Patient-Level Diagnosis of Acute Myeloid Leukemia via Deep Learning Analysis of Bone Marrow Smear

2026-06-09

AI Summary

This study addresses the low efficiency of manual review of bone marrow smears in acute myeloid leukemia (AML) and the challenge of effectively aggregating cell-level observations into patient-level diagnoses. To overcome these limitations, the authors propose an end-to-end deep learning framework that introduces expert-defined Composite Blast-like Cells (CBLC). The approach integrates YOLO-based cell detection, an EfficientNet-B0 classifier, and a two-stage training strategy, enhanced by class imbalance correction, boundary regularization, and morphological supervision to enable automated diagnosis from individual cell identification to patient-level CBLC proportion estimation. Evaluated on three external multicenter validation sets, the model achieves weighted F1 scores of 0.9076, 0.8696, and 0.9124, demonstrating strong cross-center generalizability.

Abstract

Bone marrow smear review remains important for acute myeloid leukemia (AML) assessment, but manual single-cell interpretation is labor-intensive and patient-level diagnosis requires aggregation of many cellular observations. We present a cell-to-patient deep learning pipeline for AML-assisted diagnosis from bone marrow smear images. The study included 258 patients from six anonymized centers, including a main cohort of 169 patients from Centers 1-3 and an external validation cohort of 89 patients from Centers 4-6. A 16-category cell annotation vocabulary was used to describe the global cellular composition, including granulocytic, monocytic, erythroid, lymphoid, eosinophilic, and other cells. Rather than identifying strict AML blasts or leukemic blasts, the model targets an expert-defined composite category termed Composite Blast-like Cells (CBLC), comprising N, N1, M, M1, R, R1, J, and J1 according to the project-wide morphological standard. A fixed YOLO-based segmentation module detected cells, predicted contours were matched to expert polygon annotations by contour IoU, and standardized single-cell crops were generated. An EfficientNet-B0 classifier was trained through a two-stage GT-to-YOLO and YOLO-to-YOLO strategy with class-imbalance correction, center-border regularization, and morphology-assisted supervision. Cell-level predictions were aggregated into patient-level CBLC ratios for AML-oriented diagnostic support. The pipeline achieved stable internal validation and maintained external generalization, with ensemble weighted F1-scores of 0.9076, 0.8696, and 0.9124 on Centers 4, 5, and 6, respectively.

Problem

• Acute Myeloid Leukemia
• Bone Marrow Smear
• Patient-Level Diagnosis
• Cell-to-Patient Aggregation
• Composite Blast-like Cells

Innovation

• Composite Blast-like Cells
• cell-to-patient pipeline
• YOLO-based segmentation
• morphology-assisted supervision
• multi-center generalization