Abstract:

Diabetes occurs due to destruction of Beta-cells in the pancreatic islets of Langerhans with resulting loss of insulin production. The result of insufficient action of insulin is an increase in blood glucose concentration. The diagnosis of Diabetes must always be established by a blood glucose measurement made in an accredited laboratory. The alternative way to measure a deficiency of insulin from the Beta-cells of pancreatic islets uses iris diagnosis. Evaluating the iris is done by detecting the presence of some broken tissue in iris. However, conventional iris diagnosis is always concerned with the identification of syndromes rather than with the connection between abnormal iris tissue appearances and diseases. In this paper, we present a novel computerized iris inspection method aiming to address these problems for detecting insulin deficiency from the Beta-cells of pancreatic islets. First, quantitative features, textural measures are extracted from iris images by using popular digital image processing techniques. Then, Neighborhood based Modified Backpropagation using Adaptive Learning Parameters (ANMBP) method is employed to model the relationship between quantitative features and pancreatic abnormalities as caused of insulin deficiency. The effectiveness of this method is tested on 12 patients with Diabetes, and the diagnostic results predicted by the previously trained ANMBP classifiers are compared with the calculation of HOMA-B, obtained 83.3% accuracy in detecting pancreas disorders.