Paper Key : IRJ************688
Author: Sindhu
Date Published: 03 Mar 2025
Abstract
Lung nodule detection is a critical step in the early diagnosis and treatment of lung cancer, requiring reliable and efficient methods for analyzing CT scan images. This paper aims to present a comparative evaluation of two deep learning models: a Simple Convolutional Neural Network (CNN) and a VGG-like architecture.The Simple CNN, designed with fewer layers, focuses on computational efficiency and faster training, while the VGG-like architecture employs deeper convolutional blocks to extract more complex features, potentially improving classification accuracy.This project implements a VGG-like convolutional neural network (CNN) architecture using MATLAB for the detection of lung nodules from CT scan images. The VGG-like architecture leverages multiple convolutional layers, ReLU activations, and max-pooling layers for hierarchical feature extraction. Fully connected layers and a softmax classifier are employed for final classification. Both models were evaluated on a IQ-OTHNCCD lung dataset, where the simple CNN achieved an accuracy of 81%, and the VGG-like architecture outperformed it with an accuracy of 98%. These results suggest that while both models show promising performance, the deeper VGG-like architecture provides superior accuracy, making it a more effective choice for lung nodule detection in CT scan images.
DOI Requested