A modified convolutional neural network framework for categorizing lung cell histopathological image based on residual network

Abstract

Computational technology has drastically improved the medical and Healthcare fields in the last few years. Many new diagnostic techniques are introduced with the help of computer technology. Cancer causes most death worldwide. Lung cancer is the most common and gender-independent cancer among other types of cancer. The timely detection of cancer increases survival for the patient. Cancer may be automatically diagnosed using the power of Machine Learning (ML) with medical images. ML can classify cancer cell images more accurately with less time and lower cost. This research modifies the Convolutional Neural Network (CNN) model as pre-trained Visual Geometry Group19 (VGG19) for classifying lung cancer biopsy images with improved augmentation technique. The results observed with fine-tuned VGG19 model with improved augmentation technique are up to 97.73% accuracy. The experiment results by the proposed method show better performance than existing methods.

Keywords

• Machine learning
• Convolutional neural network
• Histopathological images
• Improved augmentation
• Visual geometry group19