Metadata-independent classification of MRI sequences using convolutional neural networks: Successful application to prostate MRI.

PURPOSE: The ever-increasing volume of medical imaging data and interest in Big Data research brings challenges to data organization, categorization, and retrieval. Although the radiological value chain is almost entirely digital, data structuring has been widely performed pragmatically, but with insufficient naming and metadata standards for the stringent needs of image analysis. To enable automated data management independent of naming and metadata, this study focused on developing a convolutional neural network (CNN) that classifies medical images based solely on voxel data. METHOD: A 3D CNN (3D-ResNet18) was trained using a dataset of 31,602 prostate MRI volumes with 10 different sequence types of 1243 patients. A five-fold cross-validation approach with patient-based splits was chosen for training and testing. Training was repeated with a gradual reduction in training data assessing classification accuracies to determine the minimum training data required for sufficient performance. The trained model and developed method were tested on three external datasets. RESULTS: The model achieved an overall accuracy of 99.88 % ± 0.13 % in classifying typical prostate MRI sequence types. When being trained with approximately 10 % of the original cohort (112 patients), the CNN still achieved an accuracy of 97.43 % ± 2.10 %. In external testing the model achieved sensitivities of > 90 % for 10/15 tested sequence types. CONCLUSIONS: The herein developed CNN enabled automatic and reliable sequence identification in prostate MRI. Ultimately, such CNN models for voxel-based sequence identification could substantially enhance the management of medical imaging data, improve workflow efficiency and data quality, and allow for robust clinical AI workflows.

External validation of the diagnostic value of perihematomal edema characteristics in neoplastic and non-neoplastic intracerebral hemorrhage.

BACKGROUND AND PURPOSE: Neoplastic intracerebral hemorrhage (ICH) may be incorrectly identified as non-neoplastic ICH on imaging. Relative perihematomal edema (relPHE) on computed tomography (CT) has been proposed as a marker to discriminate neoplastic from non-neoplastic ICH but has not been externally validated. The purpose of this study was to evaluate the discriminatory power of relPHE in an independent cohort. METHODS: A total of 291 patients with acute ICH on CT and follow-up magnetic resonance imaging (MRI) were included in this single-center retrospective study. ICH subjects were dichotomized into non-neoplastic or neoplastic ICH based on the diagnosis on the follow-up MRI. ICH and PHE volumes and density values were derived from semi-manually segmented CT scans. Calculated PHE characteristics for discriminating neoplastic ICH were evaluated using receiver-operating characteristic (ROC) curves. ROC curve-associated cut-offs were calculated and compared between the initial and the validation cohort. RESULTS: A total of 116 patients (39.86%) with neoplastic ICH and 175 (60.14%) with non-neoplastic ICH were included. Median PHE volumes, relPHE, and relPHE adjusted for hematoma density were significantly higher in subjects with neoplastic ICH (all p values <0.001). ROC curves for relPHE had an area under the curve (AUC) of 0.72 (95% confidence interval [CI] 0.66-0.78) and an AUC of 0.81 (95% CI 0.76-0.87) for adjusted relPHE. The cut-offs were identical in the two cohorts, with >0.70 for relPHE and >0.01 for adjusted relPHE. CONCLUSIONS: Relative perihematomal edema and adjusted relPHE accurately discriminated neoplastic from non-neoplastic ICH on CT imaging in an external patient cohort. These results confirmed the findings of the initial study and may improve clinical decision making.