Title: CAD for demarcation of malignant and benign nodules in CT lung images of spiculated nodules

Authors: S. Saravanan; G. Selvakumar; C. Amarnath; S. Udayabaskaran; S. Manikandan

Addresses: Sathyabama University, Chennai, India ' Muthayammal Engineering College, Rasipuram 637408, India ' Stanley Medical College, Chennai, India ' Department of Mathematics, Veltech University, Chennai, India ' Underblu Technologies, Chennai, India

Abstract: This research work intents to remove the intricacies involved in demarcating the malignant and benign of the spiculated Solitary Pulmonary Nodules (SPN). Edges can be classified as irregular edge with corona radiata, lobulation, notching signs and a distinct soft, uncloudy contour edge. These edges are hardly spotted in bronchial carcinoma. This paper develops an algorithm for automatically detecting stipulated nodules using BPN algorithm, from the given computed tomography (CT) lung image. Here, to automate the detection of lung nodule, parametric active contours are used for manual segmentation. Features are extracted from gray level co-occurrence matrix (GLCM) derived from manually segmented lung nodule and used for further classification as nodule and non-nodule/normal image. This paper further classifies spiculated nodule into malignant or benign by fixing a threshold for the average image intensity after administering contrast.

Keywords: corona radiate; lobulation and notching signs; spiculated nodule; malignant; benign; GLCM texture features; contrast enhancement; active contour.

DOI: 10.1504/IJBET.2017.083815

International Journal of Biomedical Engineering and Technology, 2017 Vol.24 No.1, pp.33 - 52

Received: 11 Jan 2016
Accepted: 11 Apr 2016

Published online: 24 Apr 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article