Computational Hematopathology: From Digital Morphology to Clinical Translation

Computational pathology and artificial intelligence have advanced rapidly across oncology, enabling quantitative characterization of tissue morphology and the development of image-based biomarkers for diagnosis, molecular profiling, prognosis, and treatment selection. However, the application of these approaches to hematologic malignancies remains comparatively underdeveloped relative to solid tumors.Hematopathology presents a distinct set of biological and technical challenges. Myeloid malignancies such as acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), and myeloproliferative neoplasms (MPNs) often require integrated assessment of bone marrow biopsies and aspirates, peripheral blood, molecular testing, cytogenetics, and clinical features. Similarly, lymphoid malignancies encompass substantial morphological and molecular heterogeneity across lymph nodes, bone marrow, blood, and extranodal tissues. These sources differ considerably in cellular composition, tissue architecture, spatial organization, and diagnostic considerations, creating challenges that may not be adequately addressed by computational workflows developed primarily for solid tumors.At the same time, recent advances in machine learning (ML), deep learning, and computational pathomics have demonstrated that digital pathology images contain subvisual information associated with molecular alterations and disease subtypes. The field is therefore reaching an important transition point: moving beyond proof-of-concept algorithm development toward reproducible, interpretable, and clinically actionable systems that can be validated across institutions and incorporated into tried-and-true hematopathology workflows.This Research Topic aims to establish a dedicated forum for computational pathomics in hematopathology. We seek to address how advanced machine learning algorithms can be translated into concrete diagnostic workflows and clinical trial stratification. Furthermore, this collection will examine whether computational paradigms developed for solid tumors are suitable for hematologic diseases, or whether liquid malignancies necessitate dedicated computational frameworks. By gathering computational scientists, hematopathologists, and clinical oncologists, this collection will define standards for validation, interpretability, and clinical adoption.We welcome contributions addressing methodological developments and translational applications across hematologic malignancies. Suggested themes include, but are not limited to:• Deep learning pipelines for whole slide imaging in bone marrow biopsies and aspirates• Prediction of genetic mutations, molecular subgroups, and biomarker status from morphological features• Methodological frameworks for bridging algorithmic benchmarks to bedside clinical decision making• Comparative studies evaluating solid tumor computational models versus hematology specific approaches• Multi center validation, stain normalization, and quality control standards in digital hematopathology• Multimodal integration combining digital pathology with flow cytometry, cytogenetics, and clinical trial datasetsWe welcome Original Research, Review, Mini Review, Perspective, and Brief Research Report articles.Please note that manuscripts consisting solely of bioinformatics or computational analysis of public databases without independent cohort or biological validation are out of scope. Topic editors: Spencer Krichevsky, Rajarsi Gupta, MD/PhD Source: https://www.frontiersin.org/research-topics/86423/computational-hematopathology-from-digital-morphology-to-clinical-translation