Ahuja Abhimanyu S, Paredes Iii Alfredo A, Eisel Mallory L S, Kodwani Sejal, Wagner Isabella V, Miller Darby D, Dorairaj Syril
Department of Ophthalmology, Casey Eye Institute, Oregon Health and Science University, Portland, OR, USA.
Charles E. Schmidt College of Medicine, Florida Atlantic University, Boca Raton, FL, USA.
Clin Ophthalmol. 2024 Oct 17;18:2969-2975. doi: 10.2147/OPTH.S489054. eCollection 2024.
Cataract surgery is one of the most performed procedures worldwide, and cataracts are rising in prevalence in our aging population. With the increasing utilization of artificial intelligence (AI) in the medical field, we aimed to understand the extent of present AI applications in ophthalmic microsurgery, specifically cataract surgery. We conducted a literature search on PubMed and Google Scholar using keywords related to the application of AI in cataract surgery and included relevant articles published since 2010 in our review. The literature search yielded information on AI mechanisms such as machine learning (ML), deep learning (DL), and convolutional neural networks (CNN) as they are being incorporated into pre-operative, intraoperative, and post-operative stages of cataract surgery. AI is currently integrated in the pre-operative stage of cataract surgery to calculate intraocular lens (IOL) power and diagnose cataracts with slit-lamp microscopy and retinal imaging. During the intraoperative stage, AI has been applied to risk calculation, tracking surgical workflow, multimodal imaging data analysis, and instrument location via the use of "smart instruments". AI is also involved in predicting post-operative complications, such as posterior capsular opacification and intraocular lens dislocation, and organizing follow-up patient care. Challenges such as limited imaging dataset availability, unstandardized deep learning analysis metrics, and lack of generalizability to novel datasets currently present obstacles to the enhanced application of AI in cataract surgery. Upon addressing these barriers in upcoming research, AI stands to improve cataract screening accessibility, junior physician training, and identification of surgical complications through future applications of AI in cataract surgery.
白内障手术是全球开展最为广泛的手术之一，并且在老龄化人口中白内障的患病率正在上升。随着人工智能（AI）在医学领域的应用日益增加，我们旨在了解当前AI在眼科显微手术，特别是白内障手术中的应用程度。我们在PubMed和谷歌学术上进行了文献检索，使用了与AI在白内障手术中的应用相关的关键词，并将2010年以来发表的相关文章纳入我们的综述。文献检索得出了关于AI机制的信息，如机器学习（ML）、深度学习（DL）和卷积神经网络（CNN），因为它们正被纳入白内障手术的术前、术中和术后阶段。目前，AI被整合到白内障手术的术前阶段，用于计算人工晶状体（IOL）的度数，并通过裂隙灯显微镜检查和视网膜成像来诊断白内障。在术中阶段，AI已被应用于风险计算、跟踪手术流程、多模态成像数据分析以及通过使用“智能器械”确定器械位置。AI还参与预测术后并发症，如后囊膜混浊和人工晶状体脱位，并组织患者的随访护理。诸如成像数据集可用性有限、深度学习分析指标不规范以及对新数据集缺乏通用性等挑战，目前阻碍了AI在白内障手术中的进一步应用。在未来的研究中解决这些障碍后，AI有望通过其在白内障手术中的未来应用，提高白内障筛查的可及性、初级医师培训水平以及手术并发症的识别能力。
