Diffusion-Weighted MRI Based System for the Early Detection of Prostate Cancer
Ruba H. Alkadi 1, Fatma Taher 1, Naoufel Werghi 1, Ahmed Shalaby 2, and
Ayman Elbaz 2
1. Khalifa University of Science and Technology, Abu Dhabi, UAE
2. University of Louisville, Louisville, KY, USA
2. University of Louisville, Louisville, KY, USA
Abstract—Prostate cancer is the second most diagnosed cancer in men. In this paper, we propose a diffusionweighted MRI based computer-aided detection system for the early detection of prostate cancer. The proposed system calculates seven apparent diffusion coefficients (ADC) for each subject based on the b values at which the scans are acquired. The 3D maps are then represented in a lower dimensional space using a data-driven approach. The reduced maps are fed into seven independent artificial neural networks, each corresponding to the b value at which the ADC maps are calculated. The final decision of malignancy is obtained by aggregating the outputs of all learners in a score-fusion scheme. Essentially, this pipeline is expected to reveal discriminative 3D patterns relevant to subject malignancy. Preliminary results show that the proposed system yields an accuracy of 100% in a leave-onepatient- out cross validation scheme, competing well with state of the art methods.
Index Terms—diffusion weighted MRI, computer-aided detection, prostate cancer, artificial neural networks
