Medical image processing that includes not only two-dimensional but three-dimensional images requires massive computations. Thus, high-performance computer systems must be used, but they are very expensive. Even if the existing PC clustering technology is adopted, expensive network cards are needed. In this thesis, we implement distributed computing system based on CORBA(Common Object Request Broker Architecture) for medical image processing with low cost. For the research, we connected five PCs through an ATM switch to build distributed parallel processing facility. Such system concept allows distributed processing to be possible by using low-cost computers and operating systems without constructing new system. In this system, each server works not as a separate system but as a processing element for parallel processing. To verify correct operations of the environment, we try to actually process the sharpening filter algorithm that is one of medical image processing applications. Since the algorithm requires data communications among servers during run time, it proves any applications including interprocessor communications can be processed on the distributed computing system. The processing time on our system for the algorithm is 2.05 second, while that on a single processor system is 2.59 second, such that 1.26 times of speedup is obtained. Performance will be more improved if the time overhead due to inter-computer communication is reduced by using faster networks. The implementation technology used in this paper will be useful to construct more flexible and efficient medical information system in the future.