MYELODYSPLASTIC syndrome (MDS) is a disorder that disrupts the maturation and differentiation of haematopoietic stem cells in the bone marrow. The annual global incidence of MDS, which can progress to acute leukaemia, is currently estimated at four cases per 100,000 individuals.
Traditionally, bone marrow samples are analysed in order to identify genetic and cytogenetic aberrations in bone marrow cells and therefore diagnose MDS. Now, academics from the University of Helsinki, Helsinki, Finland, have successfully employed convolutional neural networks to extractmy features from biopsies, which are otherwise difficult to discern.
Tissue samples were initially stained with haematoxylin and eosin, before being digitised and evaluated with the aid of machine learning models. The use of computational deep learning algorithms was shown to facilitate the accurate identification of the most common genetic mutations influencing the development of the disease. The reliability of the results generated was found to increase as a function of the number of aberrant cells in the samples.
Senior author Prof Satu Mustjoki highlighted the clinical importance of the research findings: “The study confirms that computational analysis helps to identify features that elude the human eye. Moreover, data analysis helps to collect quantitative data on cellular changes and their relevance to the patient’s prognosis.”
The use of the Helsinki University Hospital data lake environment optimised the collection and analysis of extensive datasets. “We’ve developed solutions to structure and analyse data stored in the HUS [Helsinki University Hospital] data lake. Image analysis helps us analyse large quantities of biopsies and rapidly produce diverse information on disease progression. The techniques developed in this project are suited to other projects as well, and they are perfect examples of the digitalising medical science,” stated Oscar Brück, leading author and doctoral student.
Overall, this research offers new insights into the pathobiology of MDS and, going forward, provides an indication of the increased role artificial intelligence will play in the assessment and diagnosis of malignant haematological disorders.