Accurate brain meningioma segmentation and volumetric assessment are critical for serial patient follow-up, surgical planning and monitoring response to treatment. Current gold standard of manual labeling is a time-consuming process, subject to inter-user variability. Fully-automated algorithms for meningioma segmentation have the potential to bring volumetric analysis into clinical and research workflows by increasing accuracy and efficiency, reducing inter-user variability and saving time. Previous research has focused solely on segmentation tasks without assessment of impact and usability of deep learning solutions in clinical practice. Herein, we demonstrate a three-dimensional convolutional neural network (3D-CNN) that performs expert-level, automated meningioma segmentation and volume estimation on MRI scans. A 3D-CNN was initially trained by segmenting entire brain volumes using a dataset of 10,099 healthy brain MRIs. Using transfer learning, the network was then specifically trained on meningioma segmentation using 806 expert-labeled MRIs. The final model achieved a median performance of 88.2% reaching the spectrum of current inter-expert variability (82.6–91.6%). We demonstrate in a simulated clinical scenario that a deep learning approach to meningioma segmentation is feasible, highly accurate and has the potential to improve current clinical practice.

Multiple diverse pathologies result in the clinical presentation of myelopathy. The preferred way to image the spinal cord depends on clinical history, anatomic site of interest, and patient issues limiting certain imaging modalities. This radiology-focused article discusses pertinent physiological considerations, reviews basic and newer imaging techniques, and examines several distinct disease entities in order to highlight the key role of imaging in the work-up of myelopathy.

Background. Accurate brain meningioma detection, segmentation and volumetric assessment are critical for serial patient follow-up, surgical planning and monitoring response to treatment. Current gold standard of manual labeling is a time-consuming process, subject to inter-user variability. Fully-automated algorithms for meningioma detection and segmentation have the potential to bring volumetric analysis into clinical and research workflows by increasing accuracy and efficiency, reducing inter-user variability and saving time. Previous research has focused solely on segmentation tasks without assessment of impact and usability of deep learning solutions in clinical practice. Methods. We developed a three-dimensional convolutional neural network (3D-CNN) to perform expert-level, automated meningioma segmentation and volume estimation on MRI scans. A 3D-CNN was initially trained by segmenting entire brain volumes using a dataset of 10,099 healthy brain MRIs. Using transfer learning, the network was then specifically trained on meningioma segmentation using 806 expert-labeled MRIs. The algorithm tumor-labeling performance was assessed with standard metrics of tumor segmentation performance (i.e., Dice score). To evaluate clinical applicability, we compared volume estimation accuracy and segmentation time based on current practice versus the use of our automated algorithm. Findings. The final model achieved a median performance of 88.2% reaching the spectrum of current inter-expert variability (82.6% - 91.6%). Compared to current workflows, the use of the algorithm reduced processing time by 99% and produced tumor volume calculations with an almost perfect correlation with the expert manual segmentations (r=0.98, p<0.001), significantly more accurate compared to volume estimation techniques used in practice. Conclusions. We demonstrate through a prospective trial conducted in a simulated setting that a deep learning approach to meningioma segmentation is feasible, highly accurate and can substantially improve current clinical practice.