An Efficient Model for Optimizing Hyperparameters in AlexNet for Precise Malignancy Detection in Lung and Colon Histopathology Images with CSIP-EHE

Cancer, a lethal disease stemming from genetic anomalies and biochemical irregularities, presents a major global health challenge, with lung and colon cancers being significant contributors to morbidity and mortality. Timely and precise cancer detection is crucial for optimal treatment decisions, and machine learning and deep learning techniques offer a promising solution for expediting this process. In this research, a pre-trained neural network, specifically AlexNet, was fine-tuned with modifications to four layers to adapt it to a dataset comprising histopathological images of lung and colon tissues. Additionally, a Bayesian optimization approach was employed for hyperparameter tuning in Convolutional Neural Networks (CNNs) to enhance recognition accuracy while maintaining computational efficiency. The research utilized a comprehensive dataset divided into five classes, and in cases of suboptimal results, a Counteracting Suboptimal Image Processing (CSIP) strategy was applied, focusing on improving images of underperforming classes to reduce processing time and effort.