Quality Assessment of JPEG and JPEG2000 Compressed Medical Images

The goal of this project with Agfa Healthcare, a provider of diagnostic imaging and healthcare IT solutions, is to develop mathematical models for (i) assessment and (ii) improvement of compressed medical images. The rapidly increasing volume of data generated by new imaging modalities (eg CT scanner, MRI) necessitates the use of lossy compression techniques to decrease the cost of storage and improve the efficiency of transmission over networks. Increasing the degree of compression of an image, however leads to decreasing fidelity. One of the causes of this error is the introduction of artifacts by the compression method. Such distortions impede the ability of radiologists to make confident diagnoses from compressed medical images. The intern, in consultation with radiologists, will identify the important features that become degraded as the compression level is increased. In particular the research team shall study the JPEG and JPEG2000 compression schemes. The next step, of particular interest to Agfa, is to use this information to devise other, perhaps related, compression schemes that will not degrade these features at specific levels of compression. A second fundamental aspect of this work is to examine the use of “image quality measures” in the assessment of compressed medical images.

Faculty Supervisor:

Dr. Edward Vrscay

Student:

Ilona Kowalik

Partner:

Agfa Healthcare

Discipline:

Mathematics

Sector:

Medical devices

University:

University of Waterloo