 |
Pro Vice Chancellor and Head of the College of Science |
|
Research Interests Computer Vision / Artificial Intelligence / Medical Applications of Vision and Automated Surveillance |
|
|
Project: ESR 2.1 Retinal Vascular Measurement and ESR 3.1 Diagnosis and Early Screening for Diabetic Retinopathy ESR 2.1: We will develop a complete vascular segmentation algorithm that detects and accurately measures all vessels in the retina. The base measurement technique will follow Lincoln’s development of highly accurate active contour algorithms that simultaneously locate both edges of the vessels. However, this will be augmented by using explicit models of the vasculature from WP 1 to resolve uncertainties (e.g. overlapping vessels, complex junctions) by utilizing maximum likelihood principles under uncertainty. It will be further extended to include a statistical characterization of uncertainty. The model will be extensively validated on the dataset from WP 4, and used to validate models in WP 1 and for diagnostic purposes in WP 3. ESR 3.1: Screening for Diabetic Retinopathy (DR) is currently primarily performed by detection of lesions, and at advanced stages vascular changes including venous beading and neovascularisation. However, there is strong evidence that changes to detectable vascular geometry occur prior to the appearance of lesions. This project will analyze vascular changes in early DR in the light of the models developed in WP 1, and investigate features that can be extracted using techniques from WP 2. The aim is to quantify risk factors, including for initial development of diabetic retinopathy and progression to more advanced stages, including sight-threatening diabetic retinopathy. These will be correlated with findings from lesion detection algorithms previously developed by partner 1, to determine whether vascular detection can replace or augment lesion detection for DR screening. Direct collaborators: University Lincoln for the segmentation of the entire retinal vasculature from imaging data. |
|
|
Major 5 publications Al-Diri, Bashir and Hunter, Andrew and Steel, David and Habib, Maged (2010) Automated analysis of retinal vascular network connectivity. Computerized Medical Imaging and Graphics, 34 (6). pp. 462-470. ISSN: 0895-6111 Al-Diri, Bashir and Hunter, Andrew and Steel, David (2009) An active contour model for segmenting and measuring retinal vessels. IEEE transactions on medical imaging, 28 (9). ISSN: 0278-0062 Lowell, J. and Hunter, Andrew and Steel, D. and Habib, M. (2006) Retinal lesion segmentation evaluation.Investigative Ophthalmology Visual Science, 47. Lowell, J. and Hunter, Andrew and Steel, D. and Basu, A. and Ryder, R. and Kennedy, R. L. (2004) Measurement of retinal vessel widths from fundus images based on 2-D modeling . Medical Imaging, IEEE Transactions on, 23 (10). pp. 1196-1204. Hunter, Andrew (2004) Automatic computerised screening for diabetic retinopathy. Diabetes research. pp. 47-50. |
|