Lasek Julia, Stefańska Anna K, Kierońska-Siwak Sara, Obuchowicz Rafał, Krzyżak A T
AGH University of Krakow, Kraków, Poland.
Department of Neurosurgery and Neurology, Jan Biziel University Hospital No 2, Collegium Medicum, Nicolaus Copernicus University, Bydgoszcz, Poland.
Comput Biol Med. 2025 Aug;194:110503. doi: 10.1016/j.compbiomed.2025.110503. Epub 2025 Jun 2.
Diffusion Tensor Imaging (DTI) is integral to presurgical planning and early detection of neurodegenerative diseases. It reconstructs white matter pathways and enhances brain connectivity insights. However, systematic errors and noise hinder DTI's utility, disrupting the visualization of critical anatomical details necessary for understanding brain function and disorders.
This study evaluated the combined impact of denoising and B-matrix Spatial Distribution (BSD) correction on DTI accuracy and tractography quality using two datasets: a single-subject scan and a 40-subject cohort - each acquired with corresponding phantoms. Each was processed using six configurations-three preprocessing levels (RAW, DENOISED, PREPROC), with and without BSD correction.
In vivo, significant changes in FA and MD were observed across major white matter tracts, with the combined use of denoising and BSD. DTI metrics were assessed in specific brain structures, including the corpus callosum, internal capsule, putamen, and thalamus, using both manually defined and atlas-based ROIs. Visual and quantitative evaluations showed that denoising and BSD are complementary steps and should be used together to reduce both random and systematic errors. In phantom experiments, BSD correction had a substantially greater effect on improving DTI metric accuracy than the full preprocessing pipeline alone, highlighting its critical role in correcting errors associated with nonuniformity of magnetic field gradients.
This study underscores the importance of correcting spatial systematic errors and noise to ensure precise neuroimaging data. Such advancements are critical for deepening our understanding of neural connectivity and improving its clinical applications in diagnosing and treating neurological conditions.
扩散张量成像（DTI）对于手术前规划和神经退行性疾病的早期检测不可或缺。它可重建白质通路并增强对脑连接性的认识。然而，系统误差和噪声阻碍了DTI的效用，干扰了理解脑功能和疾病所需的关键解剖细节的可视化。
本研究使用两个数据集评估去噪和B矩阵空间分布（BSD）校正对DTI准确性和纤维束成像质量的综合影响：一个单受试者扫描数据集和一个40受试者队列数据集，每个数据集均与相应的体模一起采集。每个数据集都使用六种配置进行处理——三个预处理级别（原始、去噪、预处理），有无BSD校正。
在活体中，通过联合使用去噪和BSD，在主要白质束中观察到FA和MD的显著变化。使用手动定义和基于图谱的感兴趣区域，在包括胼胝体、内囊、壳核和丘脑在内的特定脑结构中评估DTI指标。视觉和定量评估表明，去噪和BSD是互补步骤，应一起使用以减少随机误差和系统误差。在体模实验中，BSD校正对提高DTI指标准确性的影响比单独的完整预处理流程大得多，突出了其在纠正与磁场梯度不均匀性相关误差方面的关键作用。
本研究强调了校正空间系统误差和噪声以确保精确神经影像数据的重要性。这些进展对于深化我们对神经连接性的理解以及改善其在神经系统疾病诊断和治疗中的临床应用至关重要。
