Morphometric MRI as a diagnostic biomarker of frontotemporal dementia: A systematic review to determine clinical applicability.
Morphometric MRI as a diagnostic biomarker of frontotemporal dementia: A systematic review to determine clinical applicability.
复制标题
DOI：
10.1016/j.nicl.2018.08.028
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Ducharme S
中科院分区：
文献类型：
--
作者：
McCarthy J;Collins DL;Ducharme S
额颞叶痴呆（FTD）是很难诊断，由于其异质性和重叠的症状与原发性精神疾病。脑萎缩的MRI是一个关键的生物标志物，但在早期缺乏敏感性。基于MRI的形态测量和机器学习技术是提高诊断准确性的有前途的工具。我们的目的是回顾使用形态测量MRI对FTD进行分类的文献现状，并评估其在临床实践中的适用性。使用Pubmed和PsychInfo完成了一项研究检索，这些研究使用单个或联合方法，在个体水平上使用形态测量MRI指标对FTD受试者与非FTD受试者（对照或其他疾病）进行了分类。纳入了28篇相关文章，并按照PRISMA指南进行了系统性综述。根据纳入的FTD受试者类型及其分类组对研究进行分类。研究在受试者选择、MRI方法学和分类方法上差异很大，结果高度异质。总体而言，许多研究表明诊断准确性良好，在区分FTD与对照时的性能（最高结果为准确度100%）高于其他痴呆（最高结果为AUC 0.874）。很少有机器学习算法在前瞻性复制中进行了测试。总之，结合机器学习的形态测量MRI显示出作为FTD早期诊断生物标志物的潜力，但是，在推荐该方法用于临床之前，必须进行使用严格方法的研究，并在独立的真实生活队列中验证结果。
Frontotemporal dementia (FTD) is difficult to diagnose, due to its heterogeneous nature and overlap in symptoms with primary psychiatric disorders. Brain MRI for atrophy is a key biomarker but lacks sensitivity in the early stage. Morphometric MRI-based measures and machine learning techniques are a promising tool to improve diagnostic accuracy. Our aim was to review the current state of the literature using morphometric MRI to classify FTD and assess its applicability for clinical practice. A search was completed using Pubmed and PsychInfo of studies which conducted a classification of subjects with FTD from non-FTD (controls or another disorder) using morphometric MRI metrics on an individual level, using single or combined approaches. 28 relevant articles were included and systematically reviewed following PRISMA guidelines. The studies were categorized based on the type of FTD subjects included and the group(s) against which they were classified. Studies varied considerably in subject selection, MRI methodology, and classification approach, and results are highly heterogeneous. Overall many studies indicate good diagnostic accuracy, with higher performance when differentiating FTD from controls (highest result was accuracy of 100%) than other dementias (highest result was AUC of 0.874). Very few machine learning algorithms have been tested in prospective replication. In conclusion, morphometric MRI with machine learning shows potential as an early diagnostic biomarker of FTD, however studies which use rigorous methodology and validate findings in an independent real-life cohort are necessary before this method can be recommended for use clinically.
登录
查看更多内容
影响因子：
2.1
作者：
Chow, Tiffany W.;Binns, Malcolm A.;Black, Sandra
通讯作者：
Black, Sandra
DOI：
10.1016/j.nicl.2017.02.003
发表时间：
2017
期刊：
NeuroImage. Clinical
影响因子：
--
作者：
Bisenius S;Mueller K;Diehl-Schmid J;Fassbender K;Grimmer T;Jessen F;Kassubek J;Kornhuber J;Landwehrmeyer B;Ludolph A;Schneider A;Anderl-Straub S;Stuke K;Danek A;Otto M;Schroeter ML;FTLDc study group
通讯作者：
FTLDc study group
影响因子：
3.7
作者：
Frings L;Yew B;Flanagan E;Lam BY;Hüll M;Huppertz HJ;Hodges JR;Hornberger M
通讯作者：
Hornberger M
DOI：
10.1016/j.nicl.2017.05.018
发表时间：
2017
期刊：
NeuroImage. Clinical
影响因子：
--
作者：
Canu E;Agosta F;Mandic-Stojmenovic G;Stojković T;Stefanova E;Inuggi A;Imperiale F;Copetti M;Kostic VS;Filippi M
通讯作者：
Filippi M
影响因子：
17.7
作者：
Ducharme, Simon;Price, Bruce H.;Dickerson, Bradford C.
通讯作者：
Dickerson, Bradford C.