Medical Image Processing in an Era of High-Performance Computing

Advanced radiology practices are already benefiting from powerful and increasingly more economical computing and networking facilities. Medical image processing methods have improved dramatically over the past five years, with sophisticated 3D display, visualization and analysis techniques allowing increased integration of multiple modalities of imaging, flexible environments for imaging analysis, and PACS (picture archiving and communication systems) for ease of transmission and retrieval. Emerging directions involve teleradiology and telesurgery virtual reality applications, the development of new image database techniques, and the building of large visual databases like that of the Visible Human Project. Challenging problems of image segmentation, registration, and multimodal image fusion are still with us. Building dynamic, flexible electronic atlases will have a profound effect on the understanding of structure and function from the level of cellular physiology to gross anatomy, but requires the development of new techniques of visual knowledge representation and more standardized ways of defining the conceptual and linguistic constructs of visual objects in biomedicine, for linkage to medical records, research results, and educational materials. Methods for reasoning with visual information in the context of multimedia information systems present an inviting challenge to the upcoming generation of researchers in medical informatics.