发表机构
Complejo Asistencial Universitario de León; Hospital Universitario Rey Juan Carlos; Health Research Institute of the Jiménez Díaz Foundation; Rey Juan Carlos University; Universidad de Oviedo; Universidad de León; IFCA-CSIC; Universidad Autónoma de Chile（莱昂大学附属医院综合体; 雷·胡安·卡洛斯大学医院; 希门尼斯·迪亚斯基金会健康研究所; 雷·胡安·卡洛斯大学; 奥维耶多大学; 莱昂大学; 西班牙国家研究委员会高级计算与电子科学研究所; 智利大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对放射组学纹理特征计算中因各向异性图像导致的问题，开发体素间距感知放射组学框架，通过修改PyRadiomics并比较四种配置进行实验，结果表明VS能有效分离几何建模与信号修正，为放射组学分析提供新方法。
AI 中文摘要
目的：放射组学纹理特征通常在体素索引邻域中计算，隐含地假设各向同性空间关系。在各向异性图像中，这可能会将体素几何与插值引起的信号变化混淆。我们开发了一个体素间距感知放射组学框架，在不重采样的情况下将物理几何纳入纹理计算。方法：我们修改了PyRadiomics以考虑体素间距，同时保留原始图像信号。比较了四种配置：原始非重采样提取（NR）、各向同性重采样（RS）、体素间距感知提取（VS）和伪各向同性预处理（FK），其中在不改变图像阵列的情况下覆盖间距元数据。实验包括685个LIDC-IDRI肺结节和209个I-SPY2乳腺MRI病例，有196个放射组学描述符。使用ICC、受试者内变异性、Friedman检验、特征选择、机器学习、多层感知器和外部验证来评估稳健性。主要结果：VS与NR显示出接近原始的一致性：CT中的中位数ICC(A,1)为0.9976，MRI中为0.9984。RS产生的一致性较低且偏差较大，而FK表现出中间行为，证实仅间距元数据就能影响放射组学特征。梯度衍生和邻域敏感描述符受预处理影响最大。在外部CT验证中，VS保留了与NR相当的预测性能，而MRI在预处理策略和分类器之间表现出更大的变异性。意义：体素间距感知提取将几何建模与插值引起的信号修正分开，同时保留原始图像信号，为各向异性CT和MRI的放射组学分析提供了一种连贯的各向同性重采样替代方法。
英文摘要
Objective: Radiomic texture features are usually computed in voxel-index neighborhoods, implicitly assuming isotropic spatial relationships. In anisotropic images, this can confound voxel geometry with interpolation-induced signal changes. We developed a voxel-spacing-aware radiomic framework that incorporates physical geometry into texture computation without resampling. Approach: We modified PyRadiomics to account for voxel spacing while preserving the native image signal. Four configurations were compared: native non-resampled extraction (NR), isotropic resampling (RS), voxel-spacing-aware extraction (VS), and fake-isotropic preprocessing (FK), in which spacing metadata were overwritten without altering the image array. Experiments included 685 LIDC-IDRI pulmonary nodules and 209 I-SPY2 breast MRI cases, with 196 radiomic descriptors. Robustness was assessed using ICC, within-subject variability, Friedman testing, feature selection, machine learning, a multilayer perceptron, and external validation. Main results: VS showed near-native agreement with NR: median ICC(A,1) was 0.9976 in CT and 0.9984 in MRI. RS produced lower agreement and larger deviations, while FK showed intermediate behavior, confirming that spacing metadata alone can affect radiomic features. Gradient-derived and neighborhood-sensitive descriptors were most affected by preprocessing. VS preserved predictive performance comparable to NR in external CT validation, whereas MRI showed greater variability across preprocessing strategies and classifiers. Significance: Voxel-spacing-aware extraction separates geometric modeling from interpolation-induced signal modification while preserving the native image signal, offering a coherent alternative to isotropic resampling for radiomic analysis of anisotropic CT and MRI.
CommentsManuscript under peer review
