Abstract
Histopathology is the study of cellular structures, tissues and their abnormalities to diagnose a wide range of diseases, with a primary focus on cancer. The recent innovations and advancements in image analysis techniques and machine learning enable the histopathologists to automate the process of detection and classification of diseases observed in histopathology images. Traditional visual analysis by pathologists, though skilled, is slow and prone to inconsistencies. By utilizing advanced techniques, such as Convolutional Neural Networks, this project aims to revolutionize disease classification and management in histopathology. Researchers are now using convolutional neural networks and other algorithms to accurately segment tissues, extract key features, and even predict cancer diagnosis and treatment response. These automated methods hold immense potential for faster, more precise cancer diagnosis and personalized care. The proposed model LuCoNet is a Convolution Neural Network Architecture that uses the publicly available dataset comprises 25,000 histopathological JPEG images, initially sourced from HIPAAcompliant datasets. It includes 750 lung tissue images and 500 colon tissue images, augmented to expand the dataset. This study underscores the transformative potential of deep learning in histopathology image analysis, promising enhanced diagnostic accuracy and personalized treatment strategies. The performance of LuCoNet was compared with other models evaluated in the literature survey and LuCoNet performed extremely well in prediction with 98.5%, 0.986, 0.988, and 0.984 for accuracy, Precision, Recall and F1-Score measures.
Keywords: Cancer Tissues, Colon Cancer, Convolutional Neural Network, Deep Learning, Histopathology, Lung Cancer