Potential of thermal imaging as a tool for prediction of cardiovascular disease

• Main Authors: Jayanthi Thiruvengadam, M Anburajan, M Menaka, B Venkatraman
• Format: Article
• Language: English
• Published: Wolters Kluwer Medknow Publications 2014-01-01
• Series: Journal of Medical Physics
• Subjects: Cardiovascular disease; framingham score; infrared imaging; lipid profile; thermal imaging; vascular dysfunction
• Online Access: http://www.jmp.org.in/article.asp?issn=0971-6203;year=2014;volume=39;issue=2;spage=98;epage=105;aulast=Thiruvengadam
• ISSN: 0971-6203; 1998-3913

Vascular dysfunction is associated with onset of cardiovascular disease (CVD). Its effect is reflected as temperature change on the skin. The aim of this work was to test the potential of thermal imaging as cost effective screening tool for prediction of CVD. Thermal imaging of various parts of the subject (N = 80, male/female =44/36, aged 25-75 years) was done using noncontact infrared (IR) camera. In each subject, total cholesterol (TC; mg/dl) and high-density lipoprotein (HDL, mg/dl) were measured according to standard biochemical analysis. Based on National Cholesterol Education Program ATP III criteria, subject with known CVD (N = 16) and age- and sex- matched normal subjects (N = 21) were included in the study. The average surface temperature of various parts from head to toe was calculated and statistical analysis was performed between the groups. In the total population (N = 37), correlation study shows TC (mg/dl) was correlated with measured surface temperature of the following regions: Temporal left (r = −0.316) and right (r = −0.417), neck left (r = 0.347) and right (r = −0.410), and hand left (r = 0.387). HDL (mg/dl) was found to be correlated with measured surface temperature of the following regions: Temporal left (r = 0.445) and right (r = 0.458), hand left (r = −0.470), and foot anterior left (r = −0.332) and right (r = −0.336). Temperature asymmetry was more significant in upper extremity in CVD group. Using the surface temperature, regression models were calculated for noninvasive estimation of TC and HDL. The predictive ability of measured surface temperature for TC and HDL was 60%. The model for noninvasive estimation gave sensitivity and specificity value of 79 and 83% for TC and 78 and 81% for HDL, respectively. Thus, the surface temperature can be one of the screening tools for prediction of CVD. The limitation of the present study is also discussed under future work.