Publication

Abstract : Cholesterol detection plays an important role in preventing heart diseases. The traditional lipid profile test for cholesterol detection is done using serum. For heart patients and diabetic patients, frequent measurement of cholesterol from serum is painful. So there is a need for a non-invasive method of cholesterol detection. Non-invasive method of cholesterol detection from various part of the human body is proved, however the measurement of cholesterol level is needed. In this paper, we proposed a novel non-invasive cholesterol level measurement method using eye image analysis. For this study, the eye images were acquired from the subjects with normal/abnormal cholesterol levels. Along with the eye images, the fasting serum cholesterol levels of the same subjects were documented. After data collection, the eye images in the study database were pre-processed and the cornel lipid deposition was segmented from the other regions of the eye image. After segmentation process, the Gray Level Mean (GLM) value of the segmented lipid deposition was computed. The Cholesterol Detection Regression (CDR) model was created using the GLM values of the lipid deposition in the input eye images and their corresponding serum cholesterol levels. In this paper, four CDR models were created to measure the total cholesterol and Low Density Lipoprotein (LDL) levels of the two age group people (40-60 years age group, 61 years and above age group). The goodness of fit of the CDR models was tested using the statistical measure R 2 and the cholesterol levels measured by the CDR models were compared with the lipid profile test results. The evaluation process indicates that the performance of the proposed method is better than the existing non-invasive cholesterol detection methods and, the total cholesterol and LDL levels measured using the proposed method is closer to the conventional lipid profile test.