, the vast majority demonstrated diminished algorithm performance on the external dataset, with some reporting a substantial performance decrease.Keywords: Meta-Analysis, Computer Applications-Detection/Diagnosis, Neural Networks, Computer Applications-General (Informatics), Epidemiology, Technology Assessment, Diagnosis, Informatics Supplemental material is available for this article. © RSNA, 2022.

Automatic Detection of Medial and Lateral Compartments from Histological Sections of Mouse Knee Joints Using the Single-Shot Multibox Detector Algorithm

OBJECTIVE

Although mouse osteoarthritis (OA) models are widely used, their histological analysis may be susceptible to arbitrariness and inter-examiner variability in conventional methods. Therefore, a method for the unbiased scoring of OA histology is needed. In this study, as the first step for establishing this system, we developed a computer-vision algorithm that automatically detects the medial and lateral compartments of mouse knee sections in a rigorous and unbiased manner.

DESIGN

A total of 706 images of coronal sections of mouse knee joints stained by hematoxylin and eosin, safranin O, or toluidine blue were randomly divided into training and validation images at a ratio of 80:20. A model to detect both compartments automatically was built by machine learning using a single-shot multibox detector (SSD) algorithm with training images. The model was tested to determine whether it could accurately detect both compartments by analyzing the validation images and 52 images of sections stained with Picrosirius red, a method not used for the training images.

RESULTS

The trained model accurately detected both medial and lateral compartments of all 140 validation images regardless of the staining method employed, severity of articular cartilage defects, and the anatomical positions and conditions of the sections. Our model also correctly detected both compartments of 50 of 52 Picrosirius red-stained images.

CONCLUSIONS

By applying deep learning based on the SSD algorithm, we successfully developed a model that detects the locations of the medial and lateral compartments of tissue sections of mouse knee joints with high accuracy.

Handling Imbalanced Data: Uncertainty-guided Virtual Adversarial Training with Batch Nuclear-norm Optimization for Semi-supervised Medical Image Classification

In many clinical settings, a lot of medical image datasets suffer from imbalance problems, which makes predictions of trained models to be biased toward majority classes. Semi-supervised Learning (SSL) algorithms trained with such imbalanced datasets become more problematic since pseudo-supervision of unlabeled data are generated from the model's biased predictions. To address these issues, in this work, we propose a novel semi-supervised deep learning method, i.e., uncertainty-guided virtual adversarial training (VAT) with batch nuclear-norm (BNN) optimization, for large-scale medical image classification. To effectively exploit useful information from both labeled and unlabeled data, we leverage VAT and BNN optimization to harness the underlying knowledge, which helps to improve discriminability, diversity and generalization of the trained models. More concretely, our network is trained by minimizing a combination of four types of losses, including a supervised cross-entropy loss, a BNN loss defined on the output matrix of labeled data batch (lBNN loss), a negative BNN loss defined on the output matrix of unlabeled data batch (uBNN loss), and a VAT loss on both labeled and unlabeled data. We additionally propose to use uncertainty estimation to filter out unlabeled samples near the decision boundary when computing the VAT loss. We conduct comprehensive experiments to evaluate the performance of our method on two publicly available datasets and one in-house collected dataset. The experimental results demonstrated that our method achieved better results than state-of-the-art SSL methods.

Multi-landmark environment analysis with reinforcement learning for pelvic abnormality detection and quantification

Morphological abnormalities of the femoroacetabular (hip) joint are among the most common human musculoskeletal disorders and often develop asymptomatically at early easily treatable stages. In this paper, we propose an automated framework for landmark-based detection and quantification of hip abnormalities from magnetic resonance (MR) images. The framework relies on a novel idea of multi-landmark environment analysis with reinforcement learning. In particular, we merge the concepts of the graphical lasso and Morris sensitivity analysis with deep neural networks to quantitatively estimate the contribution of individual landmark and landmark subgroup locations to the other landmark locations. Convolutional neural networks for image segmentation are utilized to propose the initial landmark locations, and landmark detection is then formulated as a reinforcement learning (RL) problem, where each landmark-agent can adjust its position by observing the local MR image neighborhood and the locations of the most-contributive landmarks. The framework was validated on T1-, T2- and proton density-weighted MR images of 260 patients with the aim to measure the lateral center-edge angle (LCEA), femoral neck-shaft angle (NSA), and the anterior and posterior acetabular sector angles (AASA and PASA) of the hip, and derive the quantitative abnormality metrics from these angles. The framework was successfully tested using the UNet and feature pyramid network (FPN) segmentation architectures for landmark proposal generation, and the deep Q-network (DeepQN), deep deterministic policy gradient (DDPG), twin delayed deep deterministic policy gradient (TD3), and actor-critic policy gradient (A2C) RL networks for landmark position optimization. The resulting overall landmark detection error of 1.5 mm and angle measurement error of 1.4° indicates a superior performance in comparison to existing methods. Moreover, the automatically estimated abnormality labels were in 95% agreement with those generated by an expert radiologist.

Use of machine learning in osteoarthritis research: a systematic literature review

OBJECTIVE

The aim of this systematic literature review was to provide a comprehensive and exhaustive overview of the use of machine learning (ML) in the clinical care of osteoarthritis (OA).

METHODS

A systematic literature review was performed in July 2021 using MEDLINE PubMed with key words and MeSH terms. For each selected article, the number of patients, ML algorithms used, type of data analysed, validation methods and data availability were collected.

RESULTS

From 1148 screened articles, 46 were selected and analysed; most were published after 2017. Twelve articles were related to diagnosis, 7 to prediction, 4 to phenotyping, 12 to severity and 11 to progression. The number of patients included ranged from 18 to 5749. Overall, 35% of the articles described the use of deep learning And 74% imaging analyses. A total of 85% of the articles involved knee OA and 15% hip OA. No study investigated hand OA. Most of the studies involved the same cohort, with data from the OA initiative described in 46% of the articles and the MOST and Cohort Hip and Cohort Knee cohorts in 11% and 7%. Data and source codes were described as publicly available respectively in 54% and 22% of the articles. External validation was provided in only 7% of the articles.

CONCLUSION

This review proposes an up-to-date overview of ML approaches used in clinical OA research and will help to enhance its application in this field.

Studying osteoarthritis with artificial intelligence applied to magnetic resonance imaging

The 3D nature and soft-tissue contrast of MRI makes it an invaluable tool for osteoarthritis research, by facilitating the elucidation of disease pathogenesis and progression. The recent increasing employment of MRI has certainly been stimulated by major advances that are due to considerable investment in research, particularly related to artificial intelligence (AI). These AI-related advances are revolutionizing the use of MRI in clinical research by augmenting activities ranging from image acquisition to post-processing. Automation is key to reducing the long acquisition times of MRI, conducting large-scale longitudinal studies and quantitatively defining morphometric and other important clinical features of both soft and hard tissues in various anatomical joints. Deep learning methods have been used recently for multiple applications in the musculoskeletal field to improve understanding of osteoarthritis. Compared with labour-intensive human efforts, AI-based methods have advantages and potential in all stages of imaging, as well as post-processing steps, including aiding diagnosis and prognosis. However, AI-based methods also have limitations, including the arguably limited interpretability of AI models. Given that the AI community is highly invested in uncovering uncertainties associated with model predictions and improving their interpretability, we envision future clinical translation and progressive increase in the use of AI algorithms to support clinicians in optimizing patient care.

Detecting hip osteoarthritis on clinical CT: a deep learning application based on 2-D summation images derived from CT

We developed and compared deep learning models to detect hip osteoarthritis on clinical CT. The CT-based summation images, CT-AP, that resemble X-ray radiographs can detect radiographic hip osteoarthritis and in the absence of large training data, a reliable deep learning model can be optimized by combining CT-AP and X-ray images.

INTRODUCTION

In this study, we aimed to investigate the applicability of deep learning (DL) to assess radiographic hip osteoarthritis (rHOA) on computed tomography (CT).

METHODS

The study data consisted of 94 abdominopelvic clinical CTs and 5659 hip X-ray images collected from Cohort Hip and Cohort Knee (CHECK). The CT slices were sequentially summed to create radiograph-like 2-D images named CT-AP. X-ray and CT-AP images were classified as rHOA if they had osteoarthritic changes corresponding to Kellgren-Lawrence grade 2 or higher. The study data was split into 55% training, 30% validation, and 15% test sets. A pretrained ResNet18 was optimized for a classification task of rHOA vs. no-rHOA. Five models were trained using (1) X-rays, (2) downsampled X-rays, (3) combination of CT-AP and X-ray images, (4) combination of CT-AP and downsampled X-ray images, and (5) CT-AP images.

RESULTS

Amongst the five models, Model-3 and Model-5 performed best in detecting rHOA from the CT-AP images. Model-3 detected rHOA on the test set of CT-AP images with a balanced accuracy of 82.2% and was able to discriminate rHOA from no-rHOA with an area under the receiver operating characteristic curve (ROC AUC) of 0.93 [0.75-0.99]. Model-5 detected rHOA on the test set at a balanced accuracy of 82.2% and classified rHOA from no-rHOA with an ROC AUC of 0.89 [0.67-0.97].

CONCLUSION

CT-based summation images that resemble radiographs can be used to detect rHOA. In addition, in the absence of large training data, a reliable DL model can be optimized by combining CT-AP and X-ray images.

Osteoarthritis year in review 2021: imaging

PURPOSE

To provide a narrative review of original articles on imaging of osteoarthritis (OA) published between January 1, 2020 and March 31, 2021, with a special focus on imaging of inflammation, imaging of bone, cartilage and bone-cartilage interactions, imaging of peri-articular tissues, imaging scoring methods for OA, and artificial intelligence (AI) applied to OA imaging.

METHODS

The Embase, Pubmed, Medline, Cochrane databases were searched for original research articles in the English language on human, in vivo, imaging of OA published between January 1, 2020 and March 31, 2021. Search terms related to osteoarthritis combined with all imaging modalities and artificial intelligence were applied. A selection of articles reporting on one of the focus topics was discussed further.

RESULTS

The search resulted in 651 articles, of which 214 were deemed relevant to human OA imaging. Among the articles included, the knee joint (69%) and magnetic resonance imaging (MRI) (52%) were the predominant anatomical area and imaging modality studied. There were also a substantial number of papers (n = 46) reporting on AI applications in the field of OA imaging.

CONCLUSION

Imaging continues to play an important role in the assessment of OA. Recent advances in OA imaging include quantitative, non-contrast, and hybrid imaging techniques for improved characterization of multiple tissue processes in OA. In addition, an increasing effort in AI techniques is undertaken to enhance OA imaging acquisition and analysis.

Deep learning for accurately recognizing common causes of shoulder pain on radiographs

OBJECTIVE

Training a convolutional neural network (CNN) to detect the most common causes of shoulder pain on plain radiographs and to assess its potential value in serving as an assistive device to physicians.

MATERIALS AND METHODS

We used a CNN of the ResNet-50 architecture which was trained on 2700 shoulder radiographs from clinical practice of multiple institutions. All radiographs were reviewed and labeled for six findings: proximal humeral fractures, joint dislocation, periarticular calcification, osteoarthritis, osteosynthesis, and joint endoprosthesis. The trained model was then evaluated on a separate test dataset, which was previously annotated by three independent expert radiologists. Both the training and the test datasets included radiographs of highly variable image quality to reflect the clinical situation and to foster robustness of the CNN. Performance of the model was evaluated using receiver operating characteristic (ROC) curves, the thereof derived AUC as well as sensitivity and specificity.

RESULTS

The developed CNN demonstrated a high accuracy with an area under the curve (AUC) of 0.871 for detecting fractures, 0.896 for joint dislocation, 0.945 for osteoarthritis, and 0.800 for periarticular calcifications. It also detected osteosynthesis and endoprosthesis with near perfect accuracy (AUC 0.998 and 1.0, respectively). Sensitivity and specificity were 0.75 and 0.86 for fractures, 0.95 and 0.65 for joint dislocation, 0.90 and 0.86 for osteoarthrosis, and 0.60 and 0.89 for calcification.

CONCLUSION

CNNs have the potential to serve as an assistive device by providing clinicians a means to prioritize worklists or providing additional safety in situations of increased workload.

Imaging Manifestations and Evaluation of Postoperative Complications of Bone and Joint Infections under Deep Learning

To explore and evaluate the imaging manifestations of postoperative complications of bone and joint infections based on deep learning, a retrospective study was performed on 40 patients with bone and joint infections in the Department of Orthopedics of Orthopedics Hospital of Henan Province of Luoyang City. Sensitivity and Dice similarity coefficient (DSC) were used to evaluate the image results by convolutional neural network (CNN) algorithm. Imaging features of postoperative complications in 40 patients were analyzed. Then, three imaging methods were used to diagnose the features. Sensitivity and specificity were used to evaluate the diagnostic performance of three imaging methods for imaging features. Compared with professional doctors and biomarker algorithms, the sensitivity of CNN algorithm proposed was 90.6%, and DSC was 84.1%. Compared with traditional methods, the CNN algorithm has higher image resolution and wider and more accurate lesion area recognition and division. The three manifestations of soft tissue abscess, periosteum swelling, and bone damage were postoperative imaging features of bone and joint infections. In addition, compared with X-ray, CT examination and MRI examination were better for the examination of imaging characteristics. CT and MRI had higher sensitivity and specificity than X-ray. The experimental results show that CNN algorithm can effectively identify and divide pathological images and assist doctors to diagnose the images more efficiently in clinic.

Toward Generalizability in the Deployment of Artificial Intelligence in Radiology: Role of Computation Stress Testing to Overcome Underspecification

The clinical deployment of artificial intelligence (AI) applications in medical imaging is perhaps the greatest challenge facing radiology in the next decade. One of the main obstacles to the incorporation of automated AI-based decision-making tools in medicine is the failure of models to generalize when deployed across institutions with heterogeneous populations and imaging protocols. The most well-understood pitfall in developing these AI models is overfitting, which has, in part, been overcome by optimizing training protocols. However, overfitting is not the only obstacle to the success and generalizability of AI. Underspecification is also a serious impediment that requires conceptual understanding and correction. It is well known that a single AI pipeline, with prescribed training and testing sets, can produce several models with various levels of generalizability. Underspecification defines the inability of the pipeline to identify whether these models have embedded the structure of the underlying system by using a test set independent of, but distributed identically, to the training set. An underspecified pipeline is unable to assess the degree to which the models will be generalizable. Stress testing is a known tool in AI that can limit underspecification and, importantly, assure broad generalizability of AI models. However, the application of stress tests is new in radiologic applications. This report describes the concept of underspecification from a radiologist perspective, discusses stress testing as a specific strategy to overcome underspecification, and explains how stress tests could be designed in radiology-by modifying medical images or stratifying testing datasets. In the upcoming years, stress tests should become in radiology the standard that crash tests have become in the automotive industry. Keywords: Computer Applications-General, Informatics, Computer-aided Diagnosis © RSNA, 2021.

Clinical Artificial Intelligence Applications: Musculoskeletal

We present an overview of current clinical musculoskeletal imaging applications for artificial intelligence, as well as potential future applications and techniques.

Multitask Deep Learning for Segmentation and Classification of Primary Bone Tumors on Radiographs

Background An artificial intelligence model that assesses primary bone tumors on radiographs may assist in the diagnostic workflow. Purpose To develop a multitask deep learning (DL) model for simultaneous bounding box placement, segmentation, and classification of primary bone tumors on radiographs. Materials and Methods This retrospective study analyzed bone tumors on radiographs acquired prior to treatment and obtained from patient data from January 2000 to June 2020. Benign or malignant bone tumors were diagnosed in all patients by using the histopathologic findings as the reference standard. By using split-sample validation, 70% of the patients were assigned to the training set, 15% were assigned to the validation set, and 15% were assigned to the test set. The final performance was evaluated on an external test set by using geographic validation, with accuracy, sensitivity, specificity, and 95% CIs being used for classification, the intersection over union (IoU) being used for bounding box placements, and the Dice score being used for segmentations. Results Radiographs from 934 patients (mean age, 33 years ± 19 [standard deviation]; 419 women) were evaluated in the internal data set, which included 667 benign bone tumors and 267 malignant bone tumors. Six hundred fifty-four patients were in the training set, 140 were in the validation set, and 140 were in the test set. One hundred eleven patients were in the external test set. The multitask DL model achieved 80.2% (89 of 111; 95% CI: 72.8, 87.6) accuracy, 62.9% (22 of 35; 95% CI: 47, 79) sensitivity, and 88.2% (67 of 76; CI: 81, 96) specificity in the classification of bone tumors as malignant or benign. The model achieved an IoU of 0.52 ± 0.34 for bounding box placements and a mean Dice score of 0.60 ± 0.37 for segmentations. The model accuracy was higher than that of two radiologic residents (71.2% and 64.9%; P = .002 and P < .001, respectively) and was comparable with that of two musculoskeletal fellowship-trained radiologists (83.8% and 82.9%; P = .13 and P = .25, respectively) in classifying a tumor as malignant or benign. Conclusion The developed multitask deep learning model allowed for accurate and simultaneous bounding box placement, segmentation, and classification of primary bone tumors on radiographs. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Carrino in this issue.

Lung nodule detection in chest X-rays using synthetic ground-truth data comparing CNN-based diagnosis to human performance

We present a method to generate synthetic thorax radiographs with realistic nodules from CT scans, and a perfect ground truth knowledge. We evaluated the detection performance of nine radiologists and two convolutional neural networks in a reader study. Nodules were artificially inserted into the lung of a CT volume and synthetic radiographs were obtained by forward-projecting the volume. Hence, our framework allowed for a detailed evaluation of CAD systems' and radiologists' performance due to the availability of accurate ground-truth labels for nodules from synthetic data. Radiographs for network training (U-Net and RetinaNet) were generated from 855 CT scans of a public dataset. For the reader study, 201 radiographs were generated from 21 nodule-free CT scans with altering nodule positions, sizes and nodule counts of inserted nodules. Average true positive detections by nine radiologists were 248.8 nodules, 51.7 false positive predicted nodules and 121.2 false negative predicted nodules. The best performing CAD system achieved 268 true positives, 66 false positives and 102 false negatives. Corresponding weighted alternative free response operating characteristic figure-of-merits (wAFROC FOM) for the radiologists range from 0.54 to 0.87 compared to a value of 0.81 (CI 0.75-0.87) for the best performing CNN. The CNN did not perform significantly better against the combined average of the 9 readers (p = 0.49). Paramediastinal nodules accounted for most false positive and false negative detections by readers, which can be explained by the presence of more tissue in this area.

Automatic Hip Detection in Anteroposterior Pelvic Radiographs-A Labelless Practical Framework

Automated detection of the region of interest (ROI) is a critical step in the two-step classification system in several medical image applications. However, key information such as model parameter selection, image annotation rules, and ROI confidence score are essential but usually not reported. In this study, we proposed a practical framework of ROI detection by analyzing hip joints seen on 7399 anteroposterior pelvic radiographs (PXR) from three diverse sources. We presented a deep learning-based ROI detection framework utilizing a single-shot multi-box detector with a customized head structure based on the characteristics of the obtained datasets. Our method achieved average intersection over union (IoU) = 0.8115, average confidence = 0.9812, and average precision with threshold IoU = 0.5 (AP50) = 0.9901 in the independent testing set, suggesting that the detected hip regions appropriately covered the main features of the hip joints. The proposed approach featured flexible loose-fitting labeling, customized model design, and heterogeneous data testing. We demonstrated the feasibility of training a robust hip region detector for PXRs. This practical framework has a promising potential for a wide range of medical image applications.

Detecting pathological features and predicting fracture risk from dual-energy X-ray absorptiometry images using deep learning

Dual-energy X-ray absorptiometry (DXA) is the gold standard imaging method for diagnosing osteoporosis in clinical practice. The DXA images are commonly used to estimate a numerical value for bone mineral density (BMD), which decreases in osteoporosis. Low BMD is a known risk factor for osteoporotic fractures. In this study, we used deep learning to identify lumbar scoliosis and structural abnormalities that potentially affect BMD but are often neglected in lumbar spine DXA analysis. In addition, we tested the approach's ability to predict fractures using only DXA images. A dataset of 2949 images gathered by Kuopio Osteoporosis Risk Factor and Prevention Study was used to train a convolutional neural network (CNN) for classification. The model was able to classify scoliosis with an AUC of 0.96 and structural abnormalities causing unreliable BMD measurement with an AUC of 0.91. It predicted fractures occurring within 5 years from the lumbar spine DXA scan with an AUC of 0.63, meeting the predictive performance of combined BMD measurements from the lumbar spine and hip. In an independent test set of 574 clinical patients, AUC for lumbar scoliosis was 0.93 and AUC for unreliable BMD measurements was 0.94. In each classification task, neural network visualizations indicated the model's predictive strategy. We conclude that deep learning could complement the well established DXA method for osteoporosis diagnostics by analyzing incidental findings and image reliability, and improve its predictive ability in the future.

Deep Generative Adversarial Networks: Applications in Musculoskeletal Imaging

In recent years, deep learning techniques have been applied in musculoskeletal radiology to increase the diagnostic potential of acquired images. Generative adversarial networks (GANs), which are deep neural networks that can generate or transform images, have the potential to aid in faster imaging by generating images with a high level of realism across multiple contrast and modalities from existing imaging protocols. This review introduces the key architectures of GANs as well as their technical background and challenges. Key research trends are highlighted, including: (a) reconstruction of high-resolution MRI; (b) image synthesis with different modalities and contrasts; (c) image enhancement that efficiently preserves high-frequency information suitable for human interpretation; (d) pixel-level segmentation with annotation sharing between domains; and (e) applications to different musculoskeletal anatomies. In addition, an overview is provided of the key issues wherein clinical applicability is challenging to capture with conventional performance metrics and expert evaluation. When clinically validated, GANs have the potential to improve musculoskeletal imaging. Keywords: Adults and Pediatrics, Computer Aided Diagnosis (CAD), Computer Applications-General (Informatics), Informatics, Skeletal-Appendicular, Skeletal-Axial, Soft Tissues/Skin © RSNA, 2021.

A scalable physician-level deep learning algorithm detects universal trauma on pelvic radiographs

Pelvic radiograph (PXR) is essential for detecting proximal femur and pelvis injuries in trauma patients, which is also the key component for trauma survey. None of the currently available algorithms can accurately detect all kinds of trauma-related radiographic findings on PXRs. Here, we show a universal algorithm can detect most types of trauma-related radiographic findings on PXRs. We develop a multiscale deep learning algorithm called PelviXNet trained with 5204 PXRs with weakly supervised point annotation. PelviXNet yields an area under the receiver operating characteristic curve (AUROC) of 0.973 (95% CI, 0.960-0.983) and an area under the precision-recall curve (AUPRC) of 0.963 (95% CI, 0.948-0.974) in the clinical population test set of 1888 PXRs. The accuracy, sensitivity, and specificity at the cutoff value are 0.924 (95% CI, 0.912-0.936), 0.908 (95% CI, 0.885-0.908), and 0.932 (95% CI, 0.919-0.946), respectively. PelviXNet demonstrates comparable performance with radiologists and orthopedics in detecting pelvic and hip fractures.