Semi-Automated Ranking Model For Feature Extraction And Classification With Detection Of Tumor Size In Lung Images

Abstract

Lung cancer emerges as a malignancy originating in the cells of the lungs, commonly within the epithelial cells that line the air passages. Globally prevalent and notorious for its high fatality rates, lung cancer is strongly associated with smoking as a primary risk factor. Nevertheless, individuals who do not smoke can also succumb to lung cancer, influenced by factors like exposure to environmental pollutants or genetic predisposition. The early stages of lung cancer often progress without noticeable symptoms, leading to delayed diagnoses and subsequently restricting the available treatment options.This paper presents an innovative approach utilizing the Directional Clustering Ranking Semi-Automated Classification (DCRSA-C) model for lung tumor detection and classification in medical imaging. Leveraging advanced machine learning techniques, the DCRSA-C model demonstrates a high level of accuracy, sensitivity, and specificity in distinguishing between benign and malignant tumors. Additionally, the model exhibits proficiency in size estimation, as evidenced by a commendable Intersection over Union (IoU) score. The study carefully examines the model's performance across diverse datasets, considering the variability in imaging conditions, patient demographics, and class imbalances. While celebrating the promising results, the paper also addresses the need for further validation and explores avenues for improving interpretability and seamless integration into clinical workflows. This work contributes to the evolving landscape of artificial intelligence in healthcare, offering a potential transformative tool for accurate and efficient lung cancer diagnosis with implications for improved patient care.