Precision Diagnosis of Diabetic Retinopathy Using Exudate-Focused SVM Models

2024-11-24T06:25:00+03:00

The global concern over diabetes-related eye diseases continues to grow significantly. Diabetic retinopathy, caused by heightened glucose levels in retinal capillaries, leads to vision clouding and eventual blindness. Early detection through regular screening enables intervention with medication, preventing further vision deterioration. Therefore, this study introduces a smart application that utilizes digital retinal image processing to aid in the prompt identification of diabetic retinopathy. The application streamlines the analysis of eye images, with the goal of automatically classifying the severity of diabetic retinopathy. Through initial image processing, specific features such as blood vessels, microaneurysms, and hard exudates are identified and extracted for classification using a support vector machine (SVM). Evaluation performed on a dataset of 400 retinal images graded on a 4-grade scale of non-proliferative diabetic retinopathy achieved a maximum sensitivity rate of 95%. This application holds significant potential for enabling timely intervention in the treatment of diabetic retinopathy by healthcare professionals. Additionally, the AI-driven approach proposed in this study empowers patients to easily access support services, while providing physicians and researchers with advanced tools for analyzing and predicting diabetic retinopathy data. The resulting reports play a crucial role in assessing the severity of the disease in affected individuals.

IoT Enabled Non-Invasive Glucose Monitoring Through Breath Acetone

2024-12-13T20:37:43+03:00

This paper presents a non-invasive blood glucose monitoring system integrated with Internet of Things (IoT) technology using breath acetone detection. The system utilizes a TGS822 gas sensor to detect acetone levels in exhaled breath, which are correlated with blood glucose concentration. To enhance accuracy, environmental parameters such as temperature, humidity, and pressure are measured using DHT11 and BMP180 sensors. Sensor data are processed using Arduino-based signal acquisition and regression analysis techniques to estimate glucose levels, which are displayed in real-time on an LCD and transmitted for remote monitoring. Experimental validation was conducted on 11 subjects, and results demonstrated a strong correlation with standard glucometer readings, achieving an accuracy of approximately 98%. The proposed system offers a reliable, painless, and cost-effective alternative for diabetes management.

Journal of Computers, Mechanical and Management

Precision Diagnosis of Diabetic Retinopathy Using Exudate-Focused SVM Models

IoT Enabled Non-Invasive Glucose Monitoring Through Breath Acetone