Nodule Detection and Prediction of Lung Carcinoma in CT Images: A Relative Study of Enhancement and Segmentation Methods

Abstract

AbstractImage enhancement and segmentation plays an indispensable role in the accurate analysis of affected nodules in lung Computed Tomography (CT) images. Computer-aided detection or diagnosis has become very crucial in the healthcare system for fast detection of lung cancer. The radiologist has a difficult time in correctly identifying the cancerous lung nodules. Because of the vast number of patients, radiologists frequently overlook malignant nodules in imaging. Many recent studies in the field of automated lung nodule diagnosis have revealed significant improvements in radiologist performance. When detecting pulmonary nodules, imaging quality must be taken into account. This has prompted us to investigate the pre-processing stage of lung CT images, which includes a contrast enhancement and segmentation stage. In this paper, different lung nodule enhancement and segmentation methods are compared. The different enhancement methods compared are Histogram Equalization (HE), Contrast Limited Adaptive Histogram Equalization (CLAHE), Image Complement (IC), Gamma Correction (GC) and Balanced Contrast Enhancement Technique (BCET). The five different segmentation methods compared are Adaptive Image Thresholding (AIT), Flood Fill Technique (FFT), Fast Marching Method (FMM), Grayscale Intensity Difference (GSID) and Watershed Segmentation (WS).KeywordsImage enhancementSegmentationCT imagesLung nodule