| EPSRC Reference: |
GR/J13311/01 |
| Title: |
MODEL-BASED INTERPRETATION OF 3D MEDICAL IMAGES |
| Principal Investigator: |
Taylor, Professor CJ |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Imaging Science & Biomedical Eng |
| Organisation: |
Victoria University of Manchester, The |
| Scheme: |
Standard Research (Pre-FEC) |
| Starts: |
01 April 1993 |
Ends: |
30 September 1996 |
Value (£): |
173,925
|
| EPSRC Research Topic Classifications: |
|
| EPSRC Industrial Sector Classifications: |
| Healthcare |
No relevance to Underpinning Sectors |
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
|
Develop a fully automated method of generating 3D anatomical models from examples Build a 3D anatomical model of the brain based on a training set of Magnetic Resonance images Improve existing methods for matching anatomical models to volume image data Develop methods for dealing with image containing abnormal structure Evaluate the robustness and accuracy of automated interpretation experimentallyProgress:The project is intended to build on earlier SERC-funded work on the use of statistical shape models in automated interpretation of 3D (volume) medical images, and is complementary to GR/H83676. 3D medical imaging using magnetic resonance imaging, single proton emission tomography, ultrasonography etc. is becoming increasingly commonplace. In principle these methods can give much more clinical information than their 2D predecessors, but it is difficult to display and analyse the resulting 3D images, which are extremely complex. A key requirement is for automated interpretation - the location and identification of all relevant structures in the imaged volume. Methods for representing and using anatomical knowledge are essential in achieving fully automated interpretation. The feasibility of using statistical models to interpret 3D magnetic resonance images of the brain was established in our previous work though a number of factors made the approach difficult to apply generally; the aim here is to address some of these difficulties. We have made substantial progress towards achieving four of the five specific project objectives. Our existing methods were based on placing landmark points consistently in each of a training set of images - this was difficult in 3D. We have developed several methods of automatic landmark placement based on establishing correspondences between points on object boundaries via matching curvature and similar local properties. We have also shown that an optimisation approach can be used to improve the positions of an approximate set of landmarks. The feasibility of the approach has been demonstrated in 2D and is currently being extended to 3D. We have improved the iterative model-matching scheme described previously by developing the underlying theory and implementing a more efficient and exact updating algorithm. Using these improved methods we have shown experimentally that the surfaces of brain structures in new images can be located to an accuracy of better than one voxel. We have also begun to add new structures to our brain model.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
|