Zhao Wenyi, Chen Ying, Yang Lei, Liang Chunyong, Wang Donghui, Wang Hongshui
Center for Health Science and Engineering, Hebei Key Laboratory of Biomaterials and Smart Theranostics, School of Health Sciences and Biomedical Engineering, Hebei University of Technology, Tianjin 300131, China.
Tianjin Key Laboratory of Materials Laminating Fabrication and Interface Control Technology, School of Materials Science and Engineering, Hebei University of Technology, Tianjin 300130, China.
ACS Biomater Sci Eng. 2025 Jun 9;11(6):3364-3375. doi: 10.1021/acsbiomaterials.5c00155. Epub 2025 May 28.
Bacterial infections have been demonstrated to cause the premature failure of implants. A reliable strategy for preserving biocompatibility is to physically modify the implant surface, without using chemicals, to prevent bacterial adhesion. This study employed femtosecond laser processing to generate various laser-induced periodic surface structures on Ti substrates. The antibacterial properties and osteoblast adhesion characteristics of these surfaces were investigated. Gene expression profiles and transcriptomic data were compared before and after laser treatment, and high-throughput analysis was conducted to evaluate the antibacterial performance related to different surface modifications. A small data set of Ti surface scanning electron microscopy images was compiled, and a deep learning model was trained using transfer learning to facilitate surface recognition and classification. The results demonstrated that femtosecond laser treatment disrupted bacterial adhesion and the expression of adhesion-related genes on the Ti surface, with the laser-treated samples at 5.6 W and 500 mm/s exhibiting an antibacterial efficacy exceeding 60%. In addition, the optimized deep learning model, ResNet50-TL, accurately identified and classified the structures of Ti surfaces post-treatment.
细菌感染已被证明会导致植入物过早失效。一种保持生物相容性的可靠策略是在不使用化学物质的情况下对植入物表面进行物理改性，以防止细菌粘附。本研究采用飞秒激光加工在钛基底上生成各种激光诱导的周期性表面结构。研究了这些表面的抗菌性能和成骨细胞粘附特性。比较了激光处理前后的基因表达谱和转录组数据，并进行了高通量分析以评估与不同表面改性相关的抗菌性能。编制了一个钛表面扫描电子显微镜图像的小数据集，并使用迁移学习训练了一个深度学习模型，以促进表面识别和分类。结果表明，飞秒激光处理破坏了钛表面的细菌粘附和粘附相关基因的表达，在5.6W和500mm/s的激光处理样品表现出超过60%的抗菌效果。此外，优化后的深度学习模型ResNet50-TL能够准确识别和分类处理后钛表面的结构。
