Educational Exhibit at the European Congress of Radiology, Vienna, 2019
(Note: images here are for representation only. Please write to us in case you would like the actual images)
Learning objectives
We present a review of various software and algorithms that can be used by radiologists to accurately delineate regions of interest and extract them from medical imaging data. We present a comparison of their power, efficiency, and easiness.
Background
Image Segmentation is as a sub-field of Computer Vision, wherein certain pixels in an image are extracted or ‘segmented out’ creating ‘digital partitions’ within the images. Classically, this process is used to identify objects and boundaries in images, by tagging and subsequently categorizing pixels according to certain characteristics. Another example is of a framework called an active contour model, popularly called ‘snakes’, which
helps separate the foreground object from the noisy background. While segmentation is the separation of image data into various components on the basis of a combination of properties such as grey scale level and contrast, for complete visual recognition one needs to recognize or ‘tag’ the different components so as to be clinically applicable; this latter process is called annotation.
Findings and procedure details
Introduction
Description of Software
Open-Source Software
- 3D Slicer
- ITK Snap
- ImageJ
Commercial Software
- Mimics
- Amira
- Labelbox
Conclusion
Some of the popular software has been discussed here based on their UI, features, efficiency and workflow integration into Artificial Intelligence models. While different software have their pros and cons, it is imperative to choose the one which best suits your requirements, budget as well as technical skill set