Efficiency Analysis of CNN through Different Filters for Medical Image Classification

Image classification is an important aspect in the medical domain, impacting various aspects from diagnosis to treatment. Algorithms for categorizing and classification of image data have been created to assist doctors diagnose cases faster and with fewer manual mistakes. Convolutional Neural Networks (CNN) have become crucial and widely used deep learning tool in medical image classification. Unlike traditional methods that require manual feature engineering, CNNs automatically learn relevant features from images. In this work, the performance and efficiency of CNN was analyzed through different filters like Median filter, Gaussian filter, Adaptive median filter and Wiener filter for medical image classification. The model has been experimented for classification with the publicly available Malaria datasets, pneumonia datasets and blood cell datasets. From the experimented result, it was found that the Wiener filter shows best performance for malaria and pneumonia classification task with accuracy of 96.57% and 96.93% whereas Gaussian filter shows maximum accuracy of 97.79% for Blood cell classification.