Invasive Ductal Carcinoma Classification from Whole Slide Image Based on BRISQUE and Convolutional Neural Networks.

This research aims to improve the performance of an invasive ductal carcinoma (IDC) classification system by examining how input image quality affects the dataset. To achieve this goal, we utilized the Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE) method to assess image quality and divided the dataset into subsets for training, cross-validation, and testing of the proposed Convolutional Neural Network (CNN) architecture. The study evaluated various subsets of the training dataset to determine the optimal option, which was then used to identify the appropriate training parameters and evaluate the system's performance. The results demonstrated that the proposed approaches outperformed the standard methods for IDC classification.