Neural Computing: Application in Non-Invasive Cancer Detection

Abstract

The non-invasive cancer diagnosis represents one of the most practical but difficult challenges of contemporary medicine. Modern imaging methods have been recently developed to aid cancer diagnosis by discriminating between diseased and normal tissues. Neural networks, an information processing paradigm inspired by the way the human brain processes information, can be successfully trained on sample images of tumors, to discriminate between benign or malignant tumors. The aim of this paper is to demonstrate the efficiency of neural networks in the non-invasive diagnosis of malignant diseases, using endoscopic ultrasound elastography sample movies concerning pancreatic tumors. The intended audience consists of statisticians and researchers in Data Mining (including physicians) interested in computer-aided medical applications. An intermediate level of neural computing background is needed to understand the case.