In their study "Deep learning detects acute myeloid leukemia and predicts NPM1 mutation status from bone marrow smears", published in Leukemia, the researchers showed how AI can support the initial examination of so-called cytomorphology in AML. The study analyzed 1251 AML patients and 236 healthy control samples. A multi-stage machine learning workflow was developed to segment and classify digital bone marrow images. This enabled the detection of AML cases and the accurate prediction of NPM1 mutation status, which occurs in about one third of cases.
"The results demonstrate the potential of deep learning to derive morphological characteristics that can predict mutation status," says Dr. Middeke. Future research will focus on other clinically relevant mutations. This will require close collaboration between medical professionals and software developers to develop AI models that are both accurate and generalizable.