Software provides non-invasive hemoglobin measurement with a smart phone

(Reuters Health) - An algorithm that computes hemoglobin levels from a smart phone image of the inner eyelid may help patients with anemia and other disorders avoid trips to healthcare centers for blood draws, a new study suggests.

The experimental app analyzes an eyelid image, breaking it down into its component wavelengths, then converting the resultant spectra into a hemoglobin level, according to the report published in Optica.

“Using this app, patients will be able to get a very accurate blood hemoglobin level,” said study coauthor Young Kim, an associate professor of biomedical engineering at Purdue University in West Lafayette, Indiana.

Kim and his colleagues were looking for a way to take advantage of the powerful smart phones now available.

“We have a great need for mobile health services now,” Kim said. “Any smart phone has lots of built in sensors. Let’s use some of those for some biological and physiological detection.”

The app is designed to be user-friendly. All a person needs to do is correctly position the phone’s camera and shoot. “In the current app, the user has to make sure that the eyelid is within a guided circle,” Kim said.

Kim and his colleagues selected the inner eyelid as the sensing site because the microvasculature there is quite visible. Another advantage: the inner eyelid’s spectra are not affected by skin color, which eliminates the need for any calibration for racial differences.

Once the picture is shot, the software breaks the image down to its spectral components. The app then compares the new information to what it learned from viewing a training set of images and hemoglobin levels collected from 138 volunteers who had been referred for conventional blood tests at the Moi University Teaching and Referral Hospital in Eldoret, Kenya. Of the 138 participants, 45 had cancer; 46, HIV; 8, tuberculosis; 19 sickle cell disease; 1, acute kidney failure; 4, heart failure; 1 malaria; 1 immune thrombocytopenic purpura; and 25 with no major disease.

Kim and his colleagues tested the app on a different group of 15 volunteers who also had been referred for conventional blood tests at the Kenyan hospital. When the researchers compared the hemoglobin levels calculated by the app to those actually measured from blood draws, they found the app was within 6.01% of the actual measurements. “Given the limited sample size, I would say 5%-10% though,” Kim said.

The new report is “pretty interesting,” said Dr. Girish Nadkarni, clinical director of the Hasso Plattner Institute for Digital Health at Mount Sinai, in New York City. “It’s cool to think that patients might not have to come to a center to have their blood drawn and then wait two days for results.”

“It’s a good proof of principle that blood hemoglobin can be measured non-invasively,” Dr. Nadkarni said. “That’s important. Especially now with remote care becoming more important. To the best of my knowledge this is the first study to do this in a scalable manner with readily available equipment.”

Still, Dr. Nadkarni said, the app needs to be evaluated in a larger population that is ethnically and racially diverse and who have a wide range of conditions.

SOURCE: Optica, online May 21, 2020.