Introduction: This study aimed at evaluating a new segmentation technique based on diffusion tensor decomposition into an isotropic (p) and an anisotropic component (q) to better characterize the tumour tissue heterogeneity. Enabling a voxel-by-voxel analysis, this technique allow the depiction of parametric changes at distinct timepoints.To characterize the structural heterogeneity of low-grade-gliomas (LGGs), we aimed at: i) comparing the images obtained from the p:q decomposition with the dataset acquired with FLAIR sequences; ii) correlating the results of segmentation with histopathological findings from image-guided tumour biopsies; iii) evaluating the evolution of a given lesion throughout time, when applicable, and comparing distinct imaging behaviour with neurophysiologic intraoperative data.
Methods: p and q maps were obtained from diffusion-tensor-acquisitions in 50 patients with primary LGGs. They were fused with DTI-FT reconstruction and volumetric FLAIR and post-GdT1-weigthed images, loaded into the neuronavigation system, to be available intraoperatively.
Results: p map depicted a tumour area larger than that detected in FLAIR images. The tumour ROI on q maps was located inside the area depicted on p maps. The structural heterogeneity of p and q maps did not match with the degree of heterogeneity shown by FLAIR images, or with areas of contrast enhancement.
Samples taken in p areas always revealed the presence of a LGG; samples in p areas located externally to the border depicted in FLAIR images and within p maps, were classified as infiltrative areas of LGG. Samples external to p areas was normal brain parenchyma; Samples in q areas showed regions of highly compacted tumour cells, and in tumours in which p and q maps did not match, the area of q maps showed foci of anaplasia. P:q maps also correlated with different molecular profile (MGMT status, 1p19q deletion, IDH-1 status).
When applicable, neurophysiologic changes in stimulus threshold well correlated with the degree of changes depicted by p:q functional maps.
Conclusions: p and q maps better correlate with the degree of LGG heterogeneity than conventional MR images; this data has potential clinical implication for sampling for histological and molecular diagnosis and for better quantify the response to a given treatment over time.
Patient Care: For better delineation of a given tumor mass, presumably assigned as low grade gliomas
Learning Objectives: 1. To build the awareness regaarding tumor heterogeneity through DT depiction of low grade gliomas