Algorithm 1: Preprocessing step of developing input data for a convolution neural network (CNN). Value of N is 100, 150, or 200.

Input: Prostate biopsy specimens digitized.

Output: Classified selected patches into Gleason pattern labels. 1.Apply the histogram equalization and edge enhancement on the PBSs. 2.Divide the PBSs into overlapping patches, with size N × N pixels and 75% overlapping. 3. Selecting appropriate patches for training Calculate the majority voting for the pathological patch, WL ← Winning Label. Estimate two variables for corresponding label path, RB←ratio of Background and CV ←Center value. If $(WL==CV)\&(RB\le 0.3$)  Select the patch Else  Remove the patch