Purpose: Hip fractures are a common cause of morbidity and mortality. Automatic identification and classification of hip fractures using deep learning may improve outcomes by reducing diagnostic errors and decreasing time to operation. Methods: Hip and pelvic radiographs from 1118 studies were reviewed and 3034 hips were labeled via bounding boxes and classified as normal, displaced femoral neck fracture, nondisplaced femoral neck fracture, intertrochanteric fracture, previous ORIF, or previous arthroplasty. A deep learning-based object detection model was trained to automate the placement of the bounding boxes. A Densely Connected Convolutional Neural Network (DenseNet) was trained on a subset of the bounding box images, and its performance evaluated on a held out test set and by comparison on a 100-image subset to two groups of human observers: fellowshiptrained radiologists and orthopaedists, and senior residents in emergency medicine, radiology, and orthopaedics. Results: The binary accuracy for fracture of our model was 93.8% (95% CI, 91.3-95.8%), with sensitivity of 92.7% (95% CI, 88.7-95.6%), and specificity 95.0% (95% CI, 91.5-97.3%). Multiclass classification accuracy was 90.4% (95% CI, 87.4-92.9%). When compared to human observers, our model achieved at least expert-level classification under all conditions. Additionally, when the model was used as an aid, human performance improved, with aided resident performance approximating unaided fellowship-trained expert performance. Conclusions: Our deep learning model identified and classified hip fractures with at least expert-level accuracy, and when used as an aid improved human performance, with aided resident performance approximating that of unaided fellowship-trained attendings.

Purpose The purpose of this study was to evaluate the feasibility of using a short echo time, 3D H-1 magnetic resonance spectroscopic imaging (MRSI) sequence at 7T to assess the metabolic signature of lesions for patients with glioma. Materials and Methods 29 patients with glioma were studied. MRSI data were obtained using CHESS water suppression, spectrally-selective adiabatic inversion-recovery pulses and automatically prescribed outer-volume-suppression for lipid suppression, and spin echo slice selection (TE=30ms). An interleaved flyback echo-planar trajectory was applied to shorten the total acquisition time (~10min). Relative metabolite ratios were estimated in tumor and in normal-appearing white and gray matter (NAWM, GM). Results Levels of glutamine, myo-inositol, glycine and glutathione relative to total creatine (tCr) were significantly increased in the T2 lesions for all tumor grades compared to those in the NAWM (p < 0.05), while N-acetyl aspartate to tCr were significantly decreased (p < 0.05). In grade 2 gliomas, level of total choline-containing-compounds to tCr was significantly increased (p = 0.0137), while glutamate to tCr was significantly reduced (p = 0.0012). Conclusion The improved sensitivity of MRSI and the increased number of metabolites that can be evaluated using 7T MR scanners is of interest for evaluating patients with glioma. This study has successfully demonstrated the application of a short-echo spin-echo MRSI sequence to detect characteristic differences in regions of tumor versus normal appearing brain.

No abstract

Purpose: To quantify normal, in vivo tibio‐femoral knee joint kinematics in multiple weight bearing positions using non‐invasive, high‐resolution MRI and discuss the potential of developing future kinematic methods to assess patients with abnormal joint pathologies. Methods: Ten volunteers with clinically normal knees pushed inferiorly on the footplate of a weight bearing apparatus inside the MR scanner. The volunteers held the weight (133 N) for five scans as the knee motion was evaluated from 0° to 60° of flexion. Full extension was set as the zero point for all measured parameters. Using 3D reconstructions, tibia motion relative to the femur and flexion angle was measured as varus–valgus angle, axial rotation, anterior–posterior translation, and medial–lateral translation. Medial and lateral compartment tibio‐femoral contact areas were examined and centroids of the contract areas were calculated. Results: Tibial internal rotation averaged 4.8° at 40° of flexion and then decreased. Tibial valgus increased by 8° at 60° of flexion. Femoral roll back also increased to 18.5 mm average at 60° of flexion, while the tibia translated medially 2.5 mm. Medial compartment femoro‐tibial contact area started at 374 mm2 and decreased to 308 mm2 with flexion of 60°, while lateral compartment contact area did not change significantly from 276 mm2. Conclusions: Results correlate with previous studies of knee kinematics while providing greater three‐dimensional detail. MR imaging allows excellent non‐invasive evaluation of knee joint kinematics with weight bearing. This tool may potentially be used for assessing knee kinematics in patients with knee pathology. © 2003 Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved.

Disclosures of Conflicts of Interest: C.E.v.S. disclosed no relevant relationships. J.H.S. disclosed no relevant relationships. F.L. disclosed no relevant relationships. E.O. disclosed no relevant relationships. P.M.J. disclosed no relevant relationships. L.N. disclosed no relevant relationships. M.P. disclosed no relevant relationships. S.C.F. disclosed no relevant relationships. M.C.N. disclosed no relevant relationships. T.M.L. disclosed no relevant relationships. V.P. disclosed no relevant relationships.