ENHANCING MEDICAL IMAGE CLASSIFICATION THROUGH ENSEMBLE LEARNING WITH DEEP CNN

Abstract

In recent years, deep learning-based image classification techniques have shown promising
results in various medical imaging applications. In this study, we compare the performance of two popular
convolutional neural network (CNN) architectures, Xception-Net and VGG16, for medical image classification
tasks. Specifically, we evaluate their performance on two different datasets: the ISIC dataset for skin lesion
classification and the CMNIST dataset for histology image classification. For the ISIC dataset, we use Xception-
Net to classify different stages of categories. We fine-tune the pre-trained network on a subset of the ISIC
dataset and achieve an accuracy of 91.96%. Our results show that Xception-Net is a powerful deep learning
model for skin lesion classification and has the potential to be used in clinical settings for aiding dermatologists
in the diagnosis of skin cancer. For the CMNIST dataset, we use VGG16 to classify histology images to classify
different stages of categories. We fine-tune the pre-trained network on a subset of the CMNIST dataset and
achieve an accuracy of 99.76%. Our results demonstrate the effectiveness of VGG16 for histology image
classification and suggest its potential for assisting pathologists in the diagnosis of cancer. Overall, our study
highlights the potential of deep learning-based image classification techniques in medical imaging applications
and provides insights into the performance of Xception-Net and VGG16 for different medical image
classification tasks.