Improving Cancer Classification Through an Deep Learning Framework Using Transfer Learning
D Subitha, Fayyaz Khalid, J C Kavitha, Gulisetty Abhinav · 2025
The primary objective of this work is to create an AI system that offers intelligible and transparent justifications for cancer classification. When it comes to medical imaging-based cancer detection, sophisticated image analysis techniques such as deep learning models are used to assess and interpret medical pictures while offering a transparent account of its reasoning. The performance of the three state-of-art pre-trained deep models: VGG-16, and EfficientNetB0 on a dataset of eight different cancer types obtained from Kaggle is analyzed in this work. Furthermore, a novel deep learning framework is proposed using the above mentioned three pre-trained models as the backbone network using transfer learning approach for each type of cancer. Each model's effectiveness is evaluated in classifying these cancers using metrics like accuracy, precision, recall, and F1-score. This work ultimately yields the best fit model that are tuned with the hyper parameters optimized using Powell's algorithm.