Pulse-Coupled Neural Networks applied to Human Brain Image Processing

Several experiments were carried out applying a set of Pulse-Couple Neural Network variants. In particular, we realized and proposed three modifications to the Intersecting Cortical Model (ICM) Neural Network paradigm in order to measure how effective it becomes for edge detection on human brain images. The human brain images were obtained using Magnetic Resonance and Positron Emission Tomography. We compared the ICM outputs versus the outputs obtained from two well-known computer vision algorithms: Canny and Sobel. We observed that the modifications proposed to ICM produced better edge detection than the original paradigm. We include all the ICM variants details, the experiments, the evaluation criteria and the final results of the medical images edge detection recognition.