RGB pixel analysis of fingertip video image captured from sickle cell patient with low and high level of hemoglobin

The demand for medical image processing is ever growing, especially for medical device manufacturers, researchers, and innovators. In this article, we present the image processing of a fingertip video to investigate the relationship between image pixel information and different hemoglobin (Hb) levels. We use the smartphone camera to record the fingertip videos of different sickle cell patients. We also collect their clinical Hb records. We extract the red, green and blue (RGB) pixel of the video image and make the histogram of selected frames for each video. The averaged histogram values of those selected frames are used as an input feature matrix in the regression analysis. Linear regression as well as the partial least squares (PLS) algorithm is applied to the input feature matrix. We consider five sickle cell patients who received the blood transfusion. We analyze the thirty fingertip videos from five patients where each patient gave three videos at the same time. Fifteen fingertip videos are recorded before blood transfusion, and rest of the videos are captured after two weeks of their blood transfusion. Matlab tool is used for the data analysis and visual image presentation of the RGB image histogram values, masked RGB image, and the confusion matrix of this paper. The result generated from linear regression and the goodness of fit of PLS model shows the reliable performance of this research work.