Smartphone camera could help detect anaemia

Smartphone camera could help detect anaemia
A recent study published in the open-access journal PLOS ONE  has found that using a smartphone picture of a person’s inside the lower eyelid to predict anaemia is about 72% accurate.
Anaemia is typically diagnosed by testing for the blood count using sensitive lab equipment but there is a disproportionate occurrence of anaemia in rural settings where people have inadequate access to healthcare. According to the study authors, there is a need for an inexpensive, accessible, and noninvasive point-of-care tool capable of identifying anaemia. The ideal tool would use widely available technology.
Researchers conducted a two-phase study to assess the possibility of using a smartphone camera to aid in the detection of anaemia. The first phase involved taking images of the inner lower eyelids of 142 patients in an emergency department using a smartphone. The researchers selected the inner lower eyelid, called the palpebral conjunctiva. By zooming in on a small region in each photo, the researchers were able to develop an algorithm that maximizes colour resolution and a predictive model comparing the skin and whites of the eyes to haemoglobin levels.
The second phase involved testing the algorithm on smartphone images of 202 different patients in the emergency department. The findings showed that the model was 72.6% accurate in predicting anaemia. Its accuracy in predicting severe anaemia that would necessitate a blood transfusion was higher, at between 86% and 94.4%. Lead study author Dr Selim Suner, of Brown University and Rhode Island Hospital, explained that following a diagnosis of anaemia, people just need iron supplements, which are cheap and easy to take. “Making the diagnosis is the hard part,” Suner said.
Potential limitations that the researchers noted included variable image quality. However, this could have been due to the person retracting their eyelid during the recording of the image. Also, the lighting was not standardized, and it is unknown whether varying levels of brightness played a role in image quality.