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.

No abstract

BackgroundWell-informed decisions about how to best treat patients with axial spondyloarthritis (SpA) regularly include an evaluation of the sacroiliac joints (SIJ) on plain radiographs. However, grading radiographic findings correctly has proven to be a considerable challenge to expert readers as well as to state-of-the-art convolutional neural networks (CNNs). A method to reduce image information to the clinically relevant core would undoubtedly lead to more accurate results. We, therefore, trained a CNN only to detect SIJs on radiographs and evaluated its potential as a preprocessing pipeline in the automated classification of SpA.Materials and MethodsWe employed a CNN of the RetinaNet architecture, which was trained on a total of 423 plain radiographs of the sacroiliac joints (SIJs). Images were taken from two completely independent datasets. Training and tuning were performed on image data from the Patients With Axial Spondyloarthritis (PROOF) study and testing was executed using images from the German Spondyloarthritis Inception Cohort (GESPIC). Performance was evaluated by manual review and standard object detection metrics from PASCAL and Microsoft COCO.ResultsThe CNN produced excellent results in detecting SIJs on the tuning (n =106) and on the holdout dataset (n =140). Object detection metrics for the tuning data were AP = 0.996 and mAP = 0.538; values for the independent holdout data were AP = 0.981 and mAP = 0.515. ConclusionsThe developed CNN was highly accurate in detecting SIJs on radiographs. Such a model could increase the reliability of deep learning-based algorithms in detecting and grading SpA.

Purpose Successful utilization of anatomical templates in the evaluation of diffusion-weighted neuroimaging-studies requires accurate registration of intra-individual datasets. We investigated the feasibility of structural MRI image registration onto single-shot and read-out segmented echo-planar diffusion-weighted imaging variants for use in tractographic samplings of the visual system, in particular the optic nerve. Methods Slab volumes of the optic nerve pathway from thirteen volunteers were acquired and preprocessed. Three neuroradiologists marked landmarks (ROIs, regions of interest) on two diffusion-weighted and one structural dataset. Structural ROIs were respectively registered (6/12 degrees of freedom, DOF) onto single-shot (ss-EPI) and readout-segmented (rs-EPI) volumes. All six ROI/FOD (fibre orientation distribution) combinations underwent a targeted tractography task (MRtrix3: iFOD2). Results Inter-rater reliability for ROI-placement was highest in VIBE images (0.66 to 0.95, mean 0.85) and lower in both ss-EPI (mean 0.77) and rs-EPI (0.46 to 0.84, mean 0.64). Sufficient FOD generation in the optic nerve was successful in 12/26 of all cases for ss-EPI volumes (46.2% of cases) and in 18/26 for rs-EPI volumes (69.2%). Spatial shift of VIBE-drawn ROI-coordinates after 6-DOF registration was highest for ss-EPI-targets (medians: 1.15 to 1.4 mm; rs-EPI 0.82 to 0.93 mm), whereas 12-DOF registration caused less spatial shifts (ss-EPI: 0.64 to 1.03 mm; rs-EPI: 0.58 mm to 0.79 mm). Tractography results revealed no significant differences between ss-EPI and rs-EPI sequences on cases with mutual generation of sufficient FODs (n=10). Conclusion Structurally-placed ROIs (highest inter-rater reliability) with 6-DOF registration onto rs-EPI targets (highest FOD-generation rate) seems most suitable for white matter template generation.