Combination of computer extracted shape and texture features enables discrimination of granulomas from adenocarcinoma on chest computed tomography

Mahdi Orooji, Mehdi Alilou, Sagar Rakshit, Niha Beig, Mohammad Hadi Khorrami, Prabhakar Rajiah, Rajat Thawani, Jennifer Ginsberg, Christopher Donatelli, Michael Yang, Frank Jacono, Robert Gilkeson, Vamsidhar Velcheti, Philip Linden, Anant Madabhushi

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

Differentiation between benign and malignant nodules is a problem encountered by radiologists when visualizing computed tomography (CT) scans. Adenocarcinomas and granulomas have a characteristic spiculated appearance and may be fluorodeoxyglucose avid, making them difficult to distinguish for human readers. In this retrospective study, we aimed to evaluate whether a combination of radiomic texture and shape features from noncontrast CT scans can enable discrimination between granulomas and adenocarcinomas. Our study is composed of CT scans of 195 patients from two institutions, one cohort for training (N = 139) and the other (N = 56) for independent validation. A set of 645 three-dimensional texture and 24 shape features were extracted from CT scans in the training cohort. Feature selection was employed to identify the most informative features using this set. The top ranked features were also assessed in terms of their stability and reproducibility across the training and testing cohorts and between scans of different slice thickness. Three different classifiers were constructed using the top ranked features identified from the training set. These classifiers were then validated on the test set and the best classifier (support vector machine) yielded an area under the receiver operating characteristic curve of 77.8%.

Original languageEnglish (US)
Article number024501
JournalJournal of Medical Imaging
Volume5
Issue number2
DOIs
StatePublished - Apr 1 2018

Fingerprint

Granuloma
Adenocarcinoma
Thorax
Tomography
ROC Curve
Retrospective Studies

Keywords

  • artificial intelligence
  • computed tomography scan
  • computer-Assisted diagnosis
  • lung cancer
  • phenotype.

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Combination of computer extracted shape and texture features enables discrimination of granulomas from adenocarcinoma on chest computed tomography. / Orooji, Mahdi; Alilou, Mehdi; Rakshit, Sagar; Beig, Niha; Khorrami, Mohammad Hadi; Rajiah, Prabhakar; Thawani, Rajat; Ginsberg, Jennifer; Donatelli, Christopher; Yang, Michael; Jacono, Frank; Gilkeson, Robert; Velcheti, Vamsidhar; Linden, Philip; Madabhushi, Anant.

In: Journal of Medical Imaging, Vol. 5, No. 2, 024501, 01.04.2018.

Research output: Contribution to journalArticle

Orooji, M, Alilou, M, Rakshit, S, Beig, N, Khorrami, MH, Rajiah, P, Thawani, R, Ginsberg, J, Donatelli, C, Yang, M, Jacono, F, Gilkeson, R, Velcheti, V, Linden, P & Madabhushi, A 2018, 'Combination of computer extracted shape and texture features enables discrimination of granulomas from adenocarcinoma on chest computed tomography', Journal of Medical Imaging, vol. 5, no. 2, 024501. https://doi.org/10.1117/1.JMI.5.2.024501
Orooji, Mahdi ; Alilou, Mehdi ; Rakshit, Sagar ; Beig, Niha ; Khorrami, Mohammad Hadi ; Rajiah, Prabhakar ; Thawani, Rajat ; Ginsberg, Jennifer ; Donatelli, Christopher ; Yang, Michael ; Jacono, Frank ; Gilkeson, Robert ; Velcheti, Vamsidhar ; Linden, Philip ; Madabhushi, Anant. / Combination of computer extracted shape and texture features enables discrimination of granulomas from adenocarcinoma on chest computed tomography. In: Journal of Medical Imaging. 2018 ; Vol. 5, No. 2.
@article{023cb6c303dd49d6a44ab869f2a52ff7,
title = "Combination of computer extracted shape and texture features enables discrimination of granulomas from adenocarcinoma on chest computed tomography",
abstract = "Differentiation between benign and malignant nodules is a problem encountered by radiologists when visualizing computed tomography (CT) scans. Adenocarcinomas and granulomas have a characteristic spiculated appearance and may be fluorodeoxyglucose avid, making them difficult to distinguish for human readers. In this retrospective study, we aimed to evaluate whether a combination of radiomic texture and shape features from noncontrast CT scans can enable discrimination between granulomas and adenocarcinomas. Our study is composed of CT scans of 195 patients from two institutions, one cohort for training (N = 139) and the other (N = 56) for independent validation. A set of 645 three-dimensional texture and 24 shape features were extracted from CT scans in the training cohort. Feature selection was employed to identify the most informative features using this set. The top ranked features were also assessed in terms of their stability and reproducibility across the training and testing cohorts and between scans of different slice thickness. Three different classifiers were constructed using the top ranked features identified from the training set. These classifiers were then validated on the test set and the best classifier (support vector machine) yielded an area under the receiver operating characteristic curve of 77.8{\%}.",
keywords = "artificial intelligence, computed tomography scan, computer-Assisted diagnosis, lung cancer, phenotype.",
author = "Mahdi Orooji and Mehdi Alilou and Sagar Rakshit and Niha Beig and Khorrami, {Mohammad Hadi} and Prabhakar Rajiah and Rajat Thawani and Jennifer Ginsberg and Christopher Donatelli and Michael Yang and Frank Jacono and Robert Gilkeson and Vamsidhar Velcheti and Philip Linden and Anant Madabhushi",
year = "2018",
month = "4",
day = "1",
doi = "10.1117/1.JMI.5.2.024501",
language = "English (US)",
volume = "5",
journal = "Journal of Medical Imaging",
issn = "0720-048X",
publisher = "Elsevier Ireland Ltd",
number = "2",

}

TY - JOUR

T1 - Combination of computer extracted shape and texture features enables discrimination of granulomas from adenocarcinoma on chest computed tomography

AU - Orooji, Mahdi

AU - Alilou, Mehdi

AU - Rakshit, Sagar

AU - Beig, Niha

AU - Khorrami, Mohammad Hadi

AU - Rajiah, Prabhakar

AU - Thawani, Rajat

AU - Ginsberg, Jennifer

AU - Donatelli, Christopher

AU - Yang, Michael

AU - Jacono, Frank

AU - Gilkeson, Robert

AU - Velcheti, Vamsidhar

AU - Linden, Philip

AU - Madabhushi, Anant

PY - 2018/4/1

Y1 - 2018/4/1

N2 - Differentiation between benign and malignant nodules is a problem encountered by radiologists when visualizing computed tomography (CT) scans. Adenocarcinomas and granulomas have a characteristic spiculated appearance and may be fluorodeoxyglucose avid, making them difficult to distinguish for human readers. In this retrospective study, we aimed to evaluate whether a combination of radiomic texture and shape features from noncontrast CT scans can enable discrimination between granulomas and adenocarcinomas. Our study is composed of CT scans of 195 patients from two institutions, one cohort for training (N = 139) and the other (N = 56) for independent validation. A set of 645 three-dimensional texture and 24 shape features were extracted from CT scans in the training cohort. Feature selection was employed to identify the most informative features using this set. The top ranked features were also assessed in terms of their stability and reproducibility across the training and testing cohorts and between scans of different slice thickness. Three different classifiers were constructed using the top ranked features identified from the training set. These classifiers were then validated on the test set and the best classifier (support vector machine) yielded an area under the receiver operating characteristic curve of 77.8%.

AB - Differentiation between benign and malignant nodules is a problem encountered by radiologists when visualizing computed tomography (CT) scans. Adenocarcinomas and granulomas have a characteristic spiculated appearance and may be fluorodeoxyglucose avid, making them difficult to distinguish for human readers. In this retrospective study, we aimed to evaluate whether a combination of radiomic texture and shape features from noncontrast CT scans can enable discrimination between granulomas and adenocarcinomas. Our study is composed of CT scans of 195 patients from two institutions, one cohort for training (N = 139) and the other (N = 56) for independent validation. A set of 645 three-dimensional texture and 24 shape features were extracted from CT scans in the training cohort. Feature selection was employed to identify the most informative features using this set. The top ranked features were also assessed in terms of their stability and reproducibility across the training and testing cohorts and between scans of different slice thickness. Three different classifiers were constructed using the top ranked features identified from the training set. These classifiers were then validated on the test set and the best classifier (support vector machine) yielded an area under the receiver operating characteristic curve of 77.8%.

KW - artificial intelligence

KW - computed tomography scan

KW - computer-Assisted diagnosis

KW - lung cancer

KW - phenotype.

UR - http://www.scopus.com/inward/record.url?scp=85045845665&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85045845665&partnerID=8YFLogxK

U2 - 10.1117/1.JMI.5.2.024501

DO - 10.1117/1.JMI.5.2.024501

M3 - Article

VL - 5

JO - Journal of Medical Imaging

JF - Journal of Medical Imaging

SN - 0720-048X

IS - 2

M1 - 024501

ER -