By Topic

Extraction of rounded and line objects for the improvement of medical image pattern recognition

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

7 Author(s)
S. -C. B. Lo ; Dept. of Radiol., Georgetown Univ. Hospital, Washington, DC, USA ; M. Chien ; S. Jong ; H. Li
more authors

In the field of computer-aided diagnosis (CADx), the investigators have encountered various diseases and normal anatomical structure patterns. Two major image patterns that are often targeted for extraction prior to further analyses are rounded and line objects. Here, the authors employed an enhanced Hough transform to extract both objects from the pre-defined image areas. This method can also be applied to the high frequency associated subbands of the wavelet domain where line objects are more distinct. Typically, rounded objects are associated with disease and need to be further analyzed. High intensity line objects are related to normal anatomical structures. Once the line objects are extracted and eliminated, a compensation process must be taken so that the modified pixels are filled by the gray value of the surrounding area. The authors used the ellipse extraction method to search for suspected lung nodules on chest radiographs. The line extraction method was used to detect the edge of ribs which can interfere with the final determination process analyzed by a convolution neural network (CNN). In this experiment, the authors found that the ellipse extraction method performed slightly better than the previous proposed profile matching method. The line removal technique, however, improved the performance of the convolution neural network by 4%. The receiver operating characteristic (ROC) studies indicated that the convolution neural network can achieve a performance of Az=0.90 based on the authors' database when each suspected area was processed by the line removal technique

Published in:

Nuclear Science Symposium and Medical Imaging Conference, 1994., 1994 IEEE Conference Record  (Volume:4 )

Date of Conference:

30 Oct-5 Nov 1994