179 Physiological MRI Biomarkers & Cognition after SRS for Brain Metastases The multi-delay ASL was processed using the open-source ClinicalASL toolbox37 and FSL BASIL for quantitative CBF maps38. As described before14, a T1-weighted image was generated based on the M0 images from each PLD which was used for subsequent image registration and outlier removal. Outlier removal was performed based on the SCORE method by Duloi et al (Figure 1).39 ME-GRE data were used to generate OEF maps using the integrative QQ approach of Cho et al (Figure 1), with an unsupervised machine learning algorithm, temporal clustering, tissue composition, and total variation (CCTV).40–44 QQ estimates OEF maps based on the QSM processing of the ME-GRE phase and qBOLD modeling of ME-GRE magnitude data. The robustness of QQ against noise was improved substantially by using CCTV45, which groups voxels with similar signal evolution and the same tissue type (GM/WM/CSF) into a cluster and assumes that the voxels have similar model parameters including OEF. All MRIs were registered to the T1 image acquired at baseline using FMRIB’s Linear Image Registration Tool (FLIRT; Figure 1).26,31 Next, CMRO2 maps were generated by combing the OEF and CBF data according to the following equation: (1) in which [H]α is the oxygenated heme molar concentration in the arteriole expressed in µmol/ml. In this study, an oxygenated heme molar concentration of 7.377 µmol/ml is used in accordance with research by Zhang and colleagues (Figure 1).46 Difference maps for CVR and CMRO2 were created by subtracting the follow-up (T1) MRI map from the baseline (T0) MRI map (Figure 1 - Difference maps). Hereby positive values represent an increase in either CVR or CMRO2 at follow-up (T0<T1) and negative values represent a decrease in either CVR or CMRO2 at follow-up (T0>T1). Lastly, based on previously created edema masks, new masks were generated (Figure 1) to incorporate regions only exhibiting edema at baseline (edema T0), regions only exhibiting edema at follow-up (edema T1), and regions exhibiting edema at both time-points (edema T0+T1). The GTV (i.e. brain metastases) were excluded from all edema masks. Next, masks were created that incorporated healthy appearing brain tissue at both time-points. Therefore, the edema mask (edema T0+T1) and PTV were extracted from the whole-brain tissue to make sure any possible malignant tissue was excluded. Then the healthy appearing brain tissue was subdivided into grey matter (GM) and white matter (WM). 7
RkJQdWJsaXNoZXIy MTk4NDMw