Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans

Observer performance study

Berkman Sahiner, Lubomir M. Hadjiiski, Heang Ping Chan, Jiazheng Shi, Philip N. Cascade, Ella A. Kazerooni, Chuan Zhou, Jun Wei, Aamer R. Chughtai, Chad Poopat, Thomas Song, Jadranka S. Nojkova, Luba Frank, Anil Attili

Research output: Chapter in Book/Report/Conference proceedingConference contribution

17 Citations (Scopus)

Abstract

The purpose of this study was to evaluate the effect of computer-aided diagnosis (CAD) on radiologists' performance for the detection of lung nodules on thoracic CT scans. Our computer system was designed using an independent training set of 94 CT scans in our laboratory. The data set for the observer performance study consisted of 48 CT scans. Twenty scans were collected from patient files at the University of Michigan, and 28 scans by the Lung Imaging Database Consortium (LIDC). All scans were read by multiple experienced thoracic radiologists to determine the true nodule locations, defined as any region identified by one or more expert radiologists as containing a nodule larger than 3 mm in diameter. Eighteen CT examinations were nodule-free, while the remaining 30 CT examinations contained a total of 73 nodules having a median size of 5.5 mm (range 3.0-36.4 mm). Four other study radiologists read the CT scans first without and then with CAD, and provided likelihood of nodule ratings for suspicious regions. Two of the study radiologists were fellowship trained in cardiothoracic radiology, and two were cardiothoracic radiology fellows. Free-response receiver-operating characteristic (FROC) curves were used to compare the two reading conditions. The computer system had a sensitivity of 79% (58/73) with an average of 4.9 marks per normal scan (88/18). Jackknife alternative FROC (JAFROC) analysis indicated that the improvement with CAD was statistically significant (p=0.03).

Original languageEnglish (US)
Title of host publicationMedical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment
Volume6515
DOIs
StatePublished - 2007
Externally publishedYes
EventMedical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment - San Diego, CA, United States
Duration: Feb 21 2007Feb 22 2007

Other

OtherMedical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment
CountryUnited States
CitySan Diego, CA
Period2/21/072/22/07

Fingerprint

Computer aided diagnosis
Computerized tomography
Radiology
Computer systems
Imaging techniques

Keywords

  • Computer-aided diagnosis
  • CT
  • Lung nodules
  • Observer performance evaluation

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Sahiner, B., Hadjiiski, L. M., Chan, H. P., Shi, J., Cascade, P. N., Kazerooni, E. A., ... Attili, A. (2007). Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans: Observer performance study. In Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment (Vol. 6515). [65151D] https://doi.org/10.1117/12.709851

Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans : Observer performance study. / Sahiner, Berkman; Hadjiiski, Lubomir M.; Chan, Heang Ping; Shi, Jiazheng; Cascade, Philip N.; Kazerooni, Ella A.; Zhou, Chuan; Wei, Jun; Chughtai, Aamer R.; Poopat, Chad; Song, Thomas; Nojkova, Jadranka S.; Frank, Luba; Attili, Anil.

Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment. Vol. 6515 2007. 65151D.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sahiner, B, Hadjiiski, LM, Chan, HP, Shi, J, Cascade, PN, Kazerooni, EA, Zhou, C, Wei, J, Chughtai, AR, Poopat, C, Song, T, Nojkova, JS, Frank, L & Attili, A 2007, Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans: Observer performance study. in Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment. vol. 6515, 65151D, Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment, San Diego, CA, United States, 2/21/07. https://doi.org/10.1117/12.709851
Sahiner B, Hadjiiski LM, Chan HP, Shi J, Cascade PN, Kazerooni EA et al. Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans: Observer performance study. In Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment. Vol. 6515. 2007. 65151D https://doi.org/10.1117/12.709851
Sahiner, Berkman ; Hadjiiski, Lubomir M. ; Chan, Heang Ping ; Shi, Jiazheng ; Cascade, Philip N. ; Kazerooni, Ella A. ; Zhou, Chuan ; Wei, Jun ; Chughtai, Aamer R. ; Poopat, Chad ; Song, Thomas ; Nojkova, Jadranka S. ; Frank, Luba ; Attili, Anil. / Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans : Observer performance study. Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment. Vol. 6515 2007.
@inproceedings{d558b97ade844b1d91e4a9e4803c2aa0,
title = "Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans: Observer performance study",
abstract = "The purpose of this study was to evaluate the effect of computer-aided diagnosis (CAD) on radiologists' performance for the detection of lung nodules on thoracic CT scans. Our computer system was designed using an independent training set of 94 CT scans in our laboratory. The data set for the observer performance study consisted of 48 CT scans. Twenty scans were collected from patient files at the University of Michigan, and 28 scans by the Lung Imaging Database Consortium (LIDC). All scans were read by multiple experienced thoracic radiologists to determine the true nodule locations, defined as any region identified by one or more expert radiologists as containing a nodule larger than 3 mm in diameter. Eighteen CT examinations were nodule-free, while the remaining 30 CT examinations contained a total of 73 nodules having a median size of 5.5 mm (range 3.0-36.4 mm). Four other study radiologists read the CT scans first without and then with CAD, and provided likelihood of nodule ratings for suspicious regions. Two of the study radiologists were fellowship trained in cardiothoracic radiology, and two were cardiothoracic radiology fellows. Free-response receiver-operating characteristic (FROC) curves were used to compare the two reading conditions. The computer system had a sensitivity of 79{\%} (58/73) with an average of 4.9 marks per normal scan (88/18). Jackknife alternative FROC (JAFROC) analysis indicated that the improvement with CAD was statistically significant (p=0.03).",
keywords = "Computer-aided diagnosis, CT, Lung nodules, Observer performance evaluation",
author = "Berkman Sahiner and Hadjiiski, {Lubomir M.} and Chan, {Heang Ping} and Jiazheng Shi and Cascade, {Philip N.} and Kazerooni, {Ella A.} and Chuan Zhou and Jun Wei and Chughtai, {Aamer R.} and Chad Poopat and Thomas Song and Nojkova, {Jadranka S.} and Luba Frank and Anil Attili",
year = "2007",
doi = "10.1117/12.709851",
language = "English (US)",
isbn = "0819466336",
volume = "6515",
booktitle = "Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment",

}

TY - GEN

T1 - Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans

T2 - Observer performance study

AU - Sahiner, Berkman

AU - Hadjiiski, Lubomir M.

AU - Chan, Heang Ping

AU - Shi, Jiazheng

AU - Cascade, Philip N.

AU - Kazerooni, Ella A.

AU - Zhou, Chuan

AU - Wei, Jun

AU - Chughtai, Aamer R.

AU - Poopat, Chad

AU - Song, Thomas

AU - Nojkova, Jadranka S.

AU - Frank, Luba

AU - Attili, Anil

PY - 2007

Y1 - 2007

N2 - The purpose of this study was to evaluate the effect of computer-aided diagnosis (CAD) on radiologists' performance for the detection of lung nodules on thoracic CT scans. Our computer system was designed using an independent training set of 94 CT scans in our laboratory. The data set for the observer performance study consisted of 48 CT scans. Twenty scans were collected from patient files at the University of Michigan, and 28 scans by the Lung Imaging Database Consortium (LIDC). All scans were read by multiple experienced thoracic radiologists to determine the true nodule locations, defined as any region identified by one or more expert radiologists as containing a nodule larger than 3 mm in diameter. Eighteen CT examinations were nodule-free, while the remaining 30 CT examinations contained a total of 73 nodules having a median size of 5.5 mm (range 3.0-36.4 mm). Four other study radiologists read the CT scans first without and then with CAD, and provided likelihood of nodule ratings for suspicious regions. Two of the study radiologists were fellowship trained in cardiothoracic radiology, and two were cardiothoracic radiology fellows. Free-response receiver-operating characteristic (FROC) curves were used to compare the two reading conditions. The computer system had a sensitivity of 79% (58/73) with an average of 4.9 marks per normal scan (88/18). Jackknife alternative FROC (JAFROC) analysis indicated that the improvement with CAD was statistically significant (p=0.03).

AB - The purpose of this study was to evaluate the effect of computer-aided diagnosis (CAD) on radiologists' performance for the detection of lung nodules on thoracic CT scans. Our computer system was designed using an independent training set of 94 CT scans in our laboratory. The data set for the observer performance study consisted of 48 CT scans. Twenty scans were collected from patient files at the University of Michigan, and 28 scans by the Lung Imaging Database Consortium (LIDC). All scans were read by multiple experienced thoracic radiologists to determine the true nodule locations, defined as any region identified by one or more expert radiologists as containing a nodule larger than 3 mm in diameter. Eighteen CT examinations were nodule-free, while the remaining 30 CT examinations contained a total of 73 nodules having a median size of 5.5 mm (range 3.0-36.4 mm). Four other study radiologists read the CT scans first without and then with CAD, and provided likelihood of nodule ratings for suspicious regions. Two of the study radiologists were fellowship trained in cardiothoracic radiology, and two were cardiothoracic radiology fellows. Free-response receiver-operating characteristic (FROC) curves were used to compare the two reading conditions. The computer system had a sensitivity of 79% (58/73) with an average of 4.9 marks per normal scan (88/18). Jackknife alternative FROC (JAFROC) analysis indicated that the improvement with CAD was statistically significant (p=0.03).

KW - Computer-aided diagnosis

KW - CT

KW - Lung nodules

KW - Observer performance evaluation

UR - http://www.scopus.com/inward/record.url?scp=35148823647&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=35148823647&partnerID=8YFLogxK

U2 - 10.1117/12.709851

DO - 10.1117/12.709851

M3 - Conference contribution

SN - 0819466336

SN - 9780819466334

VL - 6515

BT - Medical Imaging 2007: Image Perception, Observer Performance, and Technology Assessment

ER -