Segmentation

Last revised by Candace Makeda Moore on 5 May 2024

Segmentation, in the context of informatics for radiology, refers to the delineation of areas of interest in imaging in terms of pixels or voxels.

Segmentation is often accomplished by computerized algorithms that vary in complexity from simply selecting pixels of similar values in proximity to those that include radiomic features and/or are based on machine learning. The underlying algorithms that can be used for segmentation include, but are not limited to thresholding (simply taking all the pixels within a certain value range), K-means and Otsu's algorithm

To evaluate segmentation algorithms researchers have used ROC analysis, odds ratios, the superiority evaluation method, the deviation evaluation model, Hausdorff distances, contour mean distances, regression analysis, volumetric differences and Dice similarity coefficients to compare segmentations to expert segmented gold standards or statistically generated gold standards.

Segmentation is a necessary step for automated quantitative volumetric analysis of anatomical structures. It is also usually a necessary step in creating modifiable three-dimensional images for surface modelling, extended reality and/or medical 3D printing from imaging.

External links

Several software programs can be used to segment radiological images, usually but not exclusively CT or MRI. The following segmentation tools are optimized for radiological images that are free and open-source:

If any of these links are broken or for other problems and questions, please contact [email protected].

ADVERTISEMENT: Supporters see fewer/no ads