were produced, MRI has continued to surprise us. I well remember the first strange image of a finger that I saw almost 18 years ago. Since then progress has been incredible and extremely fast. All the supposed limits of MRI have become its strengths. We used to say (and think) that MRI was unable to see bone; nowadays it provides us with perhaps the best examination technique for evaluating bone density. We used to believe also that MRI could not visualise vessels nor read the signal from moving protons, but now MR angiography is here, to demonstrate the error of our mistrust. We were so sure of the slowness of MR acquisition times that we could never have foreseen the development of high speed imaging. Now, it is all too easy to discuss the magnificent future of MRI in medicine in terms of a seemingly endless list of opportunities. Today, it is clear that MRI provides a completely new approach to routine neuroradiological study. Highspeed acquisition times combined with functional capabilities provide new perspectives, moving from one of morphology to one of function and from statics to dynamics. The level of morphological examination with MRI is outstanding, and although functional studies are still in their infancy, they can nevertheless still be performed daily on the majority of MR systems. In the early days of digital radiology, I was fascinated by the matrix and the use of computers. At that time, I discovered that Piet Mondrian, the painter, had been able, some 70 years ahead of the game, to go inside the image (The red tree, 1908), to draw a matrix (The chessboard, 1919) (Fig. 1), and then to paint many portraits of pixels seen from a very close perspective (compositions 1917±1937). Mondrian's research fascinates me; as a radiologist I believe that our discipline, an imaging discipline, has followed the same pathway: from analogic imaging of conventional radiology to the digitised images of computed tomography (CT) and MRI. Eur. Radiol. 7 (Suppl. 5), S 164±S 165 (1997)  Springer-Verlag 1997