BM Smear AI 1.0

Sample

Class	Types of cell
E	proerythroblast, basophilic erythroblast, polychromatophilic erythroblast, orthochromatic erythroblast
E-AB	proerythroblast-AB, basophilic erythroblast-AB, polychromatophilic erythroblast-AB, orthochromatic erythroblast-AB
Blast	blast
M	promyelocyte, myelocyte, metamyelocyte
N	band neutrophil, segmented neutrophil
N-AB	band neutrophil-AB, segmented neutrophil-AB
Eosino	immature eosinophil, eosinophil
Baso	immature basophil, basophil
Mono	promonocyte, monocyte
Lymph	lymphocyte
E-M	erythroblast mitosis
Mega	megakaryocyte
Mega-AB	megakaryocyte-AB
Unknown	unknown

Target Users

• Researchers: Engaged in blood cell morphology, classification, and related studies.
• Research Institutions: Conducting hematology or pathology research seeking automated tools for cell classification.
• Biomedical Research Labs: Investigating automated solutions to analyze bone marrow and other blood-related samples.
• Educational Institutions: For teaching and research purposes in the field of cell biology and morphology.

Key Features

1. AI-Powered Cell Classification: Automatically detects and classifies bone marrow cells using advanced deep learning models trained on images captured by CellaVision DC-1.

2. Comprehensive Cell Analysis: Identifies a wide range of bone marrow cells, including erythroblasts, neutrophils, blasts, and megakaryocytes.

3. Accurate and Reliable: Achieves high precision and recall, with an overall F1 score of 0.755, ensuring consistency in research applications.

4. Handles Complex Cases: Capable of identifying atypical cells, providing robust support in the study of morphological variations.

5. User-Friendly Interface: Web-based platform for seamless image uploading and real-time results display.

Benefits

• Time-Saving: Automates the laborious process of manual cell classification, accelerating research workflows.
• Improved Accuracy: Provides consistent, AI-driven classifications, reducing variability and potential errors in cell identification.
• Supports Advanced Research: Facilitates the study of complex cell morphology, even in cases involving subtle abnormalities.
• Accessible and Versatile: Suitable for use by a wide range of researchers, from professionals to students in academic settings.

How It Works

1. Image Input: Users upload bone marrow smear images captured via CellaVision DC-1 to the platform.

2. AI Processing: The system applies Faster R-CNN deep learning algorithms to detect and classify cells within the images.

3. Classification Output: The tool classifies cells into various categories, such as erythroblasts, neutrophils, and blasts, and identifies any atypical cells present.

4. Real-Time Results: Immediate classification results are provided, along with performance metrics like precision and recall.

5. Support for Research: The platform enables detailed analysis, supporting research in blood cell morphology and related fields.

Publications

Under preparation.

Acknowledgement

This app has been jointly developed with the Department of Laboratory Medicine, Kyoto University Hospital, and the Department of Hematology and Oncology, Graduate School of Medicine, Kyoto University.

Important Note:

BM Cell Classification 1.0 is a Research Use Only (RUO) product and is not intended for clinical diagnostics or medical use. It is designed to support research in the areas of hematology and cell morphology.