Multitask Learning Based Three-Dimensional Striatal Segmentation of MRI: fMRI and PET Objective Assessments.
Multitask Learning Based Three-Dimensional Striatal Segmentation of MRI: fMRI and PET Objective Assessments.
复制标题
DOI：
10.1002/jmri.27682
复制
发表时间：
2021-11
影响因子：
4.4
通讯作者：
Huang, Chuan
中科院分区：
文献类型：
--
作者：
Serrano-Sosa, Mario;Van Snellenberg, Jared X.;Meng, Jiayan;Luceno, Jacob R.;Spuhler, Karl;Weinstein, Jodi J.;Abi-Dargham, Anissa;Slifstein, Mark;Huang, Chuan
最近的研究已经建立了一个明确的地形和功能组织的预测和复杂的细分纹状体。这些功能细分的手动分割是劳动密集型和耗时的，并且自动化方法不如手动分割可靠。利用多任务学习（MTL）作为一种方法来分割纹状体的子区域，包括前连合壳核（prePU），前连合尾状核（preCA），后连合壳核（postPU），后连合尾状核（postCA）和腹侧纹状体（VST）。回顾性分析了精神分裂症患者和匹配对照的87个数据集。将1.5 T和3.0 T T1加权（SPGR SENSE，3D BRAVO）MTL生成的分割与帝国理工学院伦敦临床成像中心（CIC）图谱进行比较。Dice相似系数（DSC）被用来比较自动化的方法，手动分割。PET成像：使用[11C]雷氯必利采集60分钟的发射数据。数据通过滤波反投影（FBP）重建，计算机断层扫描（CT）用于衰减校正。结合电位值，BPND，ROI时间序列和全脑连接使用功能磁共振成像图像之间进行了比较手动和自动分割皮尔逊相关性和配对t检验。MTL生成的分割显示出与手动的良好空间一致性（所有纹状体亚区的DSC ≥ 0.72）。MTL生成的分割的BPND值与手动分割相关性良好，在所有尾状核和壳核亚区中R2 ≥ 0.91，在VST中R2=0.69。MTL生成和手动分割之间的fMRI数据的平均Pearson相关系数在所有子区域的时间序列（≥0.86）和全脑连接（≥0.89）中也很高。在PET和fMRI基于任务的评估中，MTL生成的分割结果比CIC生成的分割结果更接近于手动绘制的ROI结果。因此，建议的MTL方法是一种快速可靠的方法，用于3D纹状体亚区分割，其结果与手动分割的ROI相当。
Recent studies have established a clear topographical and functional organization of projections to and from complex subdivisions of the striatum. Manual segmentation of these functional subdivisions is labor-intensive and time-consuming, and automated methods are not as reliable as manual segmentation. To utilize Multi-Task Learning (MTL) as a method to segment subregions of the striatum consisting of pre-commissural putamen (prePU), pre-commissural caudate (preCA), post-commissural putamen (postPU), post-commissural caudate (postCA), and ventral striatum (VST). Retrospective 87 total data sets from patients with schizophrenia and matched controls. 1.5T and 3.0T, T1-weighted (SPGR SENSE, 3D BRAVO) MTL-generated segmentations were compared to the Imperial College London Clinical Imaging Center (CIC) atlas. Dice similarity coefficient (DSC) was used to compare the automated methods to manual segmentations. PET imaging: 60min of emission data were acquired using [11C]raclopride. Data were reconstructed by filtered back projection (FBP) with computed tomography (CT) used for attenuation correction. Binding potential values, BPND, and ROI time-series and whole-brain connectivity using fMRI images were compared between manual and both automated segmentations Pearson correlation and paired t-test. MTL-generated segmentations showed excellent spatial agreement with manual (DSC ≥ 0.72 across all striatal subregions). BPND values from MTL-generated segmentations were shown to correlate well with manual segmentations with R2 ≥ 0.91 in all caudate and putamen subregions, and R2=0.69 in VST. Mean Pearson correlation coefficients of the fMRI data between MTL-generated and manual segmentations were also high in time-series (≥0.86) and whole-brain connectivity (≥0.89) across all subregions. Across both PET and fMRI task-based assessment, results from MTL-generated segmentations more closely corresponded to results from manually drawn ROIs than CIC-generated segmentations did. Therefore, the proposed MTL approach is a fast and reliable method for 3D striatal subregion segmentation with results comparable to manually segmented ROIs.
登录
查看更多内容
影响因子：
4.1
作者：
Abi-Dargham, Anissa;Xu, Xiaoyan;Slifstein, Mark
通讯作者：
Slifstein, Mark
影响因子：
6
作者：
Nakano, Katsuma;Kayahara, Tetsuro;Ushiro, Hiroshi
通讯作者：
Ushiro, Hiroshi
影响因子：
5.7
作者：
Makowski, Carolina;Beland, Sophie;Chakravarty, M. Mallar
通讯作者：
Chakravarty, M. Mallar
影响因子：
5.5
作者：
Coutinho, Joana;Ramos, Ana Filipa;Sampaio, Adriana
通讯作者：
Sampaio, Adriana
影响因子：
3.3
作者：
Spuhler, Karl D.;Ding, Jie;Huang, Chuan
通讯作者：
Huang, Chuan