Prostate texture features' statistical analysis using TRUS images

S. Mohamed, M. Salama

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

Abstract

This paper focuses mainly on the prostate TRUS images features' statistical analysis. Different texture features are constructed from the prostate TRUS images for cancerous as well as con-cancerous regions. The constructed features are then statistically analyzed to determine which features best represent the regions' texture. Moreover, the autocorrelation as well as the Mutual Information (MI) among the constructed features and between the each feature and class are also calculated. The obtained results show that the constructed features are very highly correlated and if used without feature selection will cause curse of dimensionality and may confuse the classifier. Therefore feature selection is utilized and the selected features are then used to classify the regions. The classification accuracy achieved using the Support Vector Machines classifier is 94%.

Original languageBritish English
Title of host publicationProceedings of the 2007 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2007
Pages608-614
Number of pages7
StatePublished - 2007
Event2007 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2007 - Las Vegas, NV, United States
Duration: 25 Jun 200728 Jun 2007

Publication series

NameProceedings of the 2007 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2007

Conference

Conference2007 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2007
Country/TerritoryUnited States
CityLas Vegas, NV
Period25/06/0728/06/07

Keywords

  • Cancer
  • Prostate
  • Statistical analysis
  • Statistical features

Fingerprint

Dive into the research topics of 'Prostate texture features' statistical analysis using TRUS images'. Together they form a unique fingerprint.

Cite this