A Critical Approach to the Diagnostic Work-Up of Suspected Myelodysplastic Neoplasms: Integrated Reports, Technical Pitfalls and the Role of Artificial Intelligence

The diagnosis of myelodysplastic neoplasms (MDS) traditionally relies on cytomorphological assessment of peripheral blood (PB) and bone marrow (BM). Despite recent shifts toward molecular testing and next-generation sequencing, blast percentage and dysplasia remain integral to the initial work-up. Morphology provides immediate availability and cost-effectiveness but is hindered by interobserver variability, lack of standardization and declining training in classical hematology. These factors may complicate the differentiation of MDS from non-clonal cytopenias or therapy-related changes. Technical quality of BM sampling and processing is crucial, as inadequate core length, improper fixation or artifacts may obscure interpretation. Digital morphology and artificial intelligence (AI) are reshaping hematology by reducing subjectivity, improving reproducibility and enabling remote consultation, training and archiving. Although automated analyzers demonstrate high accuracy for common cell types, expert oversight remains essential, especially for rare or pathological cells. AI and machine learning (ML) extend across diagnostic modalities, from detecting dysplasia in PB and BM smears to automating flow cytometry gating, karyotyping and variant analysis. These approaches can differentiate MDS from aplastic anemia or acute myeloid leukemia with high accuracy, although generalizability is limited by small datasets, technical variability and insufficient external validation. Ethical, regulatory and educational challenges persist, underscoring the need for explainable models and clinician AI literacy. PEER REVIEWED ARTICLE **Peer reviewers:** Prof. Dr Ulrich Germing, Department of Hematology, Oncology and Clinical Immunology, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany Prof. Dr Alexandar Tzankov, Institute of Medical Genetics and Pathology, University Hospital Basel, Basel, Switzerland Received on September 29, 2025; accepted after peer review on November 14, 2025; published online on November 25, 2025.