Lundholm Lukas, Montelius Mikael, Jalnefjord Oscar, Schoultz Elin, Forssell-Aronsson Eva, Ljungberg Maria
Department of Medical Radiation Sciences, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden.
Department of Medical Physics and Biomedical Engineering, Sahlgrenska University Hospital, Gothenburg, Sweden.
NMR Biomed. 2025 Jun;38(6):e70050. doi: 10.1002/nbm.70050.
Diffusion MRI models accounting for varying diffusion times and high b-values, such as VERDICT, hold potential for non-invasively characterizing tumor tissue types, potentially enabling improved tumor grading, and treatment evaluation. Furthermore, cluster analysis can aid in identifying multidimensional patterns in the diffusion MRI (dMRI) data that are not apparent when analyzing individual parameters in isolation. The aim of this study was to evaluate how well cluster analysis of VERDICT parameters can be used for intratumor tissue characterization compared to ADC in a mouse model of human small intestine neuroendocrine tumor (GOT1), and to validate the method by histological analysis. Mice implanted with GOT1 were irradiated and subsequently imaged using a dMRI protocol designed for estimation of VERDICT parameters and ADC values. Histological analysis using hematoxylin and eosin (H&E), Masson's trichrome, and Ki67 staining identified three distinct tumor tissue types: necrotic, fibrotic, and viable tumor tissue. ROIs were drawn on regions of high and low ADC, which spatially matched with necrosis or fibrosis, and viable tumor tissue, respectively. Among the VERDICT parameters, the cell radius index (R) was most effective in distinguishing between necrotic and fibrotic tissue, whereas the intracellular fraction (f) was the most effective in differentiating viable from non-viable tissue. A Gaussian mixture model (GMM) of three clusters, representing each tumor tissue type, was fitted to R and f of all tumor voxel data. VERDICT cluster maps corresponded well with the histology classification maps overall. Fibrotic tissue corresponded best with the cluster of low f and low R, necrotic tissue with the cluster of low f and high R, and viable tumor tissue with the cluster of high f and intermediate R. In conclusion, GMM cluster analysis of VERDICT MRI data shows potential in differentiating necrotic, fibrotic, and viable tumor tissue in irradiated GOT1 tumors.
扩散加权磁共振成像（Diffusion MRI）模型，如VERDICT，考虑了不同的扩散时间和高b值，具有无创性表征肿瘤组织类型的潜力，有可能改善肿瘤分级和治疗评估。此外，聚类分析有助于识别扩散加权磁共振成像（dMRI）数据中的多维模式，而单独分析单个参数时这些模式并不明显。本研究的目的是评估在人小肠神经内分泌肿瘤（GOT1）小鼠模型中，与表观扩散系数（ADC）相比，VERDICT参数的聚类分析用于肿瘤内组织表征的效果如何，并通过组织学分析验证该方法。将GOT1植入小鼠体内，进行照射，随后使用旨在估计VERDICT参数和ADC值的dMRI方案进行成像。使用苏木精和伊红（H&E）、Masson三色染色和Ki67染色进行组织学分析，确定了三种不同的肿瘤组织类型：坏死、纤维化和存活肿瘤组织。在高ADC和低ADC区域绘制感兴趣区（ROI），分别与坏死或纤维化区域以及存活肿瘤组织在空间上匹配。在VERDICT参数中，细胞半径指数（R）在区分坏死组织和纤维化组织方面最有效，而细胞内分数（f）在区分存活组织和非存活组织方面最有效。将代表每种肿瘤组织类型的三个聚类的高斯混合模型（GMM）拟合到所有肿瘤体素数据的R和f上。VERDICT聚类图总体上与组织学分类图吻合良好。纤维化组织与低f和低R的聚类最匹配，坏死组织与低f和高R的聚类最匹配，存活肿瘤组织与高f和中等R的聚类最匹配。总之，VERDICT MRI数据的GMM聚类分析在区分照射后GOT1肿瘤中的坏死、纤维化和存活肿瘤组织方面显示出潜力。
