Point of care detection of anemia in non-invasive manner by using image processing and convolutional neural network with mobile devices

Abstract

A medical procedure is described as non-invasive if there is no necessity for incision into the skin. Invasive methods can often be risky because there is a high chance of infection. That is why the medical sector is constantly trying to shift its momentum into non-invasive methods. The traditional method of detecting Anemia is by testing blood samples taken from the patient’s body. Anemia is a medical condition when the human body suffers from a lack of RBC. RBC is the carrier of oxygen to the blood. If not checked regularly, this may lead to kidney failure or even premature births. The situation gets worse in the rural areas of developing nations, where medical equipment is often unavailable. This study proposes a non-invasive point of care solution using images captured from mobile phone videos. Image processing is a vastly used technique around the world. Mostly because it allows various algorithms to be applied and reduces the build-up of noise and deformation during image processing. Usually, an Anemic patient’s blood contains less hemoglobin. Anemic blood transmits more light compared to normal blood, so the severity of anemia can be measured by analyzing the color of blood. Using the flashlight, the capturing of blood color was made possible without drawing blood. Hence, this paper tried to analyze the sample using python image processing that will enable the chances of a low-cost point of care solution for the detection of Anemia.