AJR
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 QUICK SEARCH:   [advanced]


     


This Article
Right arrow Abstract Freely available
Right arrow Full Text
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Li, F.
Right arrow Articles by Doi, K.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Li, F.
Right arrow Articles by Doi, K.
Social Bookmarking
 Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Radiologists' Performance for Differentiating Benign from Malignant Lung Nodules on High-Resolution CT Using Computer-Estimated Likelihood of Malignancy

Feng Li1, Masahito Aoyama2, Junji Shiraishi1, Hiroyuki Abe1, Qiang Li1, Kenji Suzuki1, Roger Engelmann1, Shusuke Sone3, Heber MacMahon1 and Kunio Doi1

1 Department of Radiology, Kurt Rossmann Laboratories for Radiologic Image Research, MC-2026, The University of Chicago, 5841 S Maryland Ave., Chicago, IL 60637.
2 Department of Intelligent Systems, Faculty of Information Sciences, Hiroshima City University, Hiroshima 731-3194, Japan.
3 Azumi General Hospital, Ikeda, Nagano 399-8695, Japan.



View larger version (114K):

[in a new window]
 
Fig. 1A. Radiologists' average ratings without and with computer output for six cases used in observer study. Note that difference in likelihood of malignancy between computer output and initial radiologists' ratings was not large in cases shown here. Radiologists' interpretation with computer-aided diagnosis (CAD) scheme was, in general, more accurate than radiologists without CAD scheme in most malignant and benign nodules. High-resolution CT (HRCT) scan of 55-year-old woman with lung cancer shows pure ground-glass opacity. Computer output was 0.66; radiologists' ratings without CAD, 0.64; and radiologists' ratings with CAD, 0.71.

 


View larger version (120K):

[in a new window]
 
Fig. 1B. Radiologists' average ratings without and with computer output for six cases used in observer study. Note that difference in likelihood of malignancy between computer output and initial radiologists' ratings was not large in cases shown here. Radiologists' interpretation with computer-aided diagnosis (CAD) scheme was, in general, more accurate than radiologists without CAD scheme in most malignant and benign nodules. HRCT scan of 57-year-old woman with benign nodule shows pure ground-glass opacity. Computer output was 0.24; radiologists' ratings without CAD, 0.32; and radiologists' ratings with CAD, 0.27.

 


View larger version (126K):

[in a new window]
 
Fig. 1C. Radiologists' average ratings without and with computer output for six cases used in observer study. Note that difference in likelihood of malignancy between computer output and initial radiologists' ratings was not large in cases shown here. Radiologists' interpretation with computer-aided diagnosis (CAD) scheme was, in general, more accurate than radiologists without CAD scheme in most malignant and benign nodules. HRCT scan of 73-year-old man with lung cancer shows mixed ground-glass opacity. Computer output was 0.90; radiologists' ratings without CAD, 0.75; and radiologists' ratings with CAD, 0.85.

 


View larger version (139K):

[in a new window]
 
Fig. 1D. Radiologists' average ratings without and with computer output for six cases used in observer study. Note that difference in likelihood of malignancy between computer output and initial radiologists' ratings was not large in cases shown here. Radiologists' interpretation with computer-aided diagnosis (CAD) scheme was, in general, more accurate than radiologists without CAD scheme in most malignant and benign nodules. HRCT scan of 79-year-old man with benign nodule shows mixed ground-glass opacity. Computer output was 0.57; radiologists' ratings without CAD, 0.48; and radiologists' ratings with CAD, 0.56.

 


View larger version (155K):

[in a new window]
 
Fig. 1E. Radiologists' average ratings without and with computer output for six cases used in observer study. Note that difference in likelihood of malignancy between computer output and initial radiologists' ratings was not large in cases shown here. Radiologists' interpretation with computer-aided diagnosis (CAD) scheme was, in general, more accurate than radiologists without CAD scheme in most malignant and benign nodules. HRCT scan of 57-year-old man with lung cancer shows solid opacity. Computer output was 0.78; radiologists' ratings without CAD, 0.66; and radiologists' ratings with CAD, 0.76.

 


View larger version (126K):

[in a new window]
 
Fig. 1F. Radiologists' average ratings without and with computer output for six cases used in observer study. Note that difference in likelihood of malignancy between computer output and initial radiologists' ratings was not large in cases shown here. Radiologists' interpretation with computer-aided diagnosis (CAD) scheme was, in general, more accurate than radiologists without CAD scheme in most malignant and benign nodules. HRCT scan of 68-year-old man with benign nodule shows solid opacity. Computer output was 0.36; radiologists' ratings without CAD, 0.37; and radiologists' ratings with CAD, 0.36.

 


View larger version (24K):

[in a new window]
 
Fig. 2. Graph shows receiver operating characteristic (ROC) curves for performance of computer alone and average performance of 16 radiologists without and with computer-aided diagnosis (CAD) scheme. Note that difference was statistically significant between radiologists without and with CAD scheme (p = 0.016), between computer alone and radiologists' performance without (p = 0.00006), and between computer alone and radiologists' performance with CAD scheme (p = 0.0005).

 


View larger version (24K):

[in a new window]
 
Fig. 3A. Graphs show correlation between computer output and average radiologists' ratings without and with computer-aided diagnosis (CAD) scheme for indicating likelihood of malignancy for lung nodules. = average radiologists' ratings for malignant nodules, {circ} = average radiologists' ratings for benign nodules, horizontal lines = range of radiologists' ratings for each nodule. Graphs show correlation between computer outputs and average radiologists' ratings without CAD (A) (r = 0.514) and with CAD (B) (r = 0.784). Note that radiologists' ratings without CAD scheme in some malignant (upper left quadrant) and benign (lower right quadrant) nodules were obviously correct, whereas likelihood of malignancy based on computer outputs alone was incorrect; even with incorrect CAD outputs, radiologists retained correct ratings. One malignant case (arrow) and one benign case (arrowhead) shown here are illustrated in Figures 4A, 4B.

 


View larger version (25K):

[in a new window]
 
Fig. 3B. Graphs show correlation between computer output and average radiologists' ratings without and with computer-aided diagnosis (CAD) scheme for indicating likelihood of malignancy for lung nodules. = average radiologists' ratings for malignant nodules, {circ} = average radiologists' ratings for benign nodules, horizontal lines = range of radiologists' ratings for each nodule. Graphs show correlation between computer outputs and average radiologists' ratings without CAD (A) (r = 0.514) and with CAD (B) (r = 0.784). Note that radiologists' ratings without CAD scheme in some malignant (upper left quadrant) and benign (lower right quadrant) nodules were obviously correct, whereas likelihood of malignancy based on computer outputs alone was incorrect; even with incorrect CAD outputs, radiologists retained correct ratings. One malignant case (arrow) and one benign case (arrowhead) shown here are illustrated in Figures 4A, 4B.

 


View larger version (126K):

[in a new window]
 
Fig. 4A. High-resolution CT (HRCT) scans show one malignant case and one benign case. Note that radiologists' interpretations without computer-aided diagnosis (CAD) scheme were correct in these cases, whereas likelihoods of malignancy based on computer outputs only were obviously incorrect; even with incorrect CAD outputs, radiologists retained correct ratings. HRCT scan shows malignant lung nodule in 68-year-old man. Computer output was 0.36; radiologists' ratings without CAD, 0.67; and radiologists' ratings with CAD, 0.61.

 


View larger version (152K):

[in a new window]
 
Fig. 4B. High-resolution CT (HRCT) scans show one malignant case and one benign case. Note that radiologists' interpretations without computer-aided diagnosis (CAD) scheme were correct in these cases, whereas likelihoods of malignancy based on computer outputs only were obviously incorrect; even with incorrect CAD outputs, radiologists retained correct ratings. HRCT scan shows benign lung nodule in 35-year-old woman. Computer output was 0.78; radiologists' ratings without CAD, 0.27; and radiologists' ratings with CAD, 0.38.

 


View larger version (22K):

[in a new window]
 
Fig. 5. Graph shows correlation (r = 0.925) between likelihood of malignancy and average change in confidence level (rating scores) for each nodule by 16 radiologists. Malignant and benign nodules are marked by black circles and white circles, respectively. = average change in confidence level for malignant nodules, {circ} = average change in confidence level for benign nodules, horizontal lines = range of radiologists' ratings for each nodule. Note that radiologists increased their confidence level when likelihood of malignancy was greater than 0.50 and decreased their confidence level when likelihood was less than 0.50 for most malignant and benign nodules.

 

Add to CiteULike CiteULike   Add to Complore Complore   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us   Add to Digg Digg   Add to Reddit Reddit   Add to Technorati Technorati    What's this?




HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Copyright © 2004 by the American Roentgen Ray Society.