Microaneurysm identification using cross sectional profile analysis with optic disc removal
| Autor: | G. Nagarjuna Reddy, K. Durga Ganga Rao |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Pixel
Contextual image classification Computer science business.industry 0206 medical engineering Pattern recognition 02 engineering and technology Fundus (eye) 020601 biomedical engineering eye diseases Standard deviation 030218 nuclear medicine & medical imaging Cross section (geometry) 03 medical and health sciences 0302 clinical medicine medicine.anatomical_structure medicine Computer vision Artificial intelligence business Smoothing Optic disc Maximum Pixel |
| Zdroj: | 2016 International Conference on Communication and Electronics Systems (ICCES). |
| DOI: | 10.1109/cesys.2016.7889924 |
| Popis: | In this paper an enhanced technique is presented for the identification of microaneurysms (MAs) in retinal fundus images. Detection of MAs is necessary for diagnosis and severity classification of diabetic retinopathy (DR). In this technique optic disc is removed using ROI processing. The suggested technique achieves the MA identification using cross section profile analysis of local maximum pixels of the smoothing image. After detecting peaks for each profile, certain characteristics like width, height, shape of the peaks are obtained. For each local maximum pixel these properties are calculated in all directions around 360. The feature set for classification constitutes Statistical properties like mean, standard deviation, coefficient of variation of above specified characteristics. KNN classifier is used for the MA classification. The proposed technique is able to identify vessel crossings, optic veins, MAs very well rather than existing methods by removing optic disc in preprocessing. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|