As of now, the evaluation of bone marrow smears by medical experts, cytomorphology, is the gold-standard in diagnosing hematologic malignancies. Still, this procedure is prone to bias due to diagnostic varibilities between experts, a lack of standardization and overall the need for qualified personnel that requires years of specialized training in order to adequately perform the assessment.
For the patients, however, this process is of utmost importance as correct and timely diagnosis is the key prerequisite for treatment initiation.
Therefore, we currently develop a software prototype utilizing computer vision to automatically provide diagnosis from bone marrow smears. We trained our prototype on > 2000 samples of bone marrow smears from multiple centers in order to provide robust and generalizble models. So far, it accurately distinguishes between acute myeloid leukemia, acute promyelocytic leukemia, myelodysplastic syndromes and healthy controls. Instead of tedious manual cell counting by human experts that usually requires ~30 minutes for ~200 cells (with multiple counts being required per diagnosis), our prototype provides a diagnosis in ~30 seconds with accuracies >95%.