Related Experiment Video
Updated: Aug 8, 2025
12:05
Database-guided Flow-cytometry for Evaluation of Bone Marrow Myeloid Cell Maturation
Published on: November 3, 2018
11.7K
An AI-Aided Diagnostic Framework for Hematologic Neoplasms Based on Morphologic Features and Medical Expertise
Nan Li1, Liquan Fan1, Hang Xu2
1Shanghai Institute of Hematology, State Key Laboratory of Medical Genomics, National Research Center for Translational Medicine at Shanghai, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China.
Summary
We developed a virtual hematological morphologist (VHM) using artificial intelligence to aid in diagnosing blood cancers. This AI tool improves diagnostic accuracy and efficiency compared to traditional methods.
Area of Science:
- Hematology
- Artificial Intelligence
- Medical Diagnostics
Background:
- Manual morphologic examination is crucial but time-consuming for diagnosing hematological diseases.
- Current diagnostic methods lack efficiency and can be laborious.
- There is a need for advanced tools to support hematological diagnoses.
Purpose of the Study:
- To develop an artificial intelligence (AI)-aided diagnostic framework, termed Virtual Hematological Morphologist (VHM).
- To integrate medical expertise with AI for diagnosing hematological neoplasms.
- To establish a reliable and interpretable hematological diagnostic tool.
Main Methods:
- Utilized a Faster Region-based Convolutional Neural Network for image-based morphologic feature extraction.
- Employed a support vector machine algorithm for feature-based case identification using diagnostic criteria.
- Integrated these models into a two-stage AI-aided diagnostic framework (VHM).
Main Results:
- VHM achieved 94.65% recall and 93.95% precision in bone marrow cell classification.
- Demonstrated high accuracy in differential diagnosis (balanced accuracy: 97.16%) and specific diagnoses (e.g., CML: 99.23%).
- Outperformed end-to-end AI frameworks in testing accuracy and generalization ability.
Conclusions:
- The VHM framework offers a comprehensive AI-aided approach to morphologic diagnosis in hematology.
- VHM's knowledge-based, multi-stage strategy provides superior performance and interpretability.
- This AI tool mimics clinical diagnostic logic, enhancing reliability and efficiency in hematological disease diagnosis.
