麻省总医院和哈佛医学院Joshua L. Roffman研究团队报道了解决脑发育的大型自动化MRI分析中的人为偏差。2025年7月1日，国际知名学术期刊《自然—神经科学》发表了这一成果。

对9-10岁时获得的11263张T1 MRI扫描图像进行视觉质量控制。多年来，通过青少年大脑认知发展研究，课题组人员发现55.1%的图像质量不理想的样本在皮质厚度和表面积的测量中存在偏差。这些偏差影响了与结构MRI和临床测量相关的分析，导致假阳性和假阴性关联。表面孔数是一种拓扑复杂性的自动指标，可重复性地识别低质量扫描，具有良好的特异性，其作为协变量的包含部分减轻了质量相关的偏差。对高质量扫描的进一步检查揭示了图像预处理过程中引入的额外拓扑错误。用手工编辑校正可重复地改变厚度测量值并加强年龄-厚度关联。研究组在这里证明，质量控制不足破坏了大样本量检测有意义关联的优势。这些偏差可以通过额外的自动化和人工干预来减轻。

据了解，大规模的、以人群为基础的青少年磁共振成像（MRI）研究有望在神经发育和精神疾病风险方面带来革命性的见解。然而，青少年MRI研究特别容易受到运动和其他引入非随机噪声的人为因素的影响。

附：英文原文

Title: Addressing artifactual bias in large, automated MRI analyses of brain development

Author: Elyounssi, Safia, Kunitoki, Keiko, Clauss, Jacqueline A., Laurent, Eline, Kane, Kristina A., Hughes, Dylan E., Hopkinson, Casey E., Bazer, Oren, Sussman, Rachel Freed, Doyle, Alysa E., Lee, Hang, Tervo-Clemmens, Brenden, Eryilmaz, Hamdi, Hirschtick, Randy L., Barch, Deanna M., Satterthwaite, Theodore D., Dowling, Kevin F., Roffman, Joshua L.

Issue&Volume: 2025-07-01

Abstract: Large, population-based magnetic resonance imaging (MRI) studies of adolescents promise transformational insights into neurodevelopment and mental illness risk. However, youth MRI studies are especially susceptible to motion and other artifacts that introduce non-random noise. After visual quality control of 11,263 T1 MRI scans obtained at age 9–10years through the Adolescent Brain Cognitive Development study, we uncovered bias in measurements of cortical thickness and surface area in 55.1% of the samples with suboptimal image quality. These biases impacted analyses relating structural MRI and clinical measures, resulting in both false-positive and false-negative associations. Surface hole number, an automated index of topological complexity, reproducibly identified lower-quality scans with good specificity, and its inclusion as a covariate partially mitigated quality-related bias. Closer examination of high-quality scans revealed additional topological errors introduced during image preprocessing. Correction with manual edits reproducibly altered thickness measurements and strengthened age–thickness associations. We demonstrate here that inadequate quality control undermines advantages of large sample size to detect meaningful associations. These biases can be mitigated through additional automated and manual interventions.

DOI: 10.1038/s41593-025-01990-7

Source: https://www.nature.com/articles/s41593-025-01990-7