AI-Based Portable White Blood Cells Classification and Counting System in POCT

作者全名：Zeng, Lanqing; Fu, Yusheng; Guo, Jiuchuan; Hu, Honghua; Li, Hongyu; Wang, Ning; Huang, Xiwei; Guo, Jinhong

作者地址：[Zeng, Lanqing; Fu, Yusheng; Li, Hongyu] Univ Elect Sci & Technol, Sch Informat & Commun Engn, Chengdu 611731, Peoples R China; [Hu, Honghua] Univ Elect Sci & Technol China, Sichuan Prov Peoples Hosp, Dept Lab Med, Chengdu 611731, Peoples R China; [Guo, Jiuchuan; Hu, Honghua] Univ Elect Sci & Technol China, Sichuan Prov Peoples Hosp, Sichuan Prov Key Lab Human Dis Gene Study, Chengdu 611731, Peoples R China; [Wang, Ning] Chongqing Med Univ, Shaoxing Keqiao Med Lab Technol Res Ctr, Chongqing 400016, Peoples R China; [Huang, Xiwei] Hangzhou Dianzi Univ, Sch Elect & Informat Engn, Hangzhou 310018, Peoples R China; [Guo, Jinhong] Shanghai Jiao Tong Univ, Sch Sensing Sci & Engn, Shanghai 200240, Peoples R China

通信作者：Huang, XW (通讯作者)，Hangzhou Dianzi Univ, Sch Elect & Informat Engn, Hangzhou 310018, Peoples R China.; Guo, JH (通讯作者)，Shanghai Jiao Tong Univ, Sch Sensing Sci & Engn, Shanghai 200240, Peoples R China.

来源：IEEE SENSORS JOURNAL

ESI学科分类：ENGINEERING

WOS号：WOS:001245592100014

JCR分区：Q1

影响因子：4.3

年份：2024

卷号：24

期号：7

开始页：11057

结束页：11068

文献类型：Article

关键词：Artificial intelligence (AI); classification; leukocyte; low-cost hardware; object detection

摘要：As an important part of the human immune system, the changes in the number of white blood cells (WBCs) play a significant role in the diagnosis of diseases, the monitoring of treatment effects, and the evaluation of immune system function. Microscopic blood cell detection is the standard procedure for WBC classification and counting, but it is expensive and lacks repeatability due to the high cost of high-magnification microscopes and the need for expert medical support. In this article, a novel artificial intelligence (AI)-based portable leukocyte intelligent detection system is proposed. The system is intended to provide patients with more convenient and rapid WBC detection services at a lower cost and to provide an intuitive and effective reference for primary healthcare workers in diagnosing and treating diseases. The system simplifies the complex operations required by traditional counting methods, and all functions are highly integrated. Users only need to place a test card containing the blood sample to be tested into the card slot and turn on the device to complete the test easily. The hardware implementation of the system uses small-sized devices uniformly, and multiperspective sampling is achieved through the design of a motion mechanism to ensure the reliability of statistical results. To test the performance of the instrument, a new dataset was collected for testing, and an AI model more suitable for WBC classification purposes was obtained by improving the loss function. The results show that the detection model carried by the system has an error rate of less than 1% in WBC counting and an error rate of about 2% in classification, and the single detection time is less than 20 s. Overall, the system is portable, fast, easy to use, and has good scalability, which helps to automate and standardize disease diagnosis in point-of-care testing (POCT).

基金机构：National Natural Science Foundation of China

基金资助正文：No Statement Available