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.

Automated assessment of cardiac pathologies on cardiac MRI using T1-mapping and late gadolinium phase sensitive inversion recovery sequences with deep learning.

BACKGROUND: A deep learning (DL) model that automatically detects cardiac pathologies on cardiac MRI may help streamline the diagnostic workflow. To develop a DL model to detect cardiac pathologies on cardiac MRI T1-mapping and late gadolinium phase sensitive inversion recovery (PSIR) sequences were used. METHODS: Subjects in this study were either diagnosed with cardiac pathology (n = 137) including acute and chronic myocardial infarction, myocarditis, dilated cardiomyopathy, and hypertrophic cardiomyopathy or classified as normal (n = 63). Cardiac MR imaging included T1-mapping and PSIR sequences. Subjects were split 65/15/20% for training, validation, and hold-out testing. The DL models were based on an ImageNet pretrained DenseNet-161 and implemented using PyTorch and fastai. Data augmentation with random rotation and mixup was applied. Categorical cross entropy was used as the loss function with a cyclic learning rate (1e-3). DL models for both sequences were developed separately using similar training parameters. The final model was chosen based on its performance on the validation set. Gradient-weighted class activation maps (Grad-CAMs) visualized the decision-making process of the DL model. RESULTS: The DL model achieved a sensitivity, specificity, and accuracy of 100%, 38%, and 88% on PSIR images and 78%, 54%, and 70% on T1-mapping images. Grad-CAMs demonstrated that the DL model focused its attention on myocardium and cardiac pathology when evaluating MR images. CONCLUSIONS: The developed DL models were able to reliably detect cardiac pathologies on cardiac MR images. The diagnostic performance of T1 mapping alone is particularly of note since it does not require a contrast agent and can be acquired quickly.

Diagnostic Delay in Patients with Osteoid Osteoma.

PURPOSE: To assess diagnostic delay in patients with osteoid osteoma and to analyze influencing factors. MATERIALS AND METHODS: All patients treated for osteoid osteoma at our tertiary referral center between December 1997 and February 2021 were retrospectively identified (n = 302). The diagnosis was verified by an expert panel of radiologists and orthopedic surgeons. The exclusion criteria were post-interventional recurrence, missing data on symptom onset, and lack of pretherapeutic CT images. Clinical parameters were retrieved from the local clinical information system. CT and MR images were assessed by a senior specialist in musculoskeletal radiology. RESULTS: After all exclusions, we studied 162 patients (mean age: 24 ±â€Š11 years, 115 men). The average diagnostic delay was 419 ±â€Š485 days (median: 275 days; range: 21-4503 days). Gender, patient age, presence of nocturnal pain, positive aspirin test, extent of bone sclerosis, and location of the tumor within bone and relative to joints did not influence diagnostic delay (p > 0.05). It was, however, positively correlated with nidus size (r = 0.26; p < 0.001) and was shorter with affection of long tubular bones compared to all other sites (p = 0.04). If osteoid osteoma was included in the initial differential diagnoses, the diagnostic delay was also shorter (p = 0.007). CONCLUSION: The diagnostic delay in patients with osteoid osteoma is independent of demographics, clinical parameters, and most imaging parameters. A long average delay of more than one year suggests low awareness of the disease among physicians. Patients with unclear imaging findings should thus be referred to a specialized musculoskeletal center or an expert in the field should be consulted in a timely manner. KEY POINTS: · In this retrospective study of 162 patients treated for osteoid osteoma, the median diagnostic delay was 275 days (range: 21-4503 days).. · Gender, age, presence of nocturnal pain, positive aspirin test, extent of bone sclerosis, and location of the tumor did not influence the diagnostic delay (p > 0.05).. · Diagnostic delay was positively correlated with nidus size (r = 0.26; p < 0.001) and was shorter with affection of long tubular bones compared to all other sites (376 ±â€Š485 vs. 560 ±â€Š462 days; p = 0.04)..

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.

Coronary plaque characterization assessed by delayed enhancement dual-layer spectral CT angiography and optical coherence tomography.

This study aimed to prospectively evaluate delayed enhancement imaging by spectral computed tomography using soluble iodine containing contrast media to improve the in vivo characterization of coronary plaque types based on the quantification of delayed iodine enhancement. Patients with known or suspected coronary artery disease (CAD) underwent spectral coronary CT-angiography (SCCTA). Absolute delayed iodine enhancement in all visible coronary plaques was assessed. Patients with significant CAD (> 50% stenosis) further underwent invasive coronary angiography (ICA) including optical coherence tomography (OCT). We identified 50 non-calcified coronary plaques in 72 patients undergoing SCCTA. 17 patients with significant CAD underwent further ICA including OCT imaging. In those, we were able to match 35 plaques by both SCCTA and OCT. Based on OCT imaging, 22/35 matched plaques (63%) were characterized as high-risk coronary plaques (thin-cap fibroatheroma n = 2, fibroatheroma n = 20), whereas 13/35 (37%) were characterized as low-risk plaques (fibrocalcific lesion n = 3, fibrous plaques n = 9, and early-onset fibroatheroma n = 1). All plaques showed similar HU's and could not be classified into high-risk or low-risk plaques by conventional CT measures. Minimal delayed iodine enhancement within plaques as quantified by SCCTA demonstrated significantly lower values in high-risk as compared to low-risk coronary plaques (1.0 ± 1.5 mg/ml vs. 2.2 ± 1.1 mg/ml, p = 0.021) which allowed estimation of high-risk plaques with high sensitivity and moderate specificity (77% and 56%). Measurement of delayed enhancement iodine uptake within stable coronary artery plaques using dual-layer SCCTA might contribute to a more precise estimation of plaque vulnerability surpassing conventional CT techniques.

Federated machine learning for a facilitated implementation of Artificial Intelligence in healthcare - a proof of concept study for the prediction of coronary artery calcification scores.

The implementation of Artificial Intelligence (AI) still faces significant hurdles and one key factor is the access to data. One approach that could support that is federated machine learning (FL) since it allows for privacy preserving data access. For this proof of concept, a prediction model for coronary artery calcification scores (CACS) has been applied. The FL was trained based on the data in the different institutions, while the centralized machine learning model was trained on one allocation of data. Both algorithms predict patients with risk scores ≥5 based on age, biological sex, waist circumference, dyslipidemia and HbA1c. The centralized model yields a sensitivity of c. 66% and a specificity of c. 70%. The FL slightly outperforms that with a sensitivity of 67% while slightly underperforming it with a specificity of 69%. It could be demonstrated that CACS prediction is feasible via both, a centralized and an FL approach, and that both show very comparable accuracy. In order to increase accuracy, additional and a higher volume of patient data is required and for that FL is utterly necessary. The developed "CACulator" serves as proof of concept, is available as research tool and shall support future research to facilitate AI implementation.

Comparison of CT, MRI, and F-18 FDG PET/CT for initial N-staging of oral squamous cell carcinoma: a cost-effectiveness analysis.

BACKGROUND AND PURPOSE: Treatment of oral squamous cell carcinoma (OSCC) is based on clinical exam, biopsy, and a precise imaging-based TNM-evaluation. A high sensitivity and specificity for magnetic resonance imaging (MRI) and F-18 FDG PET/CT are reported for N-staging. Nevertheless, staging of oral squamous cell carcinoma is most often based on computed tomography (CT) scans. This study aims to evaluate cost-effectiveness of MRI and PET/CT compared to standard of care imaging in initial staging of OSCC within the US Healthcare System. METHODS: A decision model was constructed using quality-adjusted life years (QALYs) and overall costs of different imaging strategies including a CT of the head, neck, and the thorax, MRI of the neck with CT of the thorax, and whole body F-18 FDG PET/CT using Markov transition simulations for different disease states. Input parameters were derived from literature and willingness to pay (WTP) was set to US $100,000/QALY. Deterministic sensitivity analysis of diagnostic parameters and costs was performed. Monte Carlo modeling was used for probabilistic sensitivity analysis. RESULTS: In the base-case scenario, total costs were at US $239,628 for CT, US $240,001 for MRI, and US $239,131 for F-18 FDG PET/CT whereas the model yielded an effectiveness of 5.29 QALYs for CT, 5.30 QALYs for MRI, and 5.32 QALYs for F-18 FDG PET/CT respectively. F-18 FDG PET/CT was the most cost-effective strategy over MRI as well as CT, and MRI was the cost-effective strategy over CT. Deterministic and probabilistic sensitivity analysis showed high robustness of the model with incremental cost effectiveness ratio remaining below US $100,000/QALY for a wide range of variability of input parameters. CONCLUSION: F-18 FDG PET/CT is the most cost-effective strategy in the initial N-staging of OSCC when compared to MRI and CT. Despite less routine use, both whole body PET/CT and MRI are cost-effective modalities in the N-staging of OSCC. Based on these findings, the implementation of PET/CT for initial staging could be suggested to help reduce costs while increasing effectiveness in OSCC.

Development and evaluation of machine learning models based on X-ray radiomics for the classification and differentiation of malignant and benign bone tumors.

OBJECTIVES: To develop and validate machine learning models to distinguish between benign and malignant bone lesions and compare the performance to radiologists. METHODS: In 880 patients (age 33.1 ± 19.4 years, 395 women) diagnosed with malignant (n = 213, 24.2%) or benign (n = 667, 75.8%) primary bone tumors, preoperative radiographs were obtained, and the diagnosis was established using histopathology. Data was split 70%/15%/15% for training, validation, and internal testing. Additionally, 96 patients from another institution were obtained for external testing. Machine learning models were developed and validated using radiomic features and demographic information. The performance of each model was evaluated on the test sets for accuracy, area under the curve (AUC) from receiver operating characteristics, sensitivity, and specificity. For comparison, the external test set was evaluated by two radiology residents and two radiologists who specialized in musculoskeletal tumor imaging. RESULTS: The best machine learning model was based on an artificial neural network (ANN) combining both radiomic and demographic information achieving 80% and 75% accuracy at 75% and 90% sensitivity with 0.79 and 0.90 AUC on the internal and external test set, respectively. In comparison, the radiology residents achieved 71% and 65% accuracy at 61% and 35% sensitivity while the radiologists specialized in musculoskeletal tumor imaging achieved an 84% and 83% accuracy at 90% and 81% sensitivity, respectively. CONCLUSIONS: An ANN combining radiomic features and demographic information showed the best performance in distinguishing between benign and malignant bone lesions. The model showed lower accuracy compared to specialized radiologists, while accuracy was higher or similar compared to residents. KEY POINTS: • The developed machine learning model could differentiate benign from malignant bone tumors using radiography with an AUC of 0.90 on the external test set. • Machine learning models that used radiomic features or demographic information alone performed worse than those that used both radiomic features and demographic information as input, highlighting the importance of building comprehensive machine learning models. • An artificial neural network that combined both radiomic and demographic information achieved the best performance and its performance was compared to radiology readers on an external test set.

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.

Detection of Bone Marrow Edema in Patients with Osteoid Osteoma Using Three-Material Decomposition with Dual-Layer Spectral CT.

The aim of this study is to assess whether perifocal bone marrow edema (BME) in patients with osteoid osteoma (OO) can be accurately detected on dual-layer spectral CT (DLCT) with three-material decomposition. To that end, 18 patients with OO (25.33 ± 12.44 years; 7 females) were pairwise-matched with 18 patients (26.72 ± 9.65 years; 9 females) admitted for suspected pathologies other than OO in the same anatomic location but negative imaging findings. All patients were examined with DLCT and MRI. DLCT data was decomposed into hydroxyapatite and water- and fat-equivalent volume fraction maps. Two radiologists assessed DLCT-based volume fraction maps for the presence of perifocal BME, using a Likert scale (1 = no edema; 2 = likely no edema; 3 = likely edema; 4 = edema). Accuracy, sensitivity, and specificity for the detection of BME on DLCT were analyzed using MR findings as standard of reference. For the detection of BME in patients with OO, DLCT showed a sensitivity of 0.92, a specificity of 0.94, and an accuracy of 0.92 for both radiologists. Interreader agreement for the assessment of BME with DLCT was substantial (weighted κ = 0.78; 95% CI, 0.59, 0.94). DLCT with material-specific volume fraction maps allowed accurate detection of BME in patients with OO. This may spare patients additional examinations and facilitate the diagnosis of OO.

Prognostic Assessment in High-Grade Soft-Tissue Sarcoma Patients: A Comparison of Semantic Image Analysis and Radiomics.

BACKGROUND: In patients with soft-tissue sarcomas of the extremities, the treatment decision is currently regularly based on tumor grading and size. The imaging-based analysis may pose an alternative way to stratify patients' risk. In this work, we compared the value of MRI-based radiomics with expert-derived semantic imaging features for the prediction of overall survival (OS). METHODS: Fat-saturated T2-weighted sequences (T2FS) and contrast-enhanced T1-weighted fat-saturated (T1FSGd) sequences were collected from two independent retrospective cohorts (training: 108 patients; testing: 71 patients). After preprocessing, 105 radiomic features were extracted. Semantic imaging features were determined by three independent radiologists. Three machine learning techniques (elastic net regression (ENR), least absolute shrinkage and selection operator, and random survival forest) were compared to predict OS. RESULTS: ENR models achieved the best predictive performance. Histologies and clinical staging differed significantly between both cohorts. The semantic prognostic model achieved a predictive performance with a C-index of 0.58 within the test set. This was worse compared to a clinical staging system (C-index: 0.61) and the radiomic models (C-indices: T1FSGd: 0.64, T2FS: 0.63). Both radiomic models achieved significant patient stratification. CONCLUSIONS: T2FS and T1FSGd-based radiomic models outperformed semantic imaging features for prognostic assessment.

Patients with Type 2 Diabetes Exhibit a More Mineralized Deep Cartilage Layer Compared with Nondiabetic Controls: A Pilot Study.

OBJECTIVE: To assess differences in biochemical composition of the deep cartilage layer in subjects with type 2 diabetes mellitus (T2DM) and nondiabetic controls using UTE (ultra-short echo time) T2* mapping and to investigate the association of vascular health and UTE T2* measurements. DESIGN: Ten subjects with T2DM matched for age, sex, and body mass index with 10 nondiabetic controls. A 3D UTE sequence with 6 echo times was acquired using 3T magnetic resonance imaging of the knee. For UTE T2* analysis, the deep cartilage layer was segmented and analyzed in 5 compartments (patella, medial, and lateral femur and tibia). The ankle brachial index (ABI) was obtained in all subjects. Linear regression analyses were used to assess associations of T2DM and UTE T2* relaxation times and the associations of ABI measurements and UTE measurements. RESULTS: Compared with nondiabetic controls, T2DM subjects had significantly lower mean T2*-UTE in the patella (mean difference 4.87 ms; 95% confidence interval [CI] 1.09-8.65; P = 0.015), the lateral tibia (mean difference 2.26 ms; 95% CI 0.06-4.45; P = 0.045), and the lateral femur (mean difference 4.96 ms; 95% CI 0.19-9.73; P = 0.043). Independent of diabetic status, subjects with higher ABI values, indicating better vascular health, had higher T2*-UTE of the patella (coefficient 15.2; 95% CI 3.3-21.4; P = 0.017), the medial tibia (coefficient 9.8; 95% CI 1.0-18.6; P = 0.031), and the lateral femur (coefficient 18.8; 95% CI 3.3-34.3; P = 0.021). CONCLUSIONS: T2*-UTE measurements of the deep cartilage layer were consistently lower in subjects with T2DM and in subjects with impaired vascular health, likely indicating increased mineralization of this layer.

Coronary calcium scoring assessed on native screening chest CT imaging as predictor for outcome in COVID-19: An analysis of a hospitalized German cohort.

BACKGROUND: Since the outbreak of the COVID-19 pandemic, a number of risk factors for a poor outcome have been identified. Thereby, cardiovascular comorbidity has a major impact on mortality. We investigated whether coronary calcification as a marker for coronary artery disease (CAD) is appropriate for risk prediction in COVID-19. METHODS: Hospitalized patients with COVID-19 (n = 109) were analyzed regarding clinical outcome after native computed tomography (CT) imaging for COVID-19 screening. CAC (coronary calcium score) and clinical outcome (need for intensive care treatment or death) data were calculated following a standardized protocol. We defined three endpoints: critical COVID-19 and transfer to ICU, fatal COVID-19 and death, composite endpoint critical and fatal COVID-19, a composite of ICU treatment and death. We evaluated the association of clinical outcome with the CAC. Patients were dichotomized by the median of CAC. Hazard ratios and odds ratios were calculated for the events death or ICU or a composite of death and ICU. RESULTS: We observed significantly more events for patients with CAC above the group's median of 31 for critical outcome (HR: 1.97[1.09,3.57], p = 0.026), for fatal outcome (HR: 4.95[1.07,22.9], p = 0.041) and the composite endpoint (HR: 2.31[1.28,4.17], p = 0.0056. Also, odds ratio was significantly increased for critical outcome (OR: 3.01 [1.37, 6.61], p = 0.01) and for fatal outcome (OR: 5.3 [1.09, 25.8], p = 0.02). CONCLUSION: The results indicate a significant association between CAC and clinical outcome in COVID-19. Our data therefore suggest that CAC might be useful in risk prediction in patients with COVID-19.

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

Background A multitask deep learning model might be useful in large epidemiologic studies wherein detailed structural assessment of osteoarthritis still relies on expert radiologists' readings. The potential of such a model in clinical routine should be investigated. Purpose To develop a multitask deep learning model for grading radiographic hip osteoarthritis features on radiographs and compare its performance to that of attending-level radiologists. Materials and Methods This retrospective study analyzed hip joints seen on weight-bearing anterior-posterior pelvic radiographs from participants in the Osteoarthritis Initiative (OAI). Participants were recruited from February 2004 to May 2006 for baseline measurements, and follow-up was performed 48 months later. Femoral osteophytes (FOs), acetabular osteophytes (AOs), and joint-space narrowing (JSN) were graded as absent, mild, moderate, or severe according to the Osteoarthritis Research Society International atlas. Subchondral sclerosis and subchondral cysts were graded as present or absent. The participants were split at 80% (n = 3494), 10% (n = 437), and 10% (n = 437) by using split-sample validation into training, validation, and testing sets, respectively. The multitask neural network was based on DenseNet-161, a shared convolutional features extractor trained with multitask loss function. Model performance was evaluated in the internal test set from the OAI and in an external test set by using temporal and geographic validation consisting of routine clinical radiographs. Results A total of 4368 participants (mean age, 61.0 years ± 9.2 [standard deviation]; 2538 women) were evaluated (15 364 hip joints on 7738 weight-bearing anterior-posterior pelvic radiographs). The accuracy of the model for assessing these five features was 86.7% (1333 of 1538) for FOs, 69.9% (1075 of 1538) for AOs, 81.7% (1257 of 1538) for JSN, 95.8% (1473 of 1538) for subchondral sclerosis, and 97.6% (1501 of 1538) for subchondral cysts in the internal test set, and 82.7% (86 of 104) for FOS, 65.4% (68 of 104) for AOs, 80.8% (84 of 104) for JSN, 88.5% (92 of 104) for subchondral sclerosis, and 91.3% (95 of 104) for subchondral cysts in the external test set. Conclusion A multitask deep learning model is a feasible approach to reliably assess radiographic features of hip osteoarthritis. © RSNA, 2020 Online supplemental material is available for this article.

Postoperative MRI Findings and Associated Pain Changes After Arthroscopic Surgery for Femoroacetabular Impingement.

OBJECTIVE. The purpose of this study is to describe postoperative MRI findings after femoroacetabular impingement surgery in correlation with pain changes and surgical findings. SUBJECTS AND METHODS. We prospectively enrolled 42 patients (43 hips) who were scheduled for FAI surgery. Pre- and postoperative MR images were obtained using a 3-T MRI system. Changes in pain scores were assessed using the hip dysfunction and osteoarthritis outcome score. MR images were evaluated for the presence of acetabuloplasty or femoroplasty, presence of chondral and labral repair surgery, bone marrow edema, subchondral cysts, chondral defects, labral tears, capsular defects, and effusion. The optimal orientation to detect these changes was noted. Imaging findings were compared with pain score changes using linear regression analysis. Sensitivity and specificity were assessed using surgical correlation as the reference standard. RESULTS. Increased acetabular bony débridement length was associated with decreased improvement in pain scores (coefficient, -2.07; 95% CI, -3.53 to -0.62; p = 0.008), whereas other imaging findings were not significantly different. Femoroplasty and capsular alterations were best detected on oblique axial sequences; acetabuloplasty and cartilage and labral repair were best seen on sagittal sequences. MRI showed excellent sensitivity (100%) and specificity (100%) for detecting labral repair and excellent sensitivity for detecting femoroplasty (98%). Sensitivity and specificity were lower for detecting acetabuloplasty (83% and 80%, respectively) and chondral repair (75% and 54%, respectively). CONCLUSION. Arthroscopic acetabuloplasty showed a greater association with postoperative pain than did other aspects of surgical correction for femoroacetabular impingement. Femoroplasty and labral repair were reliably diagnosed on 3-T MRI; however, limitations were found in the evaluation of acetabular chondral repair.