Segmentation: Introduction and Use Cases

We have discussed the problem statement for semantic segmentation and a few use cases for segmentation in 3D medical imaging:

Longitudinal follow up: Measuring volumes of things and monitoring how they change over time. These methods are very valuable in, e.g., oncology for tracking slow-growing tumors.
Quantifying disease severity: Quite often, it is possible to identify structures in the organism whose size correlates well with the progression of the disease. For example, the size of the hippocampus can tell clinicians about the progression of Alzheimer's disease.
Radiation Therapy Planning: One of the methods of treating cancer is exposing the tumor to ionizing radiation. In order to target the radiation, an accurate plan has to be created first, and this plan requires careful delineation of all affected organs on a CT scan
Novel Scenarios: Segmentation is a tedious process that is not quite often done in clinical practice. However, knowing the sizes and extents of the objects holds a lot of promise, especially when combined with other data types. Thus, the field of radiogenomics refers to the study of how the quantitative information obtained from radiological images can be combined with the genetic-molecular features of the organism to discover information not possible before.

Now, let’s take a look at some of the methods that are commonly used for building segmentation CNNs.

Segmentation - the problem of identifying which specific pixels within an image belong to a certain object of interest.