Applied machine learning in hematopathology
Taher Dehkharghanian1,2, Youqing Mu2, Hamid R Tizhoosh3
1Department of Nephrology, University Health Network, Toronto, Ontario, Canada.
International Journal of Laboratory Hematology
|May 31, 2023
Summary
Machine learning (ML) is advancing digital pathology, particularly in hematopathology, by aiding diagnostic workflows. ML models are being developed to address unique challenges and support pathologists using bone marrow cytology and histopathology data.
Area of Science:
- Digital Pathology
- Hematopathology
- Machine Learning Applications
Background:
- Machine learning (ML) is increasingly applied to digital pathology, especially in hematopathology.
- These tools aim to support diagnostic workflows by analyzing diverse data, including digital tissue images.
- Hematopathology presents unique challenges for ML compared to other pathology subspecialties.
Purpose of the Study:
- To discuss current trends in machine learning for hematopathology.
- To review existing ML-enabled medical devices for hematopathology workflows.
- To explore research trends, focusing on bone marrow cytology and histopathology.
Main Methods:
- Review of current machine learning applications and research in hematopathology.
- Analysis of ML-enabled medical devices supporting diagnostic workflows.
- Focus on modeling pathologist workflows to address practical problems.
Main Results:
- Growing number of ML applications in digital hematopathology.
- Development of ML tools to support diagnostic decision-making.
- Identification of specific research trends in bone marrow analysis.
Conclusions:
- Machine learning holds significant potential to enhance hematopathology diagnostics.
- Addressing unique challenges in hematopathology is key for successful ML adoption.
- The transition to digital pathology facilitates the integration of new ML tools.
