Evaluating CNN variants with Transfer Learning for Multi-Class NSCLC Diagnosis

Lung Cancer is the disease with serious concern, Non-small cell lung cancer (NSCLC) is a sub type of it, with most cases that has to be timely intervened for accurate diagnosis and also to improve patient outcomes. With recent developments in deep learning, particularly in convolutional neural networks, it leads to increased potential for accurate and automated medical image diagnosis. This study assesses the performance of various CNN variants with transfer learning for multi-class NSCLC subtype image classification using CT images. Results suggest that integrating CNN with transfer learning provides a robust approach for classification of NSCLC subtypes. Out of all the benchmark models LungNetB5 outperformed by 90% accuracy highlighting its potentiality in clinical decision making.