Publication detail

Automatic Segmentation of Myocardial Infarction in Rats Subjected to Regional Ischemia

JAKUBÍČEK, R. CHMELÍK, J. NECKÁŘ, J. KOLÁŘ, R.

Original Title

Automatic Segmentation of Myocardial Infarction in Rats Subjected to Regional Ischemia

English Title

Automatic Segmentation of Myocardial Infarction in Rats Subjected to Regional Ischemia

Type

conference paper

Language

en

Original Abstract

The experimental and preclinical studies of ischemia and reperfusion on animal models usually evaluate the sizes of area at risk (AR) of infarction and infarct area (IA) as fundamental parameters. The authors usually don’t provide any detailed information about the image processing of their data, though the IA or AR segmentation is often challenging and prone to be expert-depending. Here, we describe a new approach for automatic IA and AR segmentation based on combination of Random Forest classifier and two-step pixel-wise k-means classification of image pixels. The evaluation has been performed on the set of 16 images from 8 rat hearts. We compared sizes of normal perfused tissues, viable area and IA (normalized to percentage of total area) obtained by our method with manually segmentation by biologist. We achieved mean absolute error of 2.59% with mean standard deviation of 1.61%.

English abstract

The experimental and preclinical studies of ischemia and reperfusion on animal models usually evaluate the sizes of area at risk (AR) of infarction and infarct area (IA) as fundamental parameters. The authors usually don’t provide any detailed information about the image processing of their data, though the IA or AR segmentation is often challenging and prone to be expert-depending. Here, we describe a new approach for automatic IA and AR segmentation based on combination of Random Forest classifier and two-step pixel-wise k-means classification of image pixels. The evaluation has been performed on the set of 16 images from 8 rat hearts. We compared sizes of normal perfused tissues, viable area and IA (normalized to percentage of total area) obtained by our method with manually segmentation by biologist. We achieved mean absolute error of 2.59% with mean standard deviation of 1.61%.

Keywords

image segmentation; infarct of myocard; histological image; heart; rat

Released

12.11.2018

Publisher

Computing in Cardiology

Location

Maastricht, Netherlands

Pages from

1

Pages to

4

Pages count

4

URL

BibTex


@inproceedings{BUT151065,
  author="Roman {Jakubíček} and Jiří {Chmelík} and Jan {Neckář} and Radim {Kolář}",
  title="Automatic Segmentation of Myocardial Infarction in Rats Subjected to Regional Ischemia",
  annote="The experimental and preclinical studies of ischemia and reperfusion on animal models usually evaluate the sizes of area at risk (AR) of infarction and infarct area (IA) as fundamental parameters. The authors usually don’t provide any detailed information about the image processing of their data, though the IA or AR segmentation is often challenging and prone to be expert-depending. Here, we describe a new approach for automatic IA and AR segmentation based on combination of Random Forest classifier and two-step pixel-wise k-means classification of image pixels. The evaluation has been performed on the set of 16 images from 8 rat hearts. We compared sizes of normal perfused tissues, viable area and IA (normalized to percentage of total area) obtained by our method with manually segmentation by biologist. We achieved mean absolute error of 2.59% with mean standard deviation of 1.61%.",
  address="Computing in Cardiology",
  booktitle="Computing in Cardiology 2018",
  chapter="151065",
  doi="10.22489/CinC.2018.128",
  howpublished="online",
  institution="Computing in Cardiology",
  number="1",
  year="2018",
  month="november",
  pages="1--4",
  publisher="Computing in Cardiology",
  type="conference paper"
}