This article presents a Middleware based on Grid Technologies that addresses the problem of sharing, transferring and processing DICOM medical images in a distributed environment using an ontological schema to create virtual communities and to define common targets. It defines a distributed storage that builds-up virtual repositories integrating different individual image repositories providing global searching, progressive transmission, automatic encryption and pseudo-anonimisation and a link to remote processing services. Users from a Virtual Organisation can share the cases that are relevant for their communities or research areas, epidemiological studies or even deeper analysis of complex individual cases.Software architecture has been defined for solving the problems that has been exposed before. Briefly, the architecture comprises five layers (from the more physical layer to the more logical layer) based in Grid Thecnologies. The lowest level layers (Core Middleware Layer and Server Services Layer) are composed of Grid Services that implement the global managing of resources. The Middleware Components Layer provides a transparent view of the Grid environment and it has been the main objective of this work. Finally, the upest layer (the Application Layer) comprises the applications, and a simple application has been implemented for testing the components developed in the Components Middleware Layer.Other side-results of this work are the services developed in the Middleware Components Layer for managing DICOM images, creating virtual DICOM storages, progressive transmission, automatic encryption and pseudoanonimisation depending on the ontologies. Other results, such as the Grid Services developed in the lowest layers, are also described in this article. Finally a brief performance analysis and several snapshots from the applications developed are shown.The performance analysis proves that the components developed in this work provide image processing applications with new possibilities for large-scale sharing, management and processing of DICOM images. The results show that the components fulfil the objectives proposed. The extensibility of the system is achieved by the use of open methods and protocols, so new components can be easily added.