光声辅助超声识别的智慧医疗综合实验设计  

Comprehensive Experimental Design for Smart Healthcare of Photoacoustic Aided Ultrasound Recognition

作　　者：张建[1] 尹文杰 梁超豪 傅洪波[1] ZHANG Jian;YIN Wenjie;LIANG Chaohao;FU Hongbo(School of Biomedical Engineering,Guangzhou Medical University,Guangzhou 511436,China)

机构地区：[1]广州医科大学生物医学工程学院,广州511436

出　　处：《实验室研究与探索》2025年第2期173-177,共5页Research and Exploration In Laboratory

基　　金：国家重点研发计划课题(2022YFC2304205);广州国家实验室专项课题(GZNL2023A03002);广州市高等教育教学质量与教学改革工程项目(2024YLKC012)。

摘　　要：针对当前超声导引下的注射针穿刺图像对比度不高亟待改善的问题,设计了基于深度学习的穿刺引导综合性实验。该实验将注射针的超声图像校正过程细分为每1帧的图像识别任务,以光声图像为事实基准建立数据库,训练构建好的神经网络模型,并应用在穿刺超声引导的实施过程中。实施结果证明了该综合实验在教学中的可行性和可推广性。通过实验教学后,学生能够融会深度学习、光声成像和超声成像等多领域知识,培养了学生的实际操作能力和实验技能。Aiming at the current ultrasound guided puncture images of injection needles with low contrast that need to be improved urgently,a comprehensive deep learning-based puncture guidance experiment is designed.The experiment breaks down the injection needle correction process into injection needle segmentation tasks for each frame,builds a database with photoacoustic images as a factual baseline,and trains the constructed neural network model.It is applied to the implementation of puncture ultrasound guidance.Implementation results demonstrate the feasibility and generalizability of integrated experiments in teaching.Through experimental teaching,students are able to integrate multidisciplinary knowledge such as deep learning,photoacoustic imaging and ultrasound imaging,their practical ability and experimental skills are improved.

关 键 词：深度学习 光声成像 超声成像 实验教学 智能健康护理 

分 类 号：R318.51[医药卫生—生物医学工程]