Improved pulmonary nodule classification utilizing quantitative lung parenchyma features

Samantha K. N Dilger; Johanna Uthoff; Alexandra Judisch; Emily Hammond; Sarah L Mott; Brian J Smith; John D Newell; Eric A Hoffman; Jessica C Sieren

doi:10.1117/1.JMI.2.4.041004

Back

Improved pulmonary nodule classification utilizing quantitative lung parenchyma features

Journal article

Open access

Peer reviewed

Improved pulmonary nodule classification utilizing quantitative lung parenchyma features

Samantha K. N Dilger, Johanna Uthoff, Alexandra Judisch, Emily Hammond, Sarah L Mott, Brian J Smith, John D Newell, Eric A Hoffman and Jessica C Sieren

Journal of medical imaging (Bellingham, Wash.), Vol.2(4), pp.041004-041004

10/2015

DOI: 10.1117/1.JMI.2.4.041004

PMCID: PMC4748146

PMID: 26870744

Files and links (1)

url

https://doi.org/10.1117/1.JMI.2.4.041004View

Published (Version of record) Open Access

Abstract

Current computer-aided diagnosis (CAD) models for determining pulmonary nodule malignancy characterize nodule shape, density, and border in computed tomography (CT) data. Analyzing the lung parenchyma surrounding the nodule has been minimally explored. We hypothesize that improved nodule classification is achievable by including features quantified from the surrounding lung tissue. To explore this hypothesis, we have developed expanded quantitative CT feature extraction techniques, including volumetric Laws texture energy measures for the parenchyma and nodule, border descriptors using ray-casting and rubber-band straightening, histogram features characterizing densities, and global lung measurements. Using stepwise forward selection and leave-one-case-out cross-validation, a neural network was used for classification. When applied to 50 nodules (22 malignant and 28 benign) from high-resolution CT scans, 52 features (8 nodule, 39 parenchymal, and 5 global) were statistically significant. Nodule-only features yielded an area under the ROC curve of 0.918 (including nodule size) and 0.872 (excluding nodule size). Performance was improved through inclusion of parenchymal (0.938) and global features (0.932). These results show a trend toward increased performance when the parenchyma is included, coupled with the large number of significant parenchymal features that support our hypothesis: the pulmonary parenchyma is influenced differentially by malignant versus benign nodules, assisting CAD-based nodule characterizations.

Special Section on Radiomics and Imaging Genomics

lung cancer

computed tomography

lung parenchyma

cancer screening

Paper

lung nodules

computer-aided diagnosis

texture analysis

Details

Title: Subtitle: Improved pulmonary nodule classification utilizing quantitative lung parenchyma features
Creators: Samantha K. N Dilger - University of Iowa
Johanna Uthoff - University of Iowa
Alexandra Judisch - University of Iowa
Emily Hammond - University of Iowa
Sarah L Mott - University of Iowa
Brian J Smith - University of Iowa
John D Newell - University of Iowa
Eric A Hoffman - University of Iowa
Jessica C Sieren - University of Iowa
Resource Type: Journal article
Publication Details: Journal of medical imaging (Bellingham, Wash.), Vol.2(4), pp.041004-041004
Publisher: Society of Photo-Optical Instrumentation Engineers
DOI: 10.1117/1.JMI.2.4.041004
PMID: 26870744
PMCID: PMC4748146
ISSN: 2329-4302
eISSN: 2329-4310
Grant note: R01 HL-089856; R01 HL-089897 / NIH/NHLBI LCD-220717-N / American Lung Association U01 CA-079778; U01 CA-080098 / NIH
Language: English
Date published: 10/2015
Academic Unit: Roy J. Carver Department of Biomedical Engineering; Radiology; Biostatistics; Holden Comprehensive Cancer Center; Internal Medicine
Record Identifier: 9983997305502771

Metrics

15 Record Views

43 Times Cited - Web of Science