Automated Detection of Brain Metastases on T1-Weighted MRI Using a Convolutional Neural Network: Impact of Volume Aware Loss and Sampling Strategy.

BACKGROUND: Detection of brain metastases (BM) and segmentation for treatment planning could be optimized with machine learning methods. Convolutional neural networks (CNNs) are promising, but their trade-offs between sensitivity and precision frequently lead to missing small lesions. HYPOTHESIS: Combining volume aware (VA) loss function and sampling strategy could improve BM detection sensitivity. STUDY TYPE: Retrospective. POPULATION: A total of 530 radiation oncology patients (55% women) were split into a training/validation set (433 patients/1460 BM) and an independent test set (97 patients/296 BM). FIELD STRENGTH/SEQUENCE: 1.5 T and 3 T, contrast-enhanced three-dimensional (3D) T1-weighted fast gradient echo sequences. ASSESSMENT: Ground truth masks were based on radiotherapy treatment planning contours reviewed by experts. A U-Net inspired model was trained. Three loss functions (Dice, Dice + boundary, and VA) and two sampling methods (label and VA) were compared. Results were reported with Dice scores, volumetric error, lesion detection sensitivity, and precision. A detected voxel within the ground truth constituted a true positive. STATISTICAL TESTS: McNemar's exact test to compare detected lesions between models. Pearson's correlation coefficient and Bland-Altman analysis to compare volume agreement between predicted and ground truth volumes. Statistical significance was set at P ≤ 0.05. RESULTS: Combining VA loss and VA sampling performed best with an overall sensitivity of 91% and precision of 81%. For BM in the 2.5-6 mm estimated sphere diameter range, VA loss reduced false negatives by 58% and VA sampling reduced it further by 30%. In the same range, the boundary loss achieved the highest precision at 81%, but a low sensitivity (24%) and a 31% Dice loss. DATA CONCLUSION: Considering BM size in the loss and sampling function of CNN may increase the detection sensitivity regarding small BM. Our pipeline relying on a single contrast-enhanced T1-weighted MRI sequence could reach a detection sensitivity of 91%, with an average of only 0.66 false positives per scan. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

Inflammation and Hypervascularization in a Large Animal Model of Knee Osteoarthritis: Imaging with Pathohistologic Correlation.

PURPOSE: To evaluate if synovial inflammation and hypervascularization are present in a dog model of knee osteoarthritis and can be detected on conventional magnetic resonance imaging (MRI), dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), contrast-enhanced magnetic resonance imaging (CE-MRI), and quantitative digital subtraction angiography (Q-DSA) imaging. MATERIALS AND METHODS: Six dogs underwent MRI and angiography of both knees before and 12 weeks after right knee anterior cruciate ligament injury. Synovial vascularity was evaluated on CE-MRI, DCE-MRI, and Q-DSA by 2 independent observers. Synovial inflammation and vascularity were histologically scored independently. Cartilage lesions and osteophytes were analyzed macroscopically, and cartilage volumetry was analyzed by MRI. Vascularity and osteoarthritis markers on imaging were compared before and after osteoarthritis generation, and between the osteoarthritis model and the control knee, using linear mixed models accounting for within-dog correlation. RESULTS: In all knees, baseline imaging showed no abnormalities. Control knees did not develop significant osteoarthritis changes, synovial inflammation, or hypervascularization. In osteoarthritis knees, mean synovial enhancement score on CE-MR imaging increased by 13.1 ± 0.59 (P < .0001); mean synovial inflammation variable increased from 47.33 ± 18.61 to 407.97 ± 18.61 on DCE-MR imaging (P < .0001); and area under the curve on Q-DSA increased by 1058.58 ± 199.08 (P = .0043). Synovial inflammation, hypervascularization, and osteophyte formations were present in all osteoarthritis knees. Histology scores showed strong correlation with CE-MR imaging findings (Spearman correlation coefficient [SCC] = 0.742; P = .0002) and Q-DSA findings (SCC = 0.763; P < .0001) and weak correlation with DCE-MR imaging (SCC = -0.345; P = .329). Moderate correlation was found between CE-MR imaging and DSA findings (SCC = 0.536; P = .0004). CONCLUSIONS: In this early-stage knee osteoarthritis dog model, synovial inflammation and hypervascularization were found on imaging and confirmed by histology.