Compression of Digital Medical Images Based on Multiple Regions of Interest

Advances in digital medical imaging technologies, particularly magnetic resonance imaging and multi-detector CT (Computed Tomography), have resulted in substantial increase in the size of datasets, as a result of improvement in spatial and temporal resolution. In order to reduce the storage cost, diagnostic analysis cost and transmission time without significant reduction of the image quality, a state of the art image compression technique is required. We implemented a context-based and regions of interest (ROI) based approach to compress medical images in particular vascular images, where a high spatial resolution and contrast sensitivity is required in areas such as stenosis. The vascular image is divided into: the primary region of interest (PROI), the secondary region of interests (SROI) and the background. The PROI can be a stenosis of vessel and it is identified manually by the radiologist. The SROI is divided into other parts or regions among which the most important level is represented by vessels. The other levels are the other tissues or part of the body and the last level is the background region. The SROI is detected automatically by an in house 3D region growing algorithm. The PROI is considered as a seed for region growing. The proposed lossy-to-lossless region-based compression method is compressed these multiple ROIs at various degrees of interest and at higher precision (up to lossless) than other areas such as background. To demonstrate the result of this algorithm, this method is applied on peripheral arteries images (up to 2000 images) and the result have been compared with standard Jpeg2000 on 10 datasets. The size of compressed images can be reduced up to 67 percent.