The research findings of Dr. Hong-Rui Chen from our research team have been officially published in the international journal Digital Health. The paper, titled "Classification of hematologic malignancies from whole-slide bone marrow aspirates using a two-stage deep convolutional neural network pipeline," was officially published in April 2026.
This study focuses on the application of artificial intelligence in hematopathology, establishing a two-stage deep convolutional neural network analysis framework for whole-slide bone marrow aspirate images. The process begins with a quality assessment model to filter image regions suitable for interpretation, followed by a deep learning model to classify leukemia subtypes and lymphomas. The study included 1,022 slides, achieving a patient-level average accuracy of 95.2% and an AUC of 0.993. The DenseNet121 model demonstrated the highest balanced accuracy at 97.6% and achieved accuracies of 83.8% and 86.7% on external datasets, respectively. These research results demonstrate the research potential of interdisciplinary integration between clinical hematology, pathological imaging, and artificial intelligence, while providing new directions for computer-aided diagnosis of bone marrow smears and the optimization of clinical workflows.
Dr. Hong-Rui Chen demonstrated exceptional performance throughout the research process, contributing significantly to model development, data analysis, and manuscript writing. We extend our sincere congratulations and express our gratitude to all members of the research team for their collective dedication and support. We look forward to Dr. Chen's continued excellence and further achievements in his clinical and academic career.
Reference: PMID 42052410; DOI 10.1177/20552076261444599.