A radiograph-based deep learning model improves radiologists' performance for classification of histological types of primary bone tumors: A multicenter study.

PURPOSE: To develop a deep learning (DL) model for classifying histological types of primary bone tumors (PBTs) using radiographs and evaluate its clinical utility in assisting radiologists. METHODS: This retrospective study included 878 patients with pathologically confirmed PBTs from two centers (638, 77, 80, and 83 for the training, validation, internal test, and external test sets, respectively). We classified PBTs into five categories by histological types: chondrogenic tumors, osteogenic tumors, osteoclastic giant cell-rich tumors, other mesenchymal tumors of bone, or other histological types of PBTs. A DL model combining radiographs and clinical features based on the EfficientNet-B3 was developed for five-category classification. The area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity were calculated to evaluate model performance. The clinical utility of the model was evaluated in an observer study with four radiologists. RESULTS: The combined model achieved a macro average AUC of 0.904/0.873, with an accuracy of 67.5 %/68.7 %, a macro average sensitivity of 66.9 %/57.2 %, and a macro average specificity of 92.1 %/91.6 % on the internal/external test set, respectively. Model-assisted analysis improved accuracy, interpretation time, and confidence for junior (50.6 % vs. 72.3 %, 53.07[s] vs. 18.55[s] and 3.10 vs. 3.73 on a 5-point Likert scale [P < 0.05 for each], respectively) and senior radiologists (68.7 % vs. 75.3 %, 32.50[s] vs. 21.42[s] and 4.19 vs. 4.37 [P < 0.05 for each], respectively). CONCLUSION: The combined DL model effectively classified histological types of PBTs and assisted radiologists in achieving better classification results than their independent visual assessment.

Automatic segmentation of fat metaplasia on sacroiliac joint MRI using deep learning.

OBJECTIVE: To develop a deep learning (DL) model for segmenting fat metaplasia (FM) on sacroiliac joint (SIJ) MRI and further develop a DL model for classifying axial spondyloarthritis (axSpA) and non-axSpA. MATERIALS AND METHODS: This study retrospectively collected 706 patients with FM who underwent SIJ MRI from center 1 (462 axSpA and 186 non-axSpA) and center 2 (37 axSpA and 21 non-axSpA). Patients from center 1 were divided into the training, validation, and internal test sets (n = 455, 64, and 129). Patients from center 2 were used as the external test set. We developed a UNet-based model to segment FM. Based on segmentation results, a classification model was built to distinguish axSpA and non-axSpA. Dice Similarity Coefficients (DSC) and area under the curve (AUC) were used for model evaluation. Radiologists' performance without and with model assistance was compared to assess the clinical utility of the models. RESULTS: Our segmentation model achieved satisfactory DSC of 81.86% ± 1.55% and 85.44% ± 6.09% on the internal cross-validation and external test sets. The classification model yielded AUCs of 0.876 (95% CI: 0.811-0.942) and 0.799 (95% CI: 0.696-0.902) on the internal and external test sets, respectively. With model assistance, segmentation performance was improved for the radiological resident (DSC, 75.70% vs. 82.87%, p < 0.05) and expert radiologist (DSC, 85.03% vs. 85.74%, p > 0.05). CONCLUSIONS: DL is a novel method for automatic and accurate segmentation of FM on SIJ MRI and can effectively increase radiologist's performance, which might assist in improving diagnosis and progression of axSpA. CRITICAL RELEVANCE STATEMENT: DL models allowed automatic and accurate segmentation of FM on sacroiliac joint MRI, which might facilitate quantitative analysis of FM and have the potential to improve diagnosis and prognosis of axSpA. KEY POINTS: • Deep learning was used for automatic segmentation of fat metaplasia on MRI. • UNet-based models achieved automatic and accurate segmentation of fat metaplasia. • Automatic segmentation facilitates quantitative analysis of fat metaplasia to improve diagnosis and prognosis of axial spondyloarthritis.

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.

Improved Performance of Compartments in Detecting the Activity of Axial Spondyloarthritis Based on IVIM DWI with Optimized Threshold b Value.

PURPOSE: To explore the diagnostic performance of the optimized threshold b values on IVIM to detect the activity in axial spondyloarthritis (axSpA) patients. METHOD: 40 axSpA patients in the active group, 144 axSpA patients in the inactive group, and 20 healthy volunteers were used to evaluate the tissue diffusion coefficient (D slow), perfusion fraction (f), and pseudodiffusion coefficient (D fast) with b thresholds of 10, 20, and 30 s/mm2. The Kruskal-Wallis test and one way ANOVA test was used to compare the different activity among the three groups in axSpA patients, and receiver operating characteristic (ROC) curve analysis was applied to evaluate the performance for D slow, f, and D fast to detect the activity in axSpA patients, respectively. RESULTS: D slow demonstrated a statistical difference between two groups (P < 0.05) with all threshold b values. With the threshold b value of 30 s/mm2, f could discriminate the active from control groups (P < 0.05). D slow had similar performance between the active and the inactive groups with threshold b values of 10, 20, and 30 s/mm2 (AUC: 0.877, 0.882, and 0.881, respectively, all P < 0.017). Using the optimized threshold b value of 30 s/mm2, f showed the best performance to separate the active from the inactive and the control groups with AUC of 0.613 and 0.738 (both P < 0.017) among all threshold b values. CONCLUSION: D slow and f exhibited increased diagnostic performance using the optimized threshold b value of 30 s/mm2 compared with 10 and 20 s/mm2, whereas D fast did not.

Prediction of Treatment Response According to ASAS-EULAR Management Recommendations in 1 Year for Hip Involvement in Axial Spondyloarthritis Based on MRI and Clinical Indicators.

Background: To predict the treatment response for axial spondyloarthritis (axSpA) with hip involvement in 1 year based on MRI and clinical indicators. Methods: A total of 77 axSpA patients with hip involvement (60 males; median age, 25 years; interquartile, 22-31 years old) were treated with a drug recommended by the Assessment of SpondyloArthritis international Society and the European League Against Rheumatism (ASAS-EULAR) management. They were prospectively enrolled according to Assessment in SpondyloArthritis international Society (ASAS) criteria. Clinical indicators, including age, gender, disease duration, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP), were collected at baseline and in 3 months to 1-year follow-up. Treatment response was evaluated according to ASAS response criteria. MRI indicators consisting of bone marrow edema (BME) in acetabulum and femoral head, hip effusion, fat deposition, thickened synovium, bone erosion, bone proliferation, muscle involvement, enthesitis and bony ankylosis were assessed at baseline. Spearman's correlation analysis was utilized for indicator selection. The selected clinical and MRI indicators were integrated with previous clinical knowledge to develop multivariable logistic regression models. Receiver operator characteristic curve and area under the curve (AUC) were used to assess the performance of the constructed models. Results: The model combining MR indicators comprising hip effusion, BME in acetabulum and femoral head and clinical indicators consisting of disease duration, ESR and CRP yielded AUC values of 0.811 and 0.753 for the training and validation cohorts, respectively. Conclusion: The model combining MRI and clinical indicators could predict treatment response for axSpA with hip involvement in 1 year.

Intravoxel Incoherent Motion Imaging on Sacroiliitis in Patients With Axial Spondyloarthritis: Correlation With Perfusion Characteristics Based on Dynamic Contrast-Enhanced Magnetic Resonance Imaging.

BACKGROUND: To prospectively explore the relationship between intravoxel incoherent motion (IVIM) diffusion-weighted imaging (DWI) and dynamic contrast-enhanced MRI (DCE-MRI) parameters of sacroiliitis in patients with axial spondyloarthritis (axSpA). METHODS: Patients with initially diagnosed axSpA prospectively underwent on 3.0 T MRI of sacroiliac joint (SIJ). The IVIM parameters (D, f, D *) were calculated using biexponential analysis. K trans, K ep, V e, and V p from DCE-MRI were obtained in SIJ. The uni-variable and multi-variable linear regression analyses were used to evaluate the correlation between the parameters from these two imaging methods after controlling confounders, such as bone marrow edema (BME), age, agenda, scopes, and localization of lesions, and course of the disease. Then, their correlations were measured by calculating the Pearson's correlation coefficient (r). RESULTS: The study eventually enrolled 234 patients (178 men, 56 women; mean age, 28.51 ± 9.50 years) with axSpA. With controlling confounders, D was independently related to K trans (regression coefficient [b] = 27.593, p < 0.001), K ep (b = -6.707, p = 0.021), and V e (b = 131.074, p = 0.003), whereas f and D * had no independent correlation with the parameters from DCE MRI. The correlations above were exhibited with Pearson's correlation coefficients (r) (r = 0.662, -0.408, and 0.396, respectively, all p < 0.001). CONCLUSION: There were independent correlations between D derived from IVIM DWI and K trans, K ep, and V e derived from DCE-MRI. The factors which affect their correlations mainly included BME, gender, and scopes of lesions.