Java程序辅导

IDEA Lab application
Title: Using GPU computing technology to annotate tree-like structures in 3D volumetric data
1. Who are you? The names and affiliations of all intended applicants.
David Willshaw ANC
Seymour Knowles-Barley ANC
Adrianna Teriakidis ANC / Euan McDonald Centre for Motor Neurone Disease Research in the School of  Biomedical Sciences
In collaboration with the neuron tracing team.
2. What will you do? A brief outline of the project and its background, including the objectives and methods to be employed.
We will develop an image processing library for the discovery of tree-like structures in volumetric 3D data on a Graphics Processing 
Unit (GPU) and line tracing algorithms that can utilise this technology. Modern graphics cards have many small processing units 
designed to operate on large matricies in parellel. The cross-platform standard OpenCL will be used to implement 3D filtering and 
image analysis algorithms on the GPU. Performance will be tested on algorithms which can automatically trace neuronal 
morphologies from 3D image stacks acquired through a variety of imaging methods (confocal, light, two-photon). Identifying tree-
like structures in this way can also be used to measure blood vessels in images of retina, or large arteries in CT scans.
3. How is it novel? What is exciting about it?
A quick, flexible, accurate and automatic image processing software package will enable biomedical researchers to analyse 
orders of magnitude more data. In particular this could have a huge impact on the advancement of the field of neuroscience 
and our understanding of the brain.
For certain operations GPU processing is many times faster than performing the same operations on a standard computer CPU. 
Up to 100x performance improvement can be achieved in some cases. Large 3D volumetric data is very difficult to analyse 
because processing time required for each operation quickly becomes prohibitive. 
Neural reconstruction is such a common problem faced by many neuroscientists that the Allen Institute for Brain Sciences and 
the Howard Hughes Medical Institute launched the DIADEM (Digital Reconstruction of Axonal and Dendritic Morphology) 
competition in order to encourage the development of automatic tracing algorithms. The competition organisers have provided 
5 datasets including gold standard (manual) reconstructions and have specified metric of success, thus the problem we are 
tackling is hard but well-defined. More information about the competition can be found at http://www.diademchallenge.org/.
4. What will you do next? What opportunities will it open up?
If successful, this technology will be applied to automatic registration and neuron tracing as part of an entry for the DIADEM 
challenge. This form of analysis can also be applied to medical imaging and other medical research. In future we would like to apply 
this analysis to fruit fly brain data from our own department, electron microscopy data in collaboration with a Neurobiology group at 
Dalhousie University, and data collected by Peter Kind’s lab at the Centre for Integrative Physiology (CIP) at Edinburgh University.
Examples of volumetric data that may also benefit from this type of analysis include biomedical imaging such as MRI, fMRI, PET, 
CT, and CAT scans, and biological microscopy such as Confocal microscopy, Electron Microscopy, 2-Photon Microscopy.
5. What constitutes success? How risky is it?
Success will be a library of tools for tracing tree-like structures from 3D data. Additionally success can be measured by how much 
faster 3D volumes can be analysed by utilising this technology. Documentation and examples of how to utilise the library from Java 
and Matlab code will be provided.
Risk for this project is low. Goals are achievable and we already have knowledge of suitable tracing algorithms. Software support for 
GPU processing is well established and large speed improvements have been shown for many image processing operations. A small 
team from the ANC / DTC have already identified algorithms able to do some neuron tracing on images in the DIADEM image set. 
These algorithms are expected to run faster on a GPU and performance benefits should allow better tracing performance to be 
achieved.
6. What resources do you bring to the project?
Experience of working with 3D volumetric data such as confocal microscope data, electron microscopy data. Also there is a 
library of 3D confocal images available for analysis (see http://fruitfly.inf.ed.ac.uk/braintrap ) and in collaboration with the 
neuron tracing team we have experience with the DIADEM image dataset and challenges faced when analysing this type of 
data.
7. What resources and expertise do you need? Have you already identified sources for these, e.g. suitable staff available for 
short-term employment?
One or two people should be able to complete this investigation. There are several suitable staff available for this project from the 
ANC. Further benchmarking and analysis work may be carried out in collaboration with other 3D volume analysis groups within the 
School of Informatics. Access to PCs or laptops with OpenCL compatible graphics cards would be required to achieve this task. 
Advice and training on OpenCL programming would be also useful, but we will be able to learn this unaided if required.
8. What shared resources, if any, will the project create?
A GPU-based library of 3D volumetric data analysis operations will be created and made available as open source software. Also 
experience of how to properly integrate software with GPU technology which may be valuable to other groups. If purchased, GPU PCs 
or laptops will be available to other IDEA Lab projects.
9. What is your proposed timescale?
2.5 months, starting in January 2010.
4 weeks for OpenCL implementation and integration with Java, Matlab.
2 weeks for testing and bug fixing
3 weeks for data analysis, benchmarking and performance optimisation
1 week for documentation
10. What help do you need from IDEA lab to make this project happen?
Funding for 2.5 person months and access to GPU processing computers for implementation, testing and data analysis. The budget 
estimate below sets funding cost at £3k per month (based on RA rates).
Budget estimate:
Financial support for 2.5 months: £7.5k
GPU enabled computers: £2.0k
Total £9.5k