Meier Raphael, Lux Paula, Med B, Jung Simon, Fischer Urs, Gralla Jan, Reyes Mauricio, Wiest Roland, McKinley Richard, Kaesmacher Johannes
Support Center for Advanced Neuroimaging-University Institute of Diagnostic and Interventional Neuroradiology (R. Meier, P.L., J.G., R.W., R. McKinley, J.K.), Department of Neurology (S.J., U.F., J.K.), Institute for Surgical Technology and Biomechanics (M.R.), and Institute for Diagnostic, Interventional and Pediatric Radiology (J.K.), University Hospital Inselspital and University of Bern, Freiburgstrasse 4, 3010 Bern, Switzerland.
Radiol Artif Intell. 2019 Sep 11;1(5):e190019. doi: 10.1148/ryai.2019190019. eCollection 2019 Sep.
To perform a proof-of-concept study to investigate the clinical utility of perfusion maps derived from convolutional neural networks (CNNs) for the workup of patients with acute ischemic stroke presenting with a large vessel occlusion.
Data on endovascularly treated patients with acute ischemic stroke ( = 151; median age, 68 years [interquartile range, 59-75 years]; 82 of 151 [54.3%] women) were retrospectively extracted from a single-center institutional prospective registry (between January 2011 and December 2015). Dynamic susceptibility perfusion imaging data were processed by applying a commercially available reference method and in parallel by a recently proposed CNN method to automatically infer time to maximum of the tissue residue function (Tmax) perfusion maps. The outputs were compared by using quantitative markers of tissue at risk derived from manual segmentations of perfusion lesions from two expert raters.
Strong correlations of lesion volumes (Tmax > 4 seconds, > 6 seconds, and > 8 seconds; = 0.865-0.914; < .001) and good spatial overlap of respective lesion segmentations (Dice coefficients, 0.70-0.85) between the CNN method and reference output were observed. Eligibility for late-window reperfusion treatment was feasible with use of the CNN method, with complete interrater agreement for the CNN method (Cohen κ = 1; < .001), although slight discrepancies compared with the reference-based output were observed (Cohen κ = 0.609-0.64; < .001). The CNN method tended to underestimate smaller lesion volumes, leading to a disagreement between the CNN and reference method in five of 45 patients (9%).
Compared with standard deconvolution-based processing of raw perfusion data, automatic CNN-derived Tmax perfusion maps can be applied to patients who have acute ischemic large vessel occlusion stroke, with similar clinical utility.© RSNA, 2019
进行一项概念验证研究，以探讨源自卷积神经网络（CNN）的灌注图在急性缺血性卒中伴大血管闭塞患者检查中的临床应用价值。
回顾性提取自单中心机构前瞻性登记处（2011年1月至2015年12月）的急性缺血性卒中血管内治疗患者的数据（n = 151；中位年龄68岁[四分位间距，59 - 75岁]；151例中有82例[54.3%]为女性）。动态磁敏感灌注成像数据采用一种商用参考方法进行处理，并同时采用一种最近提出的CNN方法自动推断组织残留函数（Tmax）灌注图的达峰时间。通过使用来自两名专家评估者对灌注病变手动分割得出的组织风险定量标记物来比较输出结果。
观察到CNN方法与参考输出之间病变体积具有强相关性（Tmax > 4秒、> 6秒和> 8秒；r = 0.865 - 0.914；P <.001），且各自病变分割具有良好的空间重叠性（Dice系数，0.70 - 0.85）。使用CNN方法进行晚期窗再灌注治疗的适用性是可行的，CNN方法的评估者间完全一致（Cohen κ = 1；P <.001），尽管与基于参考的输出相比存在轻微差异（Cohen κ = 0.609 - 0.64；P <.001）。CNN方法倾向于低估较小的病变体积，导致45例患者中有5例（9%）的CNN方法与参考方法之间存在分歧。
与基于标准去卷积的原始灌注数据处理相比，源自CNN的自动Tmax灌注图可应用于急性缺血性大血管闭塞性卒中患者，具有相似的临床应用价值。© RSNA, 2019
