The 3D-PATHOLOGY proposal addresses a long acknowledged need for quantification of parameters in pathology. The principal objectives of the proposal are: using real-time embedded technologies, to develop a system which is capable of increasing diagnostic and prognostic accuracy. This is done by accumulating a colour 3D image of cells or tissue in microscopic resolution within the computer, and subsequently to expose the synthesised image to full colour quantitative image analysis, thus obtaining valuable information about the material in hand. Also intuitive ways of visualising 3D images are considered. Potential applications to be tested in the project: cervical cytology, cytological diagnosis of body cavity effusions, urine cytology, fine needle aspiration cytology (FNAC), immunocytochemical stains.
The principal objective of the 3D-Pathology proposal is to create a new real time system for pathologists that makes it possible to decrease the diagnostic and prognostic error rate. This will be done by creating digital 3D light microscopy images for subsequent visual inspection and quantitative colour image analysis. To develop ways of visualising 3D images in order to heighten the level of insight to the pathologist. Technological highlights: As a result of the development work the following features are envisaged to be made possible: ?Resolution of cell clusters and tissue components?, ?The proper 3D extension of objects can be resolved?, ?True 3D colour image algorithms can be applied?, ?Virtual reality: The surroundings of a selected image object can be made transparent making it possible to view the object from any camera angle?, ?The ?software microscope? can be applied to the digital image?
DESCRIPTION OF WORK
Workpackages: The project has been broken down into 10 workpackages as described below:
1. Image acquisition. In order to create a synthetic image a number of individual image planes need to be recorded. Doing this, advantage is taken of the limited depth of focus within a microscope, resulting in image planes that are very shallow in sharpness;
2. Image synthesis3D images should to be synthesised both horizontally and vertically;
3. Image storage: For the typical and special specimens discussed above, the uncompressed storage requirements are 72 Gb and 192 Gb, respectively. Hence a fast compression technique is essential, both for short and long storage;
4. 3D image analysis: A new set of image processing/analysis tools are required for 3D images. In particular, the patented colour segmentation algorithm and cellular detection algorithm (patent pending) of Partner 1 should be extended to work on with 3D images. Extension from 2D to 3D is a non-trivial task, especially because an explosion in execution time has to be avoided;
5. The software microscope The software microscope is used on a 3D image, basically emulating the functions of a mechanical microscope as closely as possible, and benefitting from the world of virtual reality, too;
6. Virtual reality. Selected objects within a 3D image can be described in virtual reality terms, e.g. in VRML, the Virtual Reality Model Language;
7. Optimisation of interfacing and parallelisation of embedded units Partner 1 has experience in building a fast 2D image scanner with associated image analysis using a parallelisation scheme. However, the demands on a 3D image scanning and analysis system is much higher, and it is essential to optimise/redesign the interfacing and parallelisation of the embedded units (Partner 3);
8. Test applications 1-5To test the 3D image analysis on pertinent pathology examples;
9. Exploitation and dissemination;
Funding SchemeCSC - Cost-sharing contracts