DIAGNOSIS OF THE SOLITARY PULMONARY NODULE IMAGED ON CT

University Of California Los Angeles, Los Angeles CA

Investigators

Linked publications & trials

Abstract

DESCRIPTION (Adapted from Applicant's Abstract): The specific aim of this research is to maximize the information derived from a CT imaging exam of patients with solitary pulmonary nodules (SPNs) so that an accurate diagnosis can be reached more quickly and without the need for additional imaging or more invasive tests. The applicants proposed to develop a methodology to provide a more comprehensive examination of the SPN by coupling CT acquisition techniques (including investigating the use of contrast enhanced spiral scans) with image processing to assess multiple properties of the nodule (such as the nodule's size, shape, density, texture and contrast enhancement) and computer-aided diagnosis (CAD) analyses to characterize the nodule based on the value of these properties. This integration of imaging with analytical tools may be able to assist radiologists in characterizing solitary pulmonary nodules imaged on CT. If more accurate and less equivocal tests can be performed, then the need for additional, more invasive tests may be averted. This project seeks to demonstrate the potential effectiveness of the CAD approach to characterizing SPNs based on information extracted from CT image data acquired under specific conditions. Results from these experiments will be used to determine the feasibility of these approaches in characterizing SPNs, prior to developing clinical trials to get larger sample sizes. The specific aims to accomplish this are: 1.) To extract quantitative features from CT images and perform preliminary feature selection to determine which combination of features may best discriminate between benign and malignant nodules; and 2) Utilize Computer Aided Diagnosis (CAD) techniques of image processing and pattern classification to perform a preliminary analysis to predict the diagnosis of solitary pulmonary nodules based on selected image features.