Problems And Limitations Of Knowledge-based 3d Medical Image Analysis
Ankush Dhawan · 2005
Knowledge-based image analysis and interpretation of radiological images can provide a tool for identifying and labeling each part of the image in the context of human anatomy. Further, such a computerized analysis can be used in correlation studies of images which are obtained through different imaging modalities (such as CT, MRI, PET) for the same patient for studying the pathology. Often, the medical images are characterized by low-contrast regions which utilize a narrow range of gray levels. Various methods including the gray level remapping may be used for improving the contrast in a general sense. But, even then it is difficult to obtain well defined and isolated segmented regions of interest from these images. This is one of the major problems in automatic analysis of medical images. Other problems including the improper registration, analysis, and visualization of small low-contrast regions are discussed in this paper in context of a knowledge-based analysis system which is being developed for analyzing the anatomical images of human chest cavity. The question on the issue of user interaction is raised to address these problems for efficient and better analysis.