背景:影像组学在各种眼科疾病的医学图像分析中具有巨大的潜力。最近,在这一研究领域有许多努力。本系统综述旨在全面评估影像组学在眼科中的优势和局限性。
方法:符合系统评价和荟萃分析(PRISMA)指南的首选报告项目,我们对预先注册的方案(PROSPERO:CRD42023446317)进行了系统评价.我们探索了PubMed,Embase,和Cochrane数据库对该主题进行了原始研究,并进行了全面的描述性整合。此外,纳入的研究通过影像组学质量评分(RQS)进行质量评估.
结果:最终从227项研究的初始搜索中选择了41篇文章进行进一步分析。这些文章包括五个疾病类别的研究,涵盖了七个成像模式。影像组学模型表现出强大的性能,曲线下面积(AUC)值大多在0.7-1.0范围内。中度RQS(平均得分为11.17/36)表明大多数研究是回顾性的,无需外部验证的单中心分析。
结论:影像组学在眼科领域具有广阔的应用前景,协助诊断,早期筛查,和治疗反应的预测。人工智能算法为眼科影像组学模型的构建做出了重要贡献。这项研究强调了影像组学在眼科领域的优势和挑战,并提出了未来改进的潜在途径。
结论:影像组学代表了一种产生创新成像标记的有价值的方法,提高临床诊断和治疗效率,并在许多眼科疾病的临床背景下帮助决策,从而改善患者的整体预后。
结论:影像组学在眼科领域引起了广泛的关注。文章包括七种成像模式的五种疾病类别,始终产生的AUC大多高于0.7。目前的研究很少有前瞻性和多中心的研究,强调未来高质量研究的必要性。
BACKGROUND: Radiomics holds great potential in medical image analysis for various ophthalmic diseases. In recent times, there have been numerous endeavors in this area of research. This systematic review aims to provide a comprehensive assessment of the strengths and limitations of radiomics in ophthalmology.
METHODS: Conforming to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines, we conducted a systematic review with a pre-registered protocol (PROSPERO: CRD42023446317). We explored the PubMed, Embase, and Cochrane databases for original studies on this topic and made a comprehensive descriptive integration. Furthermore, the included studies underwent quality assessment by the radiomics quality score (RQS).
RESULTS: A total of 41 articles from an initial search of 227 studies were finally selected for further analysis. These articles included research across five disease categories and covered seven imaging modalities. The radiomics models demonstrated robust performance, with area under the curve (AUC) values mostly falling within 0.7-1.0. The moderate RQS (mean score: 11.17/36) indicated that most studies were retrospectively, single-center analyses without external validation.
CONCLUSIONS: Radiomics holds promising utility in the field of ophthalmology, assisting diagnosis, early-stage screening, and prognostication of treatment response. Artificial intelligence algorithms significantly contribute to the construction of radiomics models in ophthalmology. This study highlights the strengths and challenges of radiomics in ophthalmology and suggests potential avenues for future improvement.
CONCLUSIONS: Radiomics represents a valuable approach for generating innovative imaging markers, enhancing efficiency in clinical diagnosis and treatment, and aiding decision-making in clinical contexts of many ophthalmic diseases, thereby improving overall patient prognosis.
CONCLUSIONS: Radiomics has attracted extensive attention in the field of ophthalmology. Articles included five disease categories over seven imaging modalities, consistently yielding AUCs mostly above 0.7. Current research has few prospective and multi-center studies, underlining the necessity for future high-quality studies.