The subject matter covered is intended to be computer vision as applied to medical images, with a special orientation toward visualisation and computer-assisted surgery. How do we take medical images, for example computerised tomography scans, and magnetic resonance imaging data and process them in interesting and novel ways so as to increase their diagnostic usefulness?
It is a young journal, only in its second year of publication, printed on high-quality paper with many illustrations in colour. Since many of the papers involve multimedia displays such as videos, a CD version of the journal is also provided and indeed required in order to understand fully the material presented.
A particular concern is the production of 3D displays, by combining data from different imaging equipment (modalities) so that they can be displayed and interpreted ("fused") together. The journal is more specific than, but in competition with a number of other publications, notably IEEE Transactions in Medical Imaging, Computer Graphics and Image Processing, IEEE Pattern Analysis and Machine Intelligence, and also the proceedings of a number of conferences in this area usually known only by their initials, eg IPMI (information processing in medical imaging), SPIE MI (medical imaging), VBC (visualisation in biomedical computing), MIUA (medical imaging understanding and Analysis) (a new United Kingdom meeting in this area), CARS (computer-aided radiology and surgery) and many others. The number of journals in the field, and in particular the number of conferences devoted to it, illustrate the interest of the subject.
Why so interesting? Because it is considered to be fun. The field has developed rapidly as a result of the availability of computers adequate to handle and generate displays of the medical data acquired digitally from an increasing variety of sources.
Thus MRI and CT provide in particular 3D data rather than conventional projection images (as on a normal X-ray film), which has revolutionised radiology. The 3D data, while providing much more information than was previously available, were traditionally viewed as a series of slices, but are increasingly used to generate 3D graphical images with which the observer can interact by rotating to look at them from different angles, editing to distinguish and maybe hide different tissue types, and to perform quantitative analysis such as the evaluation of lesion size, the estimation of blood flow, etc.
An exciting development has been to link these techniques with computer-aided surgery. A scenario now being imagined is that of the surgeon, equipped with a virtual-reality headset (one hopes lightweight) observing at the same time both the patient who is undergoing the operation and the 3D display generated from (for example) CT data. This could permit the surgeon to see perhaps the brain tumour to be removed plus the surrounding and obstructing blood vessels so that, before cutting into the patient's brain, the operation can be planned to minimise damage.
The subject is therefore both intellectually stimulating, satisfying (in that one has the feeling of doing something useful) and artistic. Many attractive pictures are generated.
One must ask if such techniques are in fact really useful and cost-effective. The clinical importance is still not proven. There is, however, hearsay evidence of their value and there is increasing demand for such methods from both radiologists and other clinicians including surgeons. The cost of the tools required to perform the analysis and visualisation has come down dramatically, but the cost of the imaging devices, and of the equipment that would be required in surgery, is quite high. A related field, that of picture archiving and communication systems (PACS), which is basically about providing widespread access to medical imaging on digital computer displays, has recently changed from being only of research interest, to being widely available and increasingly implemented. Medical image analysis as presented in this journal could and should follow on as the required technology develops.
The "Visible Man" project (left) is an interesting example of how rapidly changes occur in this field. When the criminal who provided that data was originally offered the chance of being preserved for posterity in this unusual manner, one suspects that he could have had no idea that, a mere three years later, people would look at images of him and say: "rather poor quality data, maybe we should try again".
In summary, this journal provides a forum for presenting high-quality papers in a very specific field at a leading edge of an exciting new discipline. It is, however, still clearly aimed at the computer scientist rather than the general clinician.
A. E. Todd-Pokropek is professor of medical physics, University College London.
Medical Image Analysis
Editor - Nicholas Ayache and Jim Duncan
ISBN - ISSN 1361 8415
Publisher - Oxford University Press
Price - £98.00 (individuals); £ 265.00 (institutions)
Pages - -