Multi-class 3D region growing algorithm
Abstract
This paper describes generalization of multi-class region growing algorithm allowing for segmentation of 3D images (series of slices). The multi-class region growing algorithm was proposed in [1]. Additionally, a new method for finding the start region was presented. As its 2D version the new algorithm does not need initial parameters since it features segmentation quality assessment. A series of segmentations is performed on a dataset, each segmentation quality is assessed and the best one is picked. Additionally, the number of classes in the image is determined automatically. The multi-class 3D region growing algorithm is tested on CT and MRI scans. Different types of MRI scans are used. The scans come from multiple sources. The results are shown in the form of 3D reconstruction accompanied by a selected 2D slice. In addition, such selected algorithm performance issues are discussed as effective algorithm implementation; fast dilation implementation is also mentioned. The paper explains all concepts and operations used by the algorithm. It includes numerous figures and algorithm pseudo-code descriptions.
Full Text:
PDFDOI: http://dx.doi.org/10.17951/ai.2004.2.1.227-236
Date of publication: 2015-01-04 00:00:00
Date of submission: 2016-04-27 10:11:15
Statistics
Total abstract view - 473
Downloads (from 2020-06-17) - PDF - 0
Indicators
Refbacks
- There are currently no refbacks.
Copyright (c) 2015 Annales UMCS Sectio AI Informatica
This work is licensed under a Creative Commons Attribution 4.0 International License.