A 2.5D Deep Learning-based Approach For Prostate Cancer Detection on T2-weighted Magnetic Resonance Imaging

Ruba Alkadi, Ayman El-Baz, Fatma Taher, Naoufel Werghi; Proceedings of the European Conference on Computer Vision (ECCV) Workshops, 2018, pp. 0-0

In this paper, we propose a fully automatic magnetic resonance image (MRI)-based computer aided diagnosis (CAD) system which simultaneously performs both prostate segmentation and prostate cancer diagnosis. The system utilizes a deep-learning approach to extract high-level features from raw T2-weighted MR volumes. Features are then remapped to the original input to assign a predicted label to each pixel. In the same context, we propose a 2.5D approach which exploits 3D spatial information without a compromise in computational cost. The system is evaluated on a public dataset. Preliminary results demonstrate that our approach outperforms current state-of-the-art in both prostate segmentation and cancer diagnosis.