Development of a computerized method for identifying view position and orientation for chest radiographs by using a template matching technique

Hidetaka Arimura, Takayuki Ishida, Shigehiko Katsuragawa, Ikuo Kawashita, Kunio Doi

Research output: Contribution to journalArticlepeer-review

2 Citations (Scopus)

Abstract

For implementation of computer-aided diagnostic systems for chest radiographs in the clinical environment, it is necessary to correctly identify two view positions [posteroanterior (PA) and lateral views] and four orientations (upward, downward, leftward, and rightward) for chest radiographs. In picture archiving and communication system (PACS), the information on the view position and orientation for chest radiographs is often not recorded or is recorded incorrectly. The purpose of this study was to develop a computerized method for correctly identifying view position and orientation for chest radiographs by using a template matching technique. Our basic approach is to find the most similar template to an unknown chest image by examining the similarity of the unknown chest image with a number of templates for PA and lateral views at the four image orientations, and to determine simultaneously the view position and orientation for the unknown chest image. All upward templates were produced from the PA or lateral images for upward orientation for various patient sizes. By rotating the upward templates, we produced the templates for downward, leftward, and rightward orientations. To evaluate the similarity of a test image with all templates, the cross-correlation values of the test image with all templates were obtained. The view position and image orientation for the unknown image was considered to be identical to those for the template with which the largest cross-correlation value was obtained. Our results indicated that all cases of 200 PA and 200 lateral views were correctly identified in terms of the view position and image orientation.

Original languageEnglish
Pages (from-to)1047-1054
Number of pages8
JournalNippon Hoshasen Gijutsu Gakkai zasshi
Volume58
Issue number8
DOIs
Publication statusPublished - Jan 1 2002
Externally publishedYes

All Science Journal Classification (ASJC) codes

  • Medicine(all)

Fingerprint

Dive into the research topics of 'Development of a computerized method for identifying view position and orientation for chest radiographs by using a template matching technique'. Together they form a unique fingerprint.

Cite this