Detection of spicules on mammogram based on skeleton analysis

Hidefumi Kobatake, Yukiyasu Yoshinaga

Research output: Contribution to journalArticle

60 Citations (Scopus)

Abstract

Existence of spicules is one of important clues of malignant tumors. This paper presents a new image processing method for t he detection of spicules on mammogram. Spicules can be recognized as line patterns radiating from the center of tumor. To detect such characteristic patterns, line skeletons and a modified Hough transform are proposed. Line skeleton processing is effective in enhancing spinal axes of spicules and in reducing the other skeletons. The modified Hough transform is applied to line skeletons and radiating line structures are obtained. Experiments were made to test the performance of the proposed method. The system was designed using 19 training images, for which one normal case was recognized to be star-shaped. The other case were recognized correctly. Another experiments using 34 test images were also performed. The correct classification rate was 74%. These results shows the effectiveness of the proposed method.

Original languageEnglish
Pages (from-to)235-245
Number of pages11
JournalIEEE Transactions on Medical Imaging
Volume15
Issue number3
DOIs
Publication statusPublished - Dec 1 1996
Externally publishedYes

    Fingerprint

All Science Journal Classification (ASJC) codes

  • Software
  • Radiological and Ultrasound Technology
  • Computer Science Applications
  • Electrical and Electronic Engineering

Cite this