Abstract:
Medical Image processing has assumed great significance as a result of rapid increase in the amount of patient data available. Many scenarios in medical treatment require specific information about the part being examined. The task of providing useful information to a practitioner requires extensive preliminary analysis. Several attempts have been made to automate this process by incorporating image processing techniques.
Because of the diverse range of medical images most of the techniques are specific to a particular set of images. Hence there arises a necessity to evaluate the algorithms on a common metric to determine their usefulness to a particular class of images. In the current work a methodology is proposed to create test image sets (synthetic images) which could be used as a common metric to evaluate the performance of algorithms for different pre-defined shapes. The tool developed in the current work also provides option for adding varying levels of noise to the test image.
The advantage provided by the tool is demonstrated by evaluating the performance of an existing algorithm. The results obtained can be used to determine the usefulness of the algorithm to specific cases.
