Page 7 - 南京医科大学自然版
P. 7

第44卷第9期                           南京医科大学学报(自然科学版)
                  2024年9月                   Journal of Nanjing Medical University(Natural Sciences)     ·1179 ·


               ·基础研究·

                基于改进型U⁃Net的变色油墨血浆判别模型



                张瀚文 ,曹维娟 ,罗刚银 ,江 浩 ,邱                香 ,许    杰 ,史蓉蓉 ,郑       然  2
                                                                1
                                                                        2
                              2*
                                               1
                                                       2
                                       1
                      1
                中国科学院苏州生物医学工程技术研究所,先进体外诊断技术工程实验室,江苏                               苏州   215163;苏州市中心血站,江苏
                1                                                                             2
                苏州 215006
               [摘   要] 目的:为了解决因主观判别尺度不一和计算响应过长,在血浆制备过程中易出现疑似溶血血浆误判输出和疑似非
                溶血血浆医学报废的现象,给患者的生命安全带来极大隐患或产生浪费的问题。方法:研制一种基于深度学习和变色油墨理
                念的限界法。利用改进型U⁃Net网络进行图像分割,引入改进型注意力机制、批量归一化和填充模块来解决空间映射关系中
                存在的估计均值偏移、计算效率低和感受野视场不足的问题,并利用自采样本数据集对该模型进行验证对比。结果:采用变色
                油墨限界法进行分类判别,在确保血浆样本识别精度为前提的同时,提升了血浆判别的计算效率、降低了判别时间,实验结果
                评价准确率为99.52%。 结论:本研究模型的血浆判别精度优于其他判别模型,有望应用于临床。
               [关键词] 血浆;疑似溶血与疑似非溶血;U⁃Net;变色油墨;机器学习

               [中图分类号] TP18;TP391.41;R318         [文献标志码] A                    [文章编号] 1007⁃4368(2024)09⁃1179⁃11
                doi: 10.7655/NYDXBNSN231104



                A plasma discrimination model for color changing ink based on improved U⁃Net

                                                                                     1
                                                                    1
                                                         1
                                                                              2
                              1
                                           2*
                                                                                                   2
                ZHANG Hanwen ,CAO Weijuan ,LUO Gangyin ,JIANG Hao ,QIU Xiang ,XU Jie ,SHI Rongrong ,ZHENG Ran  2
                1 Suzhou Institute of Biomedical Engineering and Technology,Engineering Laboratory of Advanced In Vitro
                Diagnostic Technology,Chinese Academy of Sciences,Suzhou 215163;SuZhou Municipal Health Commission,
                                                                              2
                Suzhou 215006,China
               [Abstract] Objective: Due to inconsistent subjective assessment criteria and excessively long calculation responses,there is a high
                risk of mistakenly discarding suspected hemolytic plasma and inappropriately discarding suspected non ⁃ hemolytic plasma during
                plasma preparation,posing significant risks to patient safety and leading to waste. This study aims to resolue these problems. Methods:
                A thresholding method that integrates deep learning with color⁃changing ink concepts was developed. By employing an enhanced U⁃Net
                architecture for image segmentation,the study introduces an advanced attention mechanism,batch normalization,and a padding
                module to tackle issues such as mean estimation bias,computational inefficiencies,and limited receptive field sizes in spatial mapping
                relationships. The model was validated and compared using a self⁃collected sample dataset. Results:This study employed the color⁃
                changing ink boundary method for classification,enhancing the computational efficiency of plasma discrimination and reducing
                discrimination time,while ensuring the accuracy of plasma sample identification. The accuracy rate of the experimental results is
                99.52%. Conclusion:The results indicate that the plasma discrimination accuracy of this model is superior to other discrimination
                models,and it is expected to be applied in clinical practice.
               [Key words] plasma;suspected hemolysis and non suspected hemolysis;U⁃Net;color⁃changing ink;machine learning
                                                                            [J Nanjing Med Univ,2024,44(09):1179⁃1189]





                    血浆是血液中的液体成分,占全血容量的55%~                        物等。临床主要用于补充凝血因子。血浆以全血
                60%,携带细胞和其他物质如蛋白、凝血因子和化合                          为原料,用低温离心分离的方法制备。血液离心过
                                                                  程中会出现部分红细胞相互挤压、破裂所产生的血
               [基金项目] 苏州市卫生科技项目(GWZX202102)
                ∗                                                 红蛋白游离在血浆中,临床输注会引起严重的输血不
                通信作者(Corresponding author),E⁃mail: 20214221019@stu.su⁃
                                                                        [1]
                da.edu.cn                                         良反应 。目前,血浆疑似溶血的判别主要依赖人工
   2   3   4   5   6   7   8   9   10   11   12