Non-Invasive Measurement of Hemoglobin Using Smartphone

Abstract

Anemia is a global public health issue with serious impact on health status and the productivity of human being. Currently following haemoglobin estimation methods are  invasive, expensive and requires a trained technical expert for collection and analysis of blood. Recently developed non invasive methods require either an external hardware component or additional light source for haemoglobin estimation. They are not validated with the original haemoglobin concentration through blood analysis. They are not portable. The present study aims to develop a non invasive method without an additional hardware than the existing smart phone. Digital images of eye, palm and nailbed were captured with smartphone and processed using MATLAB to estimate the concentration of hemoglobin.  The images were filtered using median filter with an algorithm and segmented. The features were extracted from the  images and used as an input data for ridge regression model. The model was validated with blood hemoglobin concentration. Confusion matrix was developed and the specificity and sensitivity of the proposed method was determined to be 80%. Accuracy can be further improved by sampling large number of blood samples with wide range of Hb concentration encompassing normal and anemic people of different age and gender.