Influence of Noise Filtering in Coding Computed Tomography with JPEG2000

Radiation exposure increases the risk of inducing cancer in a patient when Computed Tomography is performed. Radiologists reduce the radiation dose to minimize the risk of cancer, consequently, noise is introduced in the image, considerably penalizing its quality. To enhance the quality of the image, different noise filters are developed. The filtered noise can be discarded due to its insignificance for the medical diagnosis. In this research, a coding scheme for Computed Tomography with JPEG2000 including a noise filtering stage is proposed. The JPEG2000 standard is selected in our coding approach because: it is supported in DICOM, the communication protocol in the medical setting, it includes progressive lossy-to-lossless coding, and it incorporates JPIP, the interactive transmission protocol. Extensive experimental tests using different images suggest that noise filtering does not penalize the visual quality, but allows substantial improvements in coding performance.