使用机器学习识别库辛病的灰质变化:一种可解释的方法
Yue Long1, Jie Ren2, FuChao Cheng1
1College of Computer, Chengdu University, Chengdu, China.
Medical physics
|April 1, 2024
概括
这项研究引入了一个可解释的机器学习 (ML) 框架来分析库辛病 (CD) 中的大脑变化. 机器学习方法成功地确定了关键的大脑区域及其与临床症状的相关性,为改善诊断提供了潜力.
科学领域:
- 神经成像是一种神经成像.
- 机器学习 机器学习
- 内分泌学 在内分泌学.
背景情况:
- 库辛病 (CD) 涉及过度上腺皮质增生激素的分泌,导致显著的大脑变化可通过MRI检测.
- 在CD中MRI的传统统计分析在预测个体患者的结果方面存在局限性.
研究的目的:
- 开发和验证可解释的机器学习 (ML) 框架,用于对库辛病的结构性MRI数据进行全面分析.
- 为了评估模型的性能,确定预测性大脑区域,并将它们与临床症状相关联.
主要方法:
- 一个可解释的ML框架被应用于来自库辛病患者的结构性MRI数据.
- 该框架包括用于识别大脑区域的模型级评估,用于预测准确性的特征级评估,以及用于临床相关性的生物学级评估.
主要成果:
- 在CD患者中发现了脑部区域的显著变化,包括Insula,Fusiform gyrus,Superior frontal gyrus,Precuneus和Inferior frontal gyrus.
- 在临床症状和前叶,胰岛和嗅觉皮层之间发现了相关性,这与之前的研究一致.
结论:
- 拟议的ML框架显示了阐明库辛病的病理生理机制的巨大潜力.
- 这种方法在诊断其他神经和内分泌疾病方面可能具有更广泛的应用.
相关概念视频
01:26Cushing Syndrome I: Introduction
Cushing syndrome refers to the collection of clinical manifestations that arise when tissues are exposed to excessive amounts of cortisol or cortisol-like medications over an extended period. Cortisol, a glucocorticoid produced by the adrenal cortex, regulates metabolism, immune responses, and the body’s adaptation to stress. When its concentration remains chronically elevated, these physiological pathways become dysregulated, resulting in the characteristic features of the syndrome.Exogenous...
01:19Cushing Syndrome II: Pathophysiology
Cortisol production is normally governed by the hypothalamic–pituitary–adrenal (HPA) axis, which maintains hormonal balance through tightly regulated feedback mechanisms. Disruption of this regulatory system is central to the development of Cushing syndrome, whether the excess cortisol originates from external medications or internal pathology. Persistent cortisol elevation alters metabolism, immune function, and endocrine signaling, producing the characteristic clinical features of the...