Unsupervised Brain Tissue Segmentation by Using Bias Correction Fuzzy C-Means and Class-Adaptive Hidden Markov Random Field Modelling

Ziming Zeng, Chunlei Han, Liping Wang, Reyer Zwiggelaar

Research output: Chapter in Book/Report/Conference proceedingConference Proceeding (Non-Journal item)

5 Citations (Scopus)

Abstract

Unsupervised brain tissue segmentation in magnetic resonance imaging (MRI) is a key step in brain analysis, such as computer-aided surgery, clinical diagnosis, pathological analysis, surgical planning. Due to the noise and bias field in MRI, it is difficult to automatically segment brain tissue. In order to improve the segmentation accuracy, we propose an unsupervised method which combines an improved bias correction Fuzzy C-means (BCFCM) and class-adaptive hidden Markov random field Modelling (HMRF). The BCFCM segmentation result is used as the initial labelling for class-adaptive HMRF, which is utilized to refine the segmentation results. Experiments are evaluated on simulated MR images. Comparing with the ground truth, the results show that the proposed method can perform well on MR brain images with noisy MRI and bias field.
Original languageEnglish
Title of host publicationFrontier and Future Development of Information Technology in Medicine and Education, ITME 2013
Subtitle of host publicationITME 2013
EditorsShaozi Li, Qun Jin, Xiaohong Jiang, James J. (Jong Hyuk) Park
PublisherSpringer Nature
Pages579-587
Number of pages9
ISBN (Electronic)978-94-007-7618-0, 9400776187
ISBN (Print)9789400776173
DOIs
Publication statusPublished - 06 Dec 2013

Publication series

NameLecture Notes in Electrical Engineering
Volume269 LNEE
ISSN (Print)1876-1100
ISSN (Electronic)1876-1119

Keywords

  • Bias field
  • Brain
  • MRI
  • Noise
  • Segmentation

Fingerprint

Dive into the research topics of 'Unsupervised Brain Tissue Segmentation by Using Bias Correction Fuzzy C-Means and Class-Adaptive Hidden Markov Random Field Modelling'. Together they form a unique fingerprint.

Cite this