Development of a non-invasive evaluation methodology using a smartphone attachment on the diagnosis and follow-up of anemia

Anemia, which has multiple causes, such as iron deficiency, chronic blood loss, and hemolysis, is a prevalent health problem affecting an estimated two billion people or 30% of the world's population. The ability to measure hemoglobin concentration in anemic patients noninvasively and continuously has significant potential to facilitate hemoglobin monitoring, improve the detection of acute anemia, and avoid complications, expenses, and pain associated with invasive blood draws. Although some non-invasive approaches have been developed, they are less accurate than invasive methods as a clinical purpose, resulting in an unmet need for more accurate non-invasive methods. In the present study, we conducted a feasibility study for a group of subjects who were admitted to Koc University Hospital, Hematology clinics and Lokman Hekim University Hematology and Internal Diseases clinics. Patients with the symptoms of anemia and healthy subjects who were scheduled to have their complete blood count (CBC) parameters measured with hematology blood analyzers as part of their clinical care were recruited for this study. After the patient's blood was collected to conduct their CBCs, high-resolution images were taken of those patients' palm finger areas with a smartphone camera via dermatoscope attachments. By examining the color characteristics of the images, we developed an image processing methodology to correlate color characteristics in smartphone images of palm to hemoglobin concentration measured with the gold standard hemoglobin measurement system which is CBC the test. According to the results, our method gives significant predicted Hgb values. We observed that our results indicate a correlation between erythema index values and CBC hemoglobin levels. The main contribution of this thesis is proving the linear relationship between color characteristics of digital images and hemoglobin concentration measured in the laboratory. Without any reflectance measurement, this non-invasive calculation methodology is useful for objective assessment of erythema index and therefore hemoglobin levels. With the proposed methodology, in future cost-effective non- invasive point-of-care system that enables users to make self-care anemia evaluation and screening, which would replace the conventional method of anemia detection by visual observation of palm pallor colors Furthermore, with the developments in cloud technology, the study results can be improved by increasing the number of samples. This can be done with a system that will allow people who have already had blood counts to add photos to the cloud. Sensitivity can be optimized with a deep learning algorithm, which will be developed based on the correlation found, and values that are much closer to the real hemoglobin results can be obtained. The outcomes of this project can improve health services offered to patients with related diseases / disorders by supporting diagnosis, and early identification of critical situations related to disease severity and therefore decrease the percentage of severe complications of the diseases. These, in turn, are expected to yield significant economic benefits to patients by better planning treatment and avoiding unnecessary treatment schemas. Apart from savings on patient expenses, these savings also encourage health care providers to deliver high- quality, coordinated care at lower costs.