Publication detail

Multiparametric Biological Tissue Analysis: A Survey of Image Processing Tools

MIKULKA, J.

Original Title

Multiparametric Biological Tissue Analysis: A Survey of Image Processing Tools

Type

conference paper

Language

English

Original Abstract

Using magnetic resonance tomography to scan biological tissues is currently a very dynamic approach. Based on various image parameters, the method enables us to analyze tissue properties, recognize healthy and pathological tissues, and diagnose the disease or indicate its progression. These activities are then necessarily accompanied by the processing of the acquired images. The paper introduces a comparison of statistical tools for the trainable segmentation of multiparametric data obtained through magnetic resonance tomography. In this context, the author briefly compares various available tools (Weka, Slicer3D, and RapidMiner) in view of the input data training and testing, applicability of the classification models, and ability of the input/output data to be extended with other systems for further processing. The paper also describes as a multiparametric task the segmentation of a brain tumor performed with real MR data. The source of the data consists in T1 and T2-weighted images. The proposed segmentation method is carried out within the following phases: data resampling; spatial data coregistration; definition of the training points; training of the SVM classification model; testing of the model and interpretation of the classification results.

Keywords

SVM, image segmentation, data mining

Authors

MIKULKA, J.

RIV year

2014

Released

15. 9. 2014

Location

Guangzhou, Čína

ISBN

978-1-934142-28-8

Book

Proceedings of PIERS 2014 in Guangzhou

ISBN

1559-9450

Periodical

Progress In Electromagnetics

State

United States of America

Pages from

1861

Pages to

1864

Pages count

4

BibTex

@inproceedings{BUT109610,
  author="Jan {Mikulka}",
  title="Multiparametric Biological Tissue Analysis: A Survey of Image Processing Tools",
  booktitle="Proceedings of PIERS 2014 in Guangzhou",
  year="2014",
  journal="Progress In Electromagnetics",
  pages="1861--1864",
  address="Guangzhou, Čína",
  isbn="978-1-934142-28-8",
  issn="1559-9450"
}