Education- Ph.D. in Medical Image Processing and Computer Vision
Department of Computer Science
University of Bristol, UK
Automated Identification of Diabetic Retinal Exudates and the Optic Disc
Diabetic-related eye diseases are the most common cause of blindness in the world. Diabetic Retinopathy is a severe and widely spread eye disease which can be regarded as manifestation of diabetes on the retina. The screening of diabetic patients for the development of diabetic retinopathy potentially reduces the risk of blindness in these patients by 50%. Diabetic retinopathy can take two forms; background retinopathy, consisting of microaneurysms, haemorrhage, hard exudate,retinal edema, and sometimes microinfarcts of the retina (cotton wool spots),or proliferative retinopathy where new vessels develop in the retina and may bleed into the vitreous cavity. When background changes occur in the central retina, the condition is termed diabetic maculopathy, and visual acuity is at risk. This is the commonest sight threatening complication caused by diabetes. Much of this blindness can be prevented if the maculopathy is detected early enough for treatment with laser. Unfortunately, because visual loss is often a late symptom of advanced diabetic retinopathy, many patients remain undiagnosed even as their disease is causing severe retinal damage. Hence, there is an urgent need for mass-screening retinal examination for the early detection and treatment of diabetic retinopathy.
Current methods of detection and assessment of diabetic retinopathy are manual, expensive, potentially inconsistent, and require highly trained personnel to facilitate the process by searching large numbers of fundus images. Many of these images from screening programmes will be normal, but some will require grading of abnormalities (microaneurysms, haemorrhages, hard exudates, cotton wool spots) by severity and distance from the fovea (area of highest acuity), so that a judgement can be made on whether treatment is required. When abnormalities not requiring immediate treatment are found, the frequency of fundus imaging may be increased to every 4 to 6 months, and series of images compared to look for sight threatening trends requiring treatment. In contrast to this, a good, automatic method based on modern digital image processing technique will be faster, will need less may be no human intervention, and will yield consistent results.
Intraretinal fatty (hard) exudates are a visible sign of diabetic retinopathy and also a marker for the presence of co-existent retinal oedema. Detecting retinal exudate lesions in a large number of images generated by screening programmes, is very expensive in professional time and opens to human error. In this research we explore the benefits of developing an automated decision support system for the purpose of detecting and classifying exudate pathologies of diabetic retinopathy. The retinal images are automatically analysed in terms of pixel resolution and image-based diagnostic accuracies and an assessment of the level of retinopathy is derived.
A pixel-level exudate recognition approach is first attempted to discriminate the exudates from other retinal anatomical-pathological structures and artifacts. To estimate the exudate and non-exudate probability density distributions, K nearest neighbour, Gaussian quadratic and Gaussian mixture model classifiers are investigated. The preliminary pixel-level exudate recognition analysis has been used to support the fact that the development of a reliable and accurate exudate identification system is feasible. We explore another method, i.e. region-level exudate recognition to identify the retinal exudates based on an object recognition scheme. This includes colour image segmentation and region level classification based on neural network and support vector machine classifier models.
The location of the optic disc is of critical importance in retinal image analysis and is required as a prerequisite stage of exudate detection. Therefore, we also address optic disc localisation and segmentation both to improve the overall diagnostic accuracy by masking the false positive optic disc regions from the other sought exudates and to measure its boundary precisely. We develop a method based on colour mathematical morphology and active contours to accurately localise the optic disc region. More details are provided in my Thesis.
MRI Breast Image Segmentation for Treatment Planning
Developing an automatic geometric-based active contour model for segmenting chest boundaries, lung cavities and the heart from the MRI breast image sequences. The segmentation results allow the treatment planner to calculate radiation dose to sub volumes, and hence design treatment plans that minimise radiation to these critical areas
- M.Sc. in Computer Engineering (Machine Intelligence and Robotics)
- B.Sc. in Computer Engineering (Hardware)
Alireza Osareh, Majid Mirmehdi, Barry Thomas and Richard Markham. Automated identification of diabetic retinal exudates in digital colour images. British Journal of Ophthalmology,volume 87(10): 1220-1223, October 2003. Acrobat:1879487 bytes.
Richard Markham, Alireza Osareh, Majid Mirmehdi , Barry Thomas and Maria Macipe. Automated Identification of Diabetic Retinal Exudates using Support Vector Machines and Neural Networks. The Association for Research in Vision and Ophthalmology Conference, May 2003.
Alireza. Osareh, Majid. Mirmehdi, Barry. Thomas and Richard. Markham. Comparative Exudate Classification using Support Vector Machines and Neural Networks. 5th International Conference on Medical Image Computing and Computer-Assisted Intervention, T. Dohi and R. Kikinis , editors, pages 413-420. Springer LNCS 2489, September 2002. Acrobat:426256 bytes.
Alireza Osareh, Majid Mirmehdi, Barry Thomas and Richard Markham. Comparison of Colour Spaces for Optic Disc Localisation in Retinal Images. Proceedings of the 16th International Conference on Pattern Recognition , R. Kasturi, D. Laurendeau and C. Suen, editors, pages 743-746. IEEE Computer Society, August 2002. Acrobat:555867 bytes.
Alireza Osareh, Majid Mirmehdi, Barry Thomas and Richard Markham. Colour Morphology and Snakes for Optic Disc Localisation. The 6th Medical Image Understanding and Analysis Conference, A Houston and R Zwiggelaar, editors, pages 21-24. BMVA Press, July 2002. Acrobat:208159 bytes.
Alireza Osareh, Majid Mirmehdi, Barry Thomas and Richard Markham . Classification and Localisation of Diabetic-Related Eye Disease. 7th European Conference on Computer Vision, A. Heyden, G. Sparr, M. Nielsen and P. Johansen, editors, pages 502-516. Springer LNCS 2353, May 2002. Acrobat:677791 bytes.
Alireza. Osareh, Majid. Mirmehdi, Barry. Thomas and Richard. Markham. Identifying Exudates in Diabetic Maculopathy. 2nd International Workshop on Computer Assisted Fundus Image Analysis, Bjarne Ersboll, editor, pages 17-17. TU Denmark, October 2001.
Alireza. Osareh, Majid. Mirmehdi, Barry. Thomas and Richard. Markham. Locating the Optic Disk in Retinal Images. 2nd International Workshop on Computer Assisted Fundus Image Analysis, Bjarne Ersboll, editor, pages 35-35. TU Denmark, October 2001.
Alireza. Osareh, Majid. Mirmehdi, Barry. Thomas and Richard Markham. Automatic Recognition of Exudative Maculopathy using Fuzzy C-Means Clustering and Neural Networks. Medical Image Understanding and Analysis, E Claridge and J Bamber, editors, pages 49-52. BMVA Press, July 2001. Acrobat:80027 bytes.
Member of the British Machine Vision Association (BMVA)