The medical task can often split into three areas:
(1) data operations of filtering, noise removal, and contrast and feature enhancement
(2) detection of medical conditions or events
(3) quantitative analysis of the lesion or detected event.
Of these subtaskt, detection of lesions or other pathologies is often a subjective and qualitative decision, a type of process ill-suited for execution by a computer.
By contrast, the computer is vastly more capable of both quantitative measurent of the medical condition.
The natural partnership of humans and machines in medicine is to provide the clinician with powerful tools for image analysis and measurement, while relying on the magnificent capabilites of the human visual system to detect and screen for the primary findings.
We divide the probles inherent in medical image processing into thress basic catories:
Filtering: These are the basic tasks involved in filtering and preprocessing the data before detection and analysis are perfomed either by the machine or the human operator
Segmenation:This is the task of partitioning an image(2D array or volume) into contiguous regions with cohesive properties
Registration:This is the task of aligning multiple data streams or images, permitting the fusion of different information creating a more powerful diagnostic tool than any single image alone.