Image analysis-based blood smear leukocyte differential count method and system

CN122265994APending Publication Date: 2026-06-23CHENGBEI COMMUNITY HEALTH SERVICE CENTER PANLIAN TOWN MIYI COUNTY
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
CHENGBEI COMMUNITY HEALTH SERVICE CENTER PANLIAN TOWN MIYI COUNTY
Filing Date
2026-03-28
Publication Date
2026-06-23

AI Technical Summary

Technical Problem

Existing blood smear white blood cell classification and counting methods lack robustness in image preprocessing, cell segmentation, feature extraction, and anomaly identification, making it difficult to meet the high-efficiency and high-quality testing needs of primary healthcare institutions.

Method used

An image preprocessing strategy employing adaptive background correction and multi-channel color normalization is combined with a deep learning-based instance segmentation network for cell detection and segmentation. This constructs an intelligent detection architecture for joint analysis of multi-dimensional features, including a multi-stage cascaded classifier and a human-machine collaboration mechanism.

Benefits of technology

It significantly improves the robustness of image preprocessing, achieves accurate extraction and classification of single cells, meets the needs of abnormal cell identification, and ensures the reliability and efficiency of test results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122265994A_ABST
    Figure CN122265994A_ABST
Patent Text Reader

Abstract

The application discloses a blood smear white blood cell classification counting method and system based on image analysis, and belongs to the technical field of medical image processing. The method comprises the following steps: after collecting a peripheral blood smear image, performing adaptive background correction and color standardization pretreatment; detecting and segmenting each white blood cell through an instance segmentation network; extracting three-dimensional features of nuclear morphology, cytoplasmic staining and nuclear-cytoplasmic ratio and adaptively fusing; adopting a cascade classifier to perform five-class white blood cell classification and abnormal morphology identification; and counting and labeling suspected abnormal cells for review. The classification coincidence rate of the application reaches more than 95%, can effectively identify abnormal morphologies such as atypical lymphocytes, immature cells and toxic granules, and provides technical support for blood tests at the grassroots level.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • CN105205804A