Artificial Intelligence Graduate School, Gwangju Institute of Science and Technology, Gwangju, South Korea.
Department of Neurology, College of Medicine, Ewha Womans University, Seoul, South Korea.
Eur J Neurol. 2023 Jun;30(6):1574-1584. doi: 10.1111/ene.15775. Epub 2023 Mar 27.
Alzheimer disease (AD) is the most common type of dementia. Amyloid-β (Aβ) positivity is the main diagnostic marker for AD. Aβ positron emission tomography and cerebrospinal fluid are widely used in the clinical diagnosis of AD. However, these methods only assess the concentrations of Aβ, and the accessibility of these methods is thus relatively limited compared with structural magnetic resonance imaging (sMRI).
We investigated whether regions of interest (ROIs) in sMRIs can be used to predict Aβ positivity for samples with normal cognition (NC), mild cognitive impairment (MCI), and dementia. We obtained 846 Aβ negative (Aβ-) and 865 Aβ positive (Aβ+) samples from the Alzheimer's Disease Neuroimaging Initiative database. To predict which samples are Aβ+, we built five machine learning models using ROIs and apolipoprotein E (APOE) genotypes as features. To test the performance of the machine learning models, we constructed a new cohort containing 97 Aβ- and 81 Aβ+ samples.
The best performing machine learning model combining ROIs and APOE had an accuracy of 0.798, indicating that it can help predict Aβ+. Furthermore, we searched ROIs that could aid our prediction and discovered that an average left entorhinal cortical region (L-ERC) thickness is an important feature. We also noted significant differences in L-ERC thickness between the Aβ- and Aβ+ samples even in the same diagnosis of NC, MCI, and dementia.
Our findings indicate that ROIs from sMRIs along with APOE can be used as an initial screening tool in the early diagnosis of AD.
阿尔茨海默病（AD）是最常见的痴呆类型。淀粉样蛋白-β（Aβ）阳性是 AD 的主要诊断标志物。Aβ 正电子发射断层扫描和脑脊液在 AD 的临床诊断中被广泛应用。然而，这些方法仅评估 Aβ 的浓度，与结构磁共振成像（sMRI）相比，这些方法的可及性相对有限。
我们研究了 sMRI 中的感兴趣区域（ROI）是否可用于预测认知正常（NC）、轻度认知障碍（MCI）和痴呆患者的 Aβ 阳性。我们从阿尔茨海默病神经影像学倡议数据库中获得了 846 个 Aβ 阴性（Aβ-）和 865 个 Aβ 阳性（Aβ+）样本。为了预测哪些样本是 Aβ+，我们使用 ROI 和载脂蛋白 E（APOE）基因型作为特征构建了五个机器学习模型。为了测试机器学习模型的性能，我们构建了一个包含 97 个 Aβ-和 81 个 Aβ+样本的新队列。
结合 ROI 和 APOE 的性能最佳的机器学习模型的准确率为 0.798，表明它可以帮助预测 Aβ+。此外，我们搜索了有助于我们预测的 ROI，并发现左侧内嗅皮层区域（L-ERC）的平均厚度是一个重要特征。我们还注意到，即使在相同的 NC、MCI 和痴呆诊断中，Aβ-和 Aβ+样本之间的 L-ERC 厚度也存在显著差异。
我们的研究结果表明，sMRI 的 ROI 与 APOE 结合可以作为 AD 早期诊断的初始筛查工具。
