Real-time non-invasive hemoglobin prediction using deep learning-enabled smartphone imaging

被引:0
作者
Chen, Yuwen [1 ]
Hu, Xiaoyan [2 ]
Zhu, Yiziting [2 ]
Liu, Xiang [2 ]
Yi, Bin [2 ]
机构
[1] Chinese Acad Sci, Chongqing Inst Green & Intelligent Technol, Chongqing 400714, Peoples R China
[2] Third Mil Med Univ, Southwest Hosp, Dept Anaesthesiol, Chongqing 400038, Peoples R China
基金
国家重点研发计划; 美国国家科学基金会; 中国国家自然科学基金;
关键词
Non-invasive prediction; Smartphone; Hemoglobin; Deep learning; Automatic;
D O I
10.1186/s12911-024-02585-1
中图分类号
R-058 [];
学科分类号
摘要
BackgroundAccurate measurement of hemoglobin concentration is essential for various medical scenarios, including preoperative evaluations and determining blood loss. Traditional invasive methods are inconvenient and not suitable for rapid, point-of-care testing. Moreover, current models, due to their complex parameters, are not well-suited for mobile medical settings, which limits the ability to conduct frequent and rapid testing. This study aims to introduce a novel, compact, and efficient system that leverages deep learning and smartphone technology to accurately estimate hemoglobin levels, thereby facilitating rapid and accessible medical assessments.MethodsThe study employed a smartphone application to capture images of the eye, which were subsequently analyzed by a deep neural network trained on data from invasive blood test data. Specifically, the EGE-Unet model was utilized for eyelid segmentation, while the DHA(C3AE) model was employed for hemoglobin level prediction. The performance of the EGE-Unet was evaluated using statistical metrics including mean intersection over union (MIOU), F1 Score, accuracy, specificity, and sensitivity. The DHA(C3AE) model's performance was assessed using mean absolute error (MAE), mean-square error (MSE), root mean square error (RMSE), and R<^>2.ResultsThe EGE-Unet model demonstrated robust performance in eyelid segmentation, achieving an MIOU of 0.78, an F1 Score of 0.87, an accuracy of 0.97, a specificity of 0.98, and a sensitivity of 0.86. The DHA(C3AE) model for hemoglobin level prediction yielded promising outcomes with an MAE of 1.34, an MSE of 2.85, an RMSE of 1.69, and an R<^>2 of 0.34. The overall size of the model is modest at 1.08 M, with a computational complexity of 0.12 FLOPs (G).ConclusionsThis system presents a groundbreaking approach that eliminates the need for supplementary devices, providing a cost-effective, swift, and accurate method for healthcare professionals to enhance treatment planning and improve patient care in perioperative environments. The proposed system has the potential to enable frequent and rapid testing of hemoglobin levels, which can be particularly beneficial in mobile medical settings.Trial RegistrationThe clinical trial was registered on the Chinese Clinical Trial Registry (No. ChiCTR2100044138) on 20/02/2021.
引用
收藏
页数:15
相关论文
共 42 条
  • [1] Evaluation of the use of non-invasive hemoglobin measurement in early childhood
    Arai, Yoshiteru
    Shoji, Hiromichi
    Awata, Kentaro
    Inage, Eisuke
    Ikuse, Tamaki
    Shimizu, Toshiaki
    [J]. PEDIATRIC RESEARCH, 2023, 93 (04) : 1036 - 1040
  • [2] Asare Justice Williams, 2023, Informatics in Medicine Unlocked, DOI 10.1016/j.imu.2023.101283
  • [3] Non-Invasive Hemoglobin Assessment with NIR Imaging of Blood Vessels in Transmittance Geometry: Monte Carlo and Experimental Evaluation
    Bardadin, Ilia
    Petrov, Vladimir
    Denisenko, Georgy
    Armaganov, Artashes
    Rubekina, Anna
    Kopytina, Daria
    Panov, Vladimir
    Shatalov, Petr
    Khoronenko, Victoria
    Shegai, Petr
    Kaprin, Andrey
    Shkoda, Andrey
    Yakimov, Boris
    [J]. PHOTONICS, 2024, 11 (01)
  • [4] DAA: A Delta Age AdaIN operation for age estimation via binary code transformer
    Chen, Ping
    Zhang, Xingpeng
    Li, Ye
    Tao, Ju
    Xiao, Bin
    Wang, Bing
    Jiang, Zongjie
    [J]. 2023 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2023, : 15836 - 15845
  • [5] Two-stage hemoglobin prediction based on prior causality
    Chen, Yuwen
    Zhong, Kunhua
    Zhu, Yiziting
    Sun, Qilong
    [J]. FRONTIERS IN PUBLIC HEALTH, 2022, 10
  • [6] Non-Invasive Detection of Anaemia Using Digital Photographs of the Conjunctiva
    Collings, Shaun
    Thompson, Oliver
    Hirst, Evan
    Goossens, Louise
    George, Anup
    Weinkove, Robert
    [J]. PLOS ONE, 2016, 11 (04):
  • [7] Non-invasive hemoglobin estimation from conjunctival images using deep learning
    Cuvadar, Beyza
    Yilmaz, Hakan
    [J]. MEDICAL ENGINEERING & PHYSICS, 2023, 120
  • [8] Predicting the level of anemia among Ethiopian pregnant women using homogeneous ensemble machine learning algorithm
    Dejene, Belayneh Endalamaw
    Abuhay, Tesfamariam M.
    Bogale, Dawit Shibabaw
    [J]. BMC MEDICAL INFORMATICS AND DECISION MAKING, 2022, 22 (01)
  • [9] Anemia detection through non-invasive analysis of lip mucosa images
    Donmez, Turker Berk
    Mansour, Mohammed
    Kutlu, Mustafa
    Freeman, Chris
    Mahmud, Shekhar
    [J]. FRONTIERS IN BIG DATA, 2023, 6
  • [10] Howard AG, 2017, Arxiv, DOI [arXiv:1704.04861, DOI 10.48550/ARXIV.1704.04861]