Deep Learning-Based Identification of Lung and Colon Cancer

The prevalence and danger of colon and lung cancer highlight the importance of early and accurate detection to lower mortality risks. The skill of histopathologists is essential for this job because unqualified professionals could endanger patient safety. Deep learning has been increasingly popular recently in the realm of medical image interpretation. This study intends to use and improve pre-trained convolutional neural network (CNN) models for lung and colon cancer detection using histopathology pictures. In particular, the LC25000 dataset is used to train four CNN architectures: VGG16, ResNet50, EfficientNetB3, and EfficientNetB7. The effectiveness of the models is evaluated using precision, recall, score, and accuracy score. The findings show that all four algorithms successfully categorize benign and cancerous pictures with notable accuracy, varying between 97% to 99%.