Deep Learning Enables Automated Detection of Mature B Cell Neoplasms by Flow Cytometry
Artificial intelligence now spots blood cancers in flow cytometry tests with 97.5% accuracy, helping pathologists work faster and more reliably.
A three-stage deep learning architecture evaluated on 3,070 clinical specimens from Memorial Sloan Kettering achieved 97.5% case-level accuracy (95.6% sensitivity, 98.9% specificity) for mature B-cell neoplasm detection by flow cytometry; lymphoma subtype prediction AUROC 0.925; interpretable outputs including marker expression comparisons verified by pathologists. This record was retained from the prior triage attempt for PubMed pipeline handoff.
What the study was
- Study design
- retrospective_validation_study
- Category
- hematologic_malignancies
- Maturity
- Validated
- Journal
- Modern Pathology
Why it surfaced
Large-scale clinical validation on 3,070 MSK specimens of an interpretable multi-stage DL architecture for B-cell neoplasm detection; excellent sensitivity/specificity and lymphoma subtype classification (AUROC 0.925). Ablation experiments demonstrate the contribution of engineered features. Clinically deployable with transparent outputs—a near-term implementation candidate for high-volume hematopathology labs.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.