The paper describes a medical visualization system suitable for both diagnosis and training in medical practice. The work is using open source frameworks and libraries such as WebRTC (Web Real-Time Communication), JSARToolKit (JavaScript Augmented Reality Toolkit), Three.js and NanoDICOM. The system builds a 3D model from the slices of a patient's DICOM imagistic investigation files, superimposes the model on the body area of interest using Augmented Reality and sends the camera images remotely. The main advantage consists in real-time viewing of imagistic patient information by assisting physicians or medical students. Virtually attaching the 3D model to a specific area of the body allows the model to rotate and move in the same time with the patient. The proposed application does not depend on any operating system and runs inside a browser window without any closed source plug-in, allowing it to be accessed on any computer or mobile device that has a suitable browser.