Automated Classification of Lung Nodules and Precise Segmentation with Convolutional Neural Networks

Abstract

Among the deadliest cancers, lung cancer is very deadly, frequently having fatal results. Medical professionals face complex challenges when using CT scan pictures to assess cancer. As a result, there is an urgent need to create computer-aided systems that may reduce processing time and human labor while simultaneously properly detecting and classifying lung cancer. This research proposal is dedicated to the pivotal task of lung nodule detection, classification into benign or malignant categories in early cancer diagnosis, and subsequent precise segmentation of lung cancer nodules, offering insight into their spatial extent within the CT images based on robust deep learning network. Extensive parameter optimization has been undertaken to craft an efficient and powerful convolutional neural network (CNN) architecture. One of the standout features of this research is its ability to not only accurately classify lung nodules but also precisely delineate their spatial boundaries, calculate their areas, and pinpoint their exact locations within medical images. Our approach leverages a bespoke convolutional neural network (CNN) architecture, meticulously fine-tuned through empirical experimentation, to achieve a remarkable accuracy rate of 99.17%, precision rate of 99.01%, recall rate of 99.75%, and F1-score rate of 99.37%. This surpasses the performance of existing methods in the field and marks a significant breakthrough in lung cancer detection and analysis. The results offer a promising step forward in early diagnosis and treatment planning for lung cancer patients, ultimately contributing to enhanced patient care and outcomes.