A New Regression Method for Diagnosis of Lung Cancer Disease

Abstract

A radiologist typically diagnoses lung cancer by visually inspecting Computed Tomography (CT) scan images. The procedure is time-consuming, tedious, and prone to errors. Aside from that, variations in intensity in CT scan images, as well as anatomical structure misjudgment by doctors and radiologists, may make identifying cancerous cells difficult. Internationally, doctors and radiologists use the TNM (Tumor, Nodule, Metastases) method to describe the stage of lung cancer. The purpose of this study is to propose an image processing method for detecting Primary Tumour (T) stages of lung cancer by introducing new regression features extraction method for lung cancer in CT scan images. This will aid medical professionals in diagnosing and treating patients. To accomplish this, lung CT scans are processed to isolate. First, lung region with its background then the lesion region and later extract relevant features from the segmented lesion region. The study begins by proposing a new segmentation procedure for lung CT images that can segment lesion and non-lesion. Then a new regression feature of lesion and non-lesion will be extracted. This study's expected outcome is that a new regression feature can help in classifying lung cancer T staging.