Automatic segmentation, feature extraction and comparison of healthy and stroke cerebral vasculature.
Automatic segmentation, feature extraction and comparison of healthy and stroke cerebral vasculature.
复制标题
DOI：
10.1016/j.nicl.2021.102573
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Laksari K
中科院分区：
文献类型：
--
作者：
Deshpande A;Jamilpour N;Jiang B;Michel P;Eskandari A;Kidwell C;Wintermark M;Laksari K
基于活动轮廓的概率滤波自动脑血管分割。血管网络的几何特征研究脑血管疾病。脑卒中与健康脑血管的定量比较。血管随年龄和脑血管疾病的变化。CTA与MRA成像方式的比较。脑血管的准确分割及其形态学的定量评估对于各种诊断和治疗目的至关重要，并且与研究脑健康和疾病有关。然而，由于血管成像数据的复杂性，这仍然是一项具有挑战性的任务。本文提出了一种无需人工干预的自动脑血管分割方法，以及一种对二值分割图进行骨架化以提取血管几何特征和表征血管结构的方法。我们将基于hessian的概率血管增强滤波与基于主动轮廓的技术相结合，对磁共振和计算机断层扫描血管图像（MRA和CTA）进行分割，随后提取血管中心线和直径，以计算血管系统的几何特性。通过对Circle-of-Willis区域的3D幻影进行验证，我们的方法显示出84%的平均Dice相似系数（DSC）和85%的平均Pearson相关系数（PCC），最小的修正Hausdorff距离（MHD）误差（最多3个表面像素），并且在使用DSC、PCC和MHD进行定量比较时，与现有分割算法相比表现出优越的性能。我们随后应用算法的数据集40科目,包括1)MRA扫描的健康受试者(n = 10,年龄= 30±9),2)MRA扫描的中风患者(n = 10,年龄= 51±15),3)健康受试者的CTA扫描(n = 10, 12岁= 62±),和4)CTA扫描的中风患者(n = 10,年龄= 68±11),并获得了量化比较的中风和正常的血管成像模式。中风患者的血管网络相比,年龄调整健康受试者被发现有显著(p < 0.05)更高的弯曲度(3.24±0.88 rad / cm和7.17±1.61 rad / cm MRA,和4.36±1.32 rad / cm和7.80±0.92 rad / cm CTA),更高的分形维数(1.36±0.28和1.71±0.14 MRA,和1.56±0.05和1.69±0.20 CTA),较低的总长度(3.46±0.99米和2.20±0.67米CTA),较低的总容积(61.80±18.79毫升CTA和34.43±22.9毫升),平均枝径（2.4±0.21 mm， CTA为2.18±0.07 mm）较低，平均枝长（4.81±1.97 mm， MRA为8.68±2.03 mm）较低。我们还研究了血管特征的变化与年龄和成像方式的关系。虽然我们观察到由于年龄的原因，特征之间存在差异，但统计分析没有显示任何显著差异，而我们发现两种成像方式之间的分支数量有显著差异（p < 0.05） （MRA为201±73，CTA为189±69）。我们的分割和特征提取算法可以应用于任何成像方式，未来可以自动获得三维分割的血管系统，用于诊断和治疗计划，也可以用于临床研究中风和其他脑血管疾病（CVD）的形态学变化。
Probabilistic filter with active contours to automate cerebrovascular segmentation. Geometric features of vessel network to study cerebrovascular diseases. Quantitative comparison of stroke and healthy cerebral vasculature. Vascular changes with aging and cerebrovascular disease. Comparison of CTA and MRA imaging modalities. Accurate segmentation of cerebral vasculature and a quantitative assessment of its morphology is critical to various diagnostic and therapeutic purposes and is pertinent to studying brain health and disease. However, this is still a challenging task due to the complexity of the vascular imaging data. We propose an automated method for cerebral vascular segmentation without the need of any manual intervention as well as a method to skeletonize the binary segmented map to extract vascular geometric features and characterize vessel structure. We combine a Hessian-based probabilistic vessel-enhancing filtering with an active-contour-based technique to segment magnetic resonance and computed tomography angiograms (MRA and CTA) and subsequently extract the vessel centerlines and diameters to calculate the geometrical properties of the vasculature. Our method was validated using a 3D phantom of the Circle-of-Willis region, demonstrating 84% mean Dice similarity coefficient (DSC) and 85% mean Pearson’s correlation coefficient (PCC) with minimal modified Hausdorff distance (MHD) error (3 surface pixels at most), and showed superior performance compared to existing segmentation algorithms upon quantitative comparison using DSC, PCC and MHD. We subsequently applied our algorithm to a dataset of 40 subjects, including 1) MRA scans of healthy subjects (n = 10, age = 30 ± 9), 2) MRA scans of stroke patients (n = 10, age = 51 ± 15), 3) CTA scans of healthy subjects (n = 10, age = 62 ± 12), and 4) CTA scans of stroke patients (n = 10, age = 68 ± 11), and obtained a quantitative comparison between the stroke and normal vasculature for both imaging modalities. The vascular network in stroke patients compared to age-adjusted healthy subjects was found to have a significantly (p < 0.05) higher tortuosity (3.24 ± 0.88 rad/cm vs. 7.17 ± 1.61 rad/cm for MRA, and 4.36 ± 1.32 rad/cm vs. 7.80 ± 0.92 rad/cm for CTA), higher fractal dimension (1.36 ± 0.28 vs. 1.71 ± 0.14 for MRA, and 1.56 ± 0.05 vs. 1.69 ± 0.20 for CTA), lower total length (3.46 ± 0.99 m vs. 2.20 ± 0.67 m for CTA), lower total volume (61.80 ± 18.79 ml vs. 34.43 ± 22.9 ml for CTA), lower average diameter (2.4 ± 0.21 mm vs. 2.18 ± 0.07 mm for CTA), and lower average branch length (4.81 ± 1.97 mm vs. 8.68 ± 2.03 mm for MRA), respectively. We additionally studied the change in vascular features with respect to aging and imaging modality. While we observed differences between features as a result of aging, statistical analysis did not show any significant differences, whereas we found that the number of branches were significantly different (p < 0.05) between the two imaging modalities (201 ± 73 for MRA vs. 189 ± 69 for CTA). Our segmentation and feature extraction algorithm can be applied on any imaging modality and can be used in the future to automatically obtain the 3D segmented vasculature for diagnosis and treatment planning as well as to study morphological changes due to stroke and other cerebrovascular diseases (CVD) in the clinic.
登录
查看更多内容
影响因子：
3.3
作者：
Hsu CY;Schneller B;Alaraj A;Flannery M;Zhou XJ;Linninger A
通讯作者：
Linninger A
影响因子：
20.1
作者：
Donato AJ;Machin DR;Lesniewski LA
通讯作者：
Lesniewski LA
影响因子：
10.6
作者：
Chan, TF;Vese, LA
通讯作者：
Vese, LA
影响因子：
5.8
作者：
Arganda-Carreras, Ignacio;Kaynig, Verena;Seung, H. Sebastian
通讯作者：
Seung, H. Sebastian
影响因子：
2.9
作者：
DUBUC, B;QUINIOU, JF;ZUCKER, SW
通讯作者：
ZUCKER, SW