Different Regional Patterns in Gray Matter-based Age Prediction
Different Regional Patterns in Gray Matter-based Age Prediction
复制标题
基于灰质的年龄预测的不同区域模式
DOI：
10.1007/s12264-022-01016-3
复制
发表时间：
2023
影响因子：
5.6
通讯作者：
Tianzi Jiang
中科院分区：
文献类型：
--
作者：
Nianming Zuo;Tianyu Hu;Hao Liu;Jing Sui;Yong Liu;Tianzi Jiang
大脑在不同年龄段经历持续的变化，以支持大脑发育和功能重组。在成年期，虽然从神经生物学的角度来看，大脑已经成熟，但它仍然受到生活习惯，家庭环境，社会经济地位和工作环境等外部因素的影响[1]。与实足年龄（CA）相比，脑（或生物）年龄（BA）被概念化为表征衰老过程和神经心理状态以及个体认知表现的重要指标。越来越多的证据表明，BA可以通过神经成像技术（包括MRI）进行评估[2]。由于T1加权MRI的采集时间短、图像特征稳定且在临床诊断期间（通常）可用，因此T1加权MRI被认为是估计BA的首选，其结构特征包括灰质（GM）和白色物质（WM）的局部/全局体积、大脑皮质的几何特征以及边界处GM和WM之间的区别[3]。在基于T1加权MRI的BA预测中存在两个难以解决的问题。第一个问题是如何提高预测精度和保留更少的参数。目前，研究要么受到相对较差的准确性的限制[4]，要么受到不完整的年龄跨度数据集的限制。因此，深度学习和生命周期数据集在这个领域逐渐流行起来。当前深度学习方法的一个明显缺点是它们不太关心模型规模和参数数量。最近将全局-局部Transformer应用于BA估计，8379例受试者（年龄范围0-97岁）的平均绝对误差（MAE）为2.70，但参数数量达到2041万[5]。过多的参数意味着更大的计算负担，更低的训练效率，更弱的泛化性能，以及相当大的内存占用，使模型的设备友好性降低。第二个问题是如何理解年龄预测中的神经生物学原理。虽然左念明和胡天宇对这项工作做出了同样的贡献。
The brain experiences ongoing changes across different ages to support brain development and functional reorganization. During the span of adulthood, although the brain has matured from a neurobiological perspective, it is still continuously shaped by external factors such as living habit, family setting, socioeconomic status, and working environment [1]. In contrast to chronological age (CA), brain (or biological) age (BA) is conceptualized as an important index for characterizing the aging process and neuropsychological state, as well as individual cognitive performance. Growing evidence indicates that BA can be assessed by neuroimaging techniques, including MRIs [2]. Due to their short collection time, stable image features, and (usual) availability during clinical diagnosis, T1-weighted MRIs are considered the first choice for estimating BA, with structural features including local/global volumes of gray matter (GM) and white matter (WM), geometrical characteristics of the cerebral cortex, and distinctions between GM and WM at the boundary [3]. There are two elusive questions in T1-weighted MRIbased BA prediction. The first question is how to improve prediction accuracy and retain fewer parameters. At present, research is either limited by relatively poor accuracy [4] or by incomplete age-span datasets. So deep learning and lifespan datasets are gradually becoming popular in this field. A clear drawback of current deep learning methodologies is that they are less concerned with model scale and number of parameters. Recent application of a global-local transformer to BA estimation achieved a mean absolute error (MAE) of 2.70 for 8379 subjects (age range 0–97 years), but the number of parameters reached 20.41 million [5]. Too many parameters mean a greater computing burden, lower training efficiency, weaker generalization performance, and considerable memory occupation, making the model less device-friendly. The second question is how to understand the neurobiological principles in age prediction. Although Nianming Zuo and Tianyu Hu contributed equally to this work.
影响因子：
3.7
作者：
Erus, Guray;Battapady, Harsha;Gur, Ruben C.
通讯作者：
Gur, Ruben C.