Automated segmentation of white matter lesions using global neighbourhood given contrast feature-based random forest and markov random field

Pallab Kanti Roy, Alauddin Bhuiyan, Andrew Janke, Patricia M Desmond, Tien Yin Wong, Elsdon Storey, Walter Abhayaratna, Kotagiri Ramamohanarao

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

2 Citations (Scopus)

Abstract

Recent studies show that, cerebral White MatterLesion (WML) is related to cerebrovascular diseases,cardiovascular diseases, dementia and psychiatric disorders.Manual segmentation of WML is not appropriate for long termlongitudinal studies because it is time consuming and it showshigh intra- and inter-rater variability. In this paper, a fullyautomated segmentation method is utilized to segment WMLfrom brain Magnetic Resonance Imaging (MRI). The segmentationmethod uses a combination of global neighbourhoodgiven contrast feature-based Random Forest (RF) classifier andMarkov Random Field (MRF) to segment WML. To removefalse positive lesions we use a rule based morphological postprocessingoperation. Quantitative evaluation of the proposedmethod was performed on 24 subjects of ENVIS-ion study.The segmentation results were validated against the manualsegmentation, performed by an experienced radiologist andwere compared to a recenlty published WML segmentationmethod. The results show a dice similarity index of 0.75 forhigh lesion load, 0.71 for medium lesion load and 0.60 for lowlesion load.

Original languageEnglish
Title of host publicationProceedings - 2014 IEEE International Conference on Healthcare Informatics, ICHI 2014
Place of PublicationPiscataway NJ USA
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages1-6
Number of pages6
ISBN (Electronic)9781479957019
DOIs
Publication statusPublished - 2 Mar 2014
EventIEEE International Conference on Healthcare Informatics (ICHI 2014) - Verona, Italy
Duration: 15 Sep 201417 Sep 2014
Conference number: 2nd
http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=7051630

Conference

ConferenceIEEE International Conference on Healthcare Informatics (ICHI 2014)
Abbreviated titleICHI 2014
CountryItaly
CityVerona
Period15/09/1417/09/14
OtherICHI 2014 was co-located with TIME 2014 (21st International Symposium on Temporal Representation and Reasoning), September 8 -10, 2014, and GandALF 2014 (the 5th International Symposium on Games, Automata, Logics and Formal Verification), September 10 – 12, 2014. We hope that this co-location could be a further occasion for improving cross-fertilization among different areas of
computer science with respect to the important application field of Healthcare.
Internet address

Keywords

  • magnetic resonance imaging (MRI)
  • markov random field (MRF)
  • random forest (RF)
  • white matter lesion

Cite this

Roy, P. K., Bhuiyan, A., Janke, A., Desmond, P. M., Wong, T. Y., Storey, E., Abhayaratna, W., & Ramamohanarao, K. (2014). Automated segmentation of white matter lesions using global neighbourhood given contrast feature-based random forest and markov random field. In Proceedings - 2014 IEEE International Conference on Healthcare Informatics, ICHI 2014 (pp. 1-6). [7052463] IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/ICHI.2014.75