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 Fukushima, A.
Right arrow Articles by Hayashi, K.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fukushima, A.
Right arrow Articles by Hayashi, 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?

Application of an Artificial Neural Network to High-Resolution CT: Usefulness in Differential Diagnosis of Diffuse Lung Disease

Aya Fukushima1, Kazuto Ashizawa1, Tetsuji Yamaguchi1, Naohiro Matsuyama1, Hideyuki Hayashi1, Isao Kida1, Yoshihiro Imafuku1, Akiko Egawa1, Seigo Kimura1, Kenji Nagaoki1, Sumihisa Honda2, Shigehiko Katsuragawa3, Kunio Doi4 and Kuniaki Hayashi1

1 Department of Radiology and Radiation Oncology, Division of Radiological Science, Nagasaki University Graduate School of Biomedical Sciences, 1-7-1, Sakamoto, Nagasaki 852-8501, Japan.
2 Department of Radiation Epidemiology, Atomic Bomb Disease Institute, Nagasaki University School of Medicine, 1-12-4 Sakamoto, Nagasaki 852-8501, Japan.
3 General Research Center, Nippon Bunri University, Ichiki 1727, Oita 870-0397, Japan.
4 Department of Radiology, Kurt Rossmann Laboratories for Radiologic Image Research, The University of Chicago, 5841 S Maryland Ave., Chicago, IL 60637.



View larger version (42K):

[in a new window]
  		Fig. 1. —Diagram shows basic structure of artificial neural network (ANN). Although only 13 hidden units are shown for illustration, ANN actually consists of 22 hidden units. HRCT = high-resolution CT, BOOP = bronchiolitis obliterans with organizing pneumonia, CMV = cytomegalovirus.

 


View larger version (165K):

[in a new window]
  		Fig. 2. —CT images of 22-year-old man with sarcoidosis. These images can serve as examples of images used in this study. Two radiologists' ratings for this case are shown in Table 2.

 


View larger version (16K):

[in a new window]
  		Fig. 3A. —Artificial neural network (ANN) output obtained by two radiologists' ratings of features for case shown in Figure 2. NSIP = nonspecific interstitial pneumonia, UIP = usual interstitial pneumonia, BOOP = bronchiolitis obliterans with organizing pneumonia, CEP = chronic eosinophilic pneumonia, ca = carcinomatosis, Tb. = tuberculosis, DPB = diffuse panbronchiolitis, PAP = pulmonary alveolar proteinosis, LAM = lymphangiomyomatosis, PCP = Pneumocystis carinii pneumonia, CMV = cytomegalovirus. Graphs show largest output values among 11 diseases correspond to correct diagnosis for observer B, a chest radiologist (A), Chest Imaging and observer G, a general radiologist (B).

 


View larger version (16K):

[in a new window]
  		Fig. 3B. —Artificial neural network (ANN) output obtained by two radiologists' ratings of features for case shown in Figure 2. NSIP = nonspecific interstitial pneumonia, UIP = usual interstitial pneumonia, BOOP = bronchiolitis obliterans with organizing pneumonia, CEP = chronic eosinophilic pneumonia, ca = carcinomatosis, Tb. = tuberculosis, DPB = diffuse panbronchiolitis, PAP = pulmonary alveolar proteinosis, LAM = lymphangiomyomatosis, PCP = Pneumocystis carinii pneumonia, CMV = cytomegalovirus. Graphs show largest output values among 11 diseases correspond to correct diagnosis for observer B, a chest radiologist (A), Chest Imaging and observer G, a general radiologist (B).

 


View larger version (18K):

[in a new window]
  		Fig. 4. —Graph illustrates range (black line) and average (white bar) of area under receiver operating characteristic curve (Az) values of artificial neural network (ANN) performance by eight radiologists for each disease. NSIP = nonspecific interstitial pneumonia, UIP = usual interstitial pneumonia, BOOP = bronchiolitis obliterans with organizing pneumonia, CEP = chronic eosinophilic pneumonia, ca. = carcinomatosis, Tb. = tuberculosis, DPB = diffuse panbronchiolitis, PAP = pulmonary alveolar proteinosis, LAM = lymphangiomyomatosis, PCP = Pneumocystis carinii pneumonia, CMV = cytomegalovirus.

 


View larger version (22K):

[in a new window]
  		Fig. 5A. —Average areas under receiver operating characteristic curves (Az) for observers with and without artificial neural network (ANN) output. Graph shows Az values for chest radiologists.

 


View larger version (21K):

[in a new window]
  		Fig. 5B. —Average areas under receiver operating characteristic curves (Az) for observers with and without artificial neural network (ANN) output. Graph shows Az values for general radiologists. Note that observer performance with artificial neural network (ANN) output improved significantly.

 


View larger version (11K):

[in a new window]
  		Fig. 6. —Graph shows number of correctly diagnosed cases for which observers' rankings changed because of artificial neural network (ANN) output. Black bars indicate number of cases in which ANN output was beneficial; white bars, number of cases ANN in which output was detrimental. ANN output clearly improved performance of observers.

 

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.