Diabetes Detection Using Iris Image

The diabetes detection utilizing iris images was an extensive investigation into the application of machine learning techniques for medical image analysis. The primary objective was to develop an effective system capable of distinguishing between diabetic and non-diabetic individuals based on iris images. The Proposed system comprised several phases, including image preprocessing, feature extraction using Histogram of Oriented Gradients (HOG), and classification using the AdaBoost algorithm. Iris images were preprocessed by converting them to grayscale, resizing them to a uniform size, and normalizing their pixel values. HOG was employed to extract meaningful features from the images, capturing the essential gradient and edge information necessary for classification. The AdaBoost classifier, with a decision tree as the base learner, was then trained to differentiate between diabetic and non-diabetic iris images. A comprehensive dataset of iris images was used, split into training and testing sets to ensure robust model validation. Cross-validation was performed to assess the model's generalization capability, resulting in a low cross-validated loss, indicative of the model's accuracy and reliability. For feature extraction and classification, the suggested solution successfully integrates Python's machine learning packages, including scikit-learn and scikit-image. A user-friendly interface for classifying test images was also included, demonstrating the proposed system's usefulness.