An Efficient Convolutional Neural Network for Classification of Multi-Class Colorectal Tissue Using Histopathological Images
Anurodh Kumar, Amit Vishwakarma, Varun Bajaj · 2022 IEEE 6th Conference on Information and Communication Technology (CICT) · 2022
Colorectal cancer (CRC) has a high fatality rate that continuously influences human lives worldwide. Early diagnosis of CRC prolongs human life and helps to prevent the disease. Histopathological inspection is routinely utilized to diagnose CRC. Manual inspection of histological diagnosis needs more assessment time, and the decision is based on clinicians' subjective opinions. In this work, a less complex, lightweight convolutional neural network (CNN)-based model is proposed for the automatic classification of multiclass colorectal tissue utilizing histopatho-logical images. Histopathological images are denoised using a median filter. The obtained denoised histopathological images are fed as input to the developed network and two pre-trained models. The proposed CNN achieved a classification accuracy of 93.50%. Compared to Xception and ResNet50 the proposed CNN is lightweight as it requires lesser learnable parameters, which makes it computationally fast. The effectiveness of the developed model is compared with the other existing methods to classify colorectal tissue on a similar dataset.