Automated Detection of Vascular Leakage in Fluorescein Angiography - A Proof of Concept.
Automated Detection of Vascular Leakage in Fluorescein Angiography - A Proof of Concept.
复制标题
DOI：
10.1167/tvst.11.7.19
复制
发表时间：
2022-07-08
影响因子：
3
通讯作者：
Sen, H. Nida
中科院分区：
文献类型：
--
作者：
Young, LeAnne H.;Kim, Jongwoo;Yakin, Mehmet;Lin, Henry;Dao, David T.;Kodati, Shilpa;Sharma, Sumit;Lee, Aaron Y.;Lee, Cecilia S.;Sen, H. Nida
本文的目的是开发一种深度学习算法来检测葡萄膜炎患者荧光素血管造影（FA）中的视网膜血管渗漏（渗漏），并使用经过训练的算法来确定临床上显着的渗漏变化。算法经过训练和测试，可在一组 200 张 FA 图像（61 名患者）上检测泄漏，并在单独的 50 幅图像测试集（21 名患者）上进行评估。基本事实是两名临床医生进行的泄漏分割。 Dice 相似系数 (DSC) 用于测量一致性。在训练期间，该算法实现了 0.572 的最佳平均 DSC（95% 置信区间 [CI] = 0.548–0.596）。在另外一组 50 张图像上进行测试时，经过训练的算法获得了 0.563 的 DSC（95% CI = 0.543–0.582）。然后使用经过训练的算法来检测纵向患者就诊中的成对 FA 图像的泄漏。纵向渗漏随访显示，渗漏覆盖的可见视网膜区域（由算法检测到）发生 >2.21% 的变化，与临床专家评估的金标准相比，检测临床显着变化的灵敏度和特异性为 90%（曲线下面积 [AUC] = 0.95）。与真实情况相比，这种深度学习算法在识别血管渗漏方面表现出适度的一致性，但能够帮助识别血管 FA 渗漏随时间的变化。这是一项概念验证研究，表明可以以更标准化的方式检测血管渗漏，并且可以开发工具来帮助临床医生更客观地比较 FA 之间的血管渗漏。
The purpose of this paper was to develop a deep learning algorithm to detect retinal vascular leakage (leakage) in fluorescein angiography (FA) of patients with uveitis and use the trained algorithm to determine clinically notable leakage changes. An algorithm was trained and tested to detect leakage on a set of 200 FA images (61 patients) and evaluated on a separate 50-image test set (21 patients). The ground truth was leakage segmentation by two clinicians. The Dice Similarity Coefficient (DSC) was used to measure concordance. During training, the algorithm achieved a best average DSC of 0.572 (95% confidence interval [CI] = 0.548–0.596). The trained algorithm achieved a DSC of 0.563 (95% CI = 0.543–0.582) when tested on an additional set of 50 images. The trained algorithm was then used to detect leakage on pairs of FA images from longitudinal patient visits. Longitudinal leakage follow-up showed a >2.21% change in the visible retina area covered by leakage (as detected by the algorithm) had a sensitivity and specificity of 90% (area under the curve [AUC] = 0.95) of detecting a clinically notable change compared to the gold standard, an expert clinician's assessment. This deep learning algorithm showed modest concordance in identifying vascular leakage compared to ground truth but was able to aid in identifying vascular FA leakage changes over time. This is a proof-of-concept study that vascular leakage can be detected in a more standardized way and that tools can be developed to help clinicians more objectively compare vascular leakage between FAs.
登录
查看更多内容
DOI：
10.1097/00006982-200212000-00001
发表时间：
2002-12-01
影响因子：
3.3
作者：
Kaiser, RS;Berger, JW;Margherio, RM
通讯作者：
Margherio, RM
影响因子：
5.7
作者：
Shattuck DW;Prasad G;Mirza M;Narr KL;Toga AW
通讯作者：
Toga AW
影响因子：
6.1
作者：
Souza, Johnatan Carvalho;Bandeira Diniz, Joao Otavio;de Paiva, Anselmo Cardoso
通讯作者：
de Paiva, Anselmo Cardoso
影响因子：
4.4
作者：
Rabbani, Hossein;Allingham, Michael J.;Farsiu, Sina
通讯作者：
Farsiu, Sina
影响因子：
4.4
作者：
Park, Jongha;Yun, Jihye;Seo, Joon Beom
通讯作者：
Seo, Joon Beom