CANCER DETECTION IN HISTOPATHOLOGY OF LYMPH NODES

Abstract: Cancer is one of the main diseases around the world. Many cancer patients die every year due to late diagnosis and treatment. Hence, lately, early cancer recognition frameworks dependent on histopathology slides are sought after. With the advancements in technology, deep learning has rooted its applications in various domains. In this project, a comparison between pre-trained models with customized models is performed, which adapts the CNN model and ResNet for feature extraction to the problem at hand. The convolutional neural network which is known as profound learning architecture has accomplished amazing outcomes in numerous applications. Working on histopathology images requires high tuning while pre-processing images to obtain high accuracy while classifying. While using CNN, due to the increase in parameters the model might get highly complex. Hence, we experiment using ResNet to overcome these problems and evaluate the models. Broad examinations on a freely accessible histopathologic cancer dataset of lymph nodes are completed and the precision scores are determined for execution assessment on the two models.

Keywords: CNN, ResNet, Histopathology, Cancer.