Developing medical avatars
The idea of an exact virtual replica of yourself, accurate in every detail of your anatomy and physiology, might sound like the stuff of science fiction, but a Europe-wide £72million research programme is developing patient-specific computer models with the ultimate aim of achieving truly personalised healthcare.
The Medical Physics Group at the University are participating in several EU funded research projects within this programme , and are working to develop computer models tailored to an individual patient, which could one day be used by doctors to determine more effective treatments which take into account each patient´s unique physiology.
The EU funded SimBio project started in 2001 with nine participating academic and industrial partners. The aim was to develop computational tools for building personalised biomechanical models. The Sheffield component of this project, led by Professor Rod Hose, developed a methodology for constructing patient-specific computer models of the knee. The idea is that the model can be used by surgeons to evaluate possible surgical interventions, giving them the opportunity to `practice´ on a virtual knee before the actual operation.
To construct the model a CT or MRI scan of the knee is performed. The individual anatomical components of the knee, the bones and the various soft tissues, need to be identified within these images so that a computational model of the need can be constructed. Although segmentation of the image into anatomical components can be done manually this is both time consuming and not very accurate and this is not practical for routine clinical use. Professor Rod Hose realised that it should be possible to perform segmentation of the knee automatically using an image registration algorithm developed by his colleague Professor David Barber, then Scientific Director of the Directorate of Medical Imaging and Medical Physics, Sheffield Teaching Hospitals, and a collaboration was set up to do this.
Professor Hose explains: "The biggest challenge is identifying the segmentation of the joint, or how the bones and soft tissues are arranged within the knee. This is very difficult for a machine to do, and it looked like our research staff would have to segment each 3D image by hand, which was a very costly and time-consuming process. This simply couldn´t be done on a large scale, but we realised that, having segmented one image by hand, which we called our reference image, we could use our Image Registration Algorithm software to segment further image. This has proved crucial in making patient specific computer models possible".
Put simply, the process is as follows. An image is taken of a patient´s knee. Using the Image Registration Algorithm this image is then registered to a reference knee image previously segmented by hand, and these segments are then mapped back on to the patient image, effectively segmenting this image in a fraction of the time it would have taken to perform this manually. The same reference knee can be used for all patients.
This approach to segmentation has proved to be quite general; the same approach has been used to segment the heart and aorta as part of the EU funded EuHeart project (started in 2008). This €17m project, which includes seventeen industrial, clinical and academic partners is aimed at the development of individualized, computer-based, human cardio-vascular models.
The research forms part of a larger EU funded programme, which has invested £72m to develop the `Virtual Physiological Human´. Teams of researchers across Europe are working on different aspects of human physiology with the ultimate aim of producing computer models of individual patients.
Professor David Barber: "Although many decades away, the ultimate aim of the Virtual Physiological Human project is that you will have your own unique digital avatar, which will be an exact virtual replica of your anatomy and physiology. Medical professionals will be able to use your `avatar´ to test out various medical interventions, including surgery and the efficacy of medicines, to determine how your own unique physiology will respond, thereby ensuring a tailored, truly personalised medical intervention."
The team´s work is already having real-world applications, with a global orthopaedic company having acquired the licence to use the registration and segmentation software in orthopaedic planning for knee and hip surgery.
For further information, please contact:
email : d.r.hose@sheffield.ac.uk
email : d.barber@sheffield.ac.uk