CarbGeM Inc. Launches “BM Smear AI 1.0”, an AI-Powered Tool for Automatic Classification of Bone Marrow Smears

– Integrated with the CarbConnect® Cloud-Based AI Imaging Platform –
– Co-Developed with Kyoto University Hospital to Streamline Hematological Cell Classification –

CarbGeM Inc. (Headquarters: Shibuya, Tokyo; CEO: Masakazu Nakajima) has officially launched “BM Smear AI 1.0”, an AI-powered tool for the automatic classification of bone marrow smears, co-developed in collaboration with Kyoto University Hospital, Graduate School of Medicine, Kyoto University and NextGeM Inc. (Headquarters: Shibuya, Tokyo; President: Kiyoharu Hoshino).

This application integrates with the CarbConnect® cloud-based AI imaging platform and is designed not only to standardize diagnostic accuracy by minimizing variations caused by technicians’ individual skill levels but also to alleviate the workload of even highly experienced clinical laboratory technicians. Developed using deep learning algorithms and images captured via the CELLAVISION DC-1, the tool is available as a Research Use Only (RUO) product. Its primary purpose is to enhance the efficiency of hematological cell classification, including dysplasia, and to support hematology researchers in diagnosing conditions such as myelodysplastic syndromes.

About BM Smear AI 1.0

Development Background

The classification of bone marrow smear samples is an essential process in hematology and pathology research. However, traditional manual classification is time-consuming and its accuracy often depends on the experience and skills of the technician, posing challenges in standardizing diagnostic quality. BM Smear AI 1.0 leverages AI technology to streamline this process, enabling researchers to obtain more reliable and accurate data.

This application was co-developed with the technical support of NextGeM Inc., Kyoto University Hospital, and the Department of Hematology, Graduate School of Medicine, Kyoto University, under the leadership of Medicine Professor Akifumi Takaori (Director of Kyoto University Hospital) and Medicine Senior Lecturer Junya Kanda. Additionally, the tool was created as part of a project supported by the Kyoto University Incubation Program and has emerged as a groundbreaking research support solution in the field of hematological cell classification.

Key Features

  • AI-Powered Cell Classification
    Utilizes advanced deep learning algorithms to automatically detect and classify hematological cells. Supports the classification of a wide range of cells, including erythroblasts, neutrophils, blasts, and megakaryocytes.
  • High-Precision Performance
    Achieves an F1 score of 0.755, demonstrating high accuracy and consistency. It also supports the classification of dysplastic cells, making it a reliable tool for studying morphological changes.
  • Seamless Usability
    A web-based platform that allows users to upload images captured via the CELLAVISION DC-1 and instantly receive classification results in real time.
  • Handles Complex Cases
    Provides stable and accurate results even for challenging cases involving dysplastic cells, meeting diverse research needs.

For more information about BM Smear AI 1.0, please visit here.

Intended Use Cases

  • Support for Hematology and Pathology Research
    Ensures consistency in bone marrow smear classification, allowing researchers to focus on data analysis.
  • Educational Tool for Training
    Serves as a training tool for cell classification in educational institutions, supporting the skill development of students and novice researchers.
  • Application in Large-Scale Research Projects
    Accelerates the analysis of large volumes of samples, improving the efficiency of high-throughput projects.

Expected Benefits

  • Time Savings
    Automates the labor-intensive process of hematological cell classification, streamlining research workflows.
  • Improved Accuracy and Consistency
    AI-powered classification minimizes variability, providing reliable results.
  • Expanding Research Potential
    Supports the study of dysplastic cells and subtle morphological changes, enabling advanced research.
  • Broad Applicability
    Suitable for a wide range of users, from professionals to students and researchers in academic settings.

About CarbConnect®

CarbConnect® facilitates secure image storage, real-time communication, and seamless collaboration among experts, even in remote environments. By enabling image-driven workflows across PCs, smartphones, and tablets, the platform supports imaging applications in research, medical, and other professional fields.

For more details, please visit: CarbConnect Details
Register for CarbConnect® here: https://www.carb-connect.com/create-account

Future Outlook

Through the release of BM Smear AI 1.0, CarbGeM aims to improve the efficiency and accuracy of hematology and pathology research. By leveraging the CarbConnect® platform, the company is committed to developing next-generation analytical tools that enhance value in the medical and research fields.

About CarbGeM Inc.

CarbGeM Inc. is committed to applying its proprietary AI analysis technologies to combat bacterial infections through the convergence of biology and digital innovation. Through open innovation with leading domestic and international research institutions, the company aims to address global challenges, such as antimicrobial resistance.

Headquarters: 1-5-13 Jinnan, Shibuya, Tokyo

CEO: Masakazu Nakajima

Website: https://carbgem.com/en/

Go to News