马氏距离
连接体
计算机科学
人类连接体项目
人工智能
可靠性(半导体)
特征(语言学)
模式识别(心理学)
机器学习
功能连接
神经科学
生物
物理
哲学
量子力学
功率(物理)
语言学
作者
Xun‐Heng Wang,Jimmy Yun,Lihua Li
标识
DOI:10.1016/j.mri.2019.11.003
摘要
Building individual brain networks form the single volume of anatomical MRI is a challenging task. Furthermore, the high-order connectivity of morphological networks remains unexplored. This paper aimed to investigate the individual high-order morphological connectivity from anatomical MRI. Towards this goal, a unified framework based on six feature distances (euclidean, seuclidean, mahalanobis, cityblock, minkowski, and chebychev) was proposed to derive high-order interregional morphological features. The test-retest datasets and the healthy aging datasets were applied to analyze the reliability and the inter-subject variability of the novel features. In addition, the predictive models based on these novel features were established for age estimation. The proposed six neuroanatomical features exhibited significant high-to-excellent reliability. Certain connections were significantly correlated to biological age based on the six novel metrics (p < .05, FDR corrected). Moreover, the predicted age were significantly correlated to the original age in each regression task (r > 0.5, p < 10−6). The results suggested that the novel high-order metrics were reliable and could reflect individual differences, which could be beneficial for current methods of individual brain connectomes.
科研通智能强力驱动
Strongly Powered by AbleSci AI