Functional and structural fingerprinting for glioma characterization: a multimodal perspective

Recently, cancer neuroscience has paid increasing attention to studying the intricate interactions between the brain and glioma, seen from a global perspective. This primary brain tumor, with its aggressive invasiveness and spreading infiltration into the surrounding brain tissues, is known to modify white matter (WM) axonal integrity, glucose metabolism, and brain functionalities through invasion, expansion, and intra-tumoral changes. Brain connectivity can provide a whole-brain characterization of glioma to explore these intricate connections, giving insights into cancer growth and spread. Additionally, glucose metabolism in brain tumors, quantified with positron emission tomography (PET), offers a different view of brain function. Therefore, understanding the interplay between glioma, metabolism, and brain connectivity could be crucial to developing strategies that minimize neurological damage, contribute to personalized treatment for tailored surgery or radiotherapy, and improve patient outcomes. Advanced imaging techniques derived from functional magnetic resonance imaging (fMRI) and diffusion MRI (dMRI) are at the basis of the measurement of functional connectivity (FC) and structural connectivity (SC). Their study was usually developed separately, mostly focusing on specific WM tracts or brain regions. In addition, the literature has recently emphasized SC-FC relationship, becoming a challenging concept in engineering and natural sciences. Structural-functional integration is particularly demanding in the field of glioma tumors, yet complementary information between them can be jointly leveraged to improve the knowledge of this brain pathology. The principal aim of this dissertation is to understand whether glioma can be characterized by the interplay between brain structure and function, with an integrated connectivity perspective and an interest in brain function as described by glucose metabolism. In fact, the preliminary goal of this thesis is to investigate how glioma induces WM structural impairments that are further linked to changes in glucose metabolism. Glucose metabolism can refine our understanding of the metabolic basis of glioma-induced WM connectivity changes, with a complementary vision of brain functions. The SC-FC integrated perspective can improve glioma knowledge yielding the complementary information provided by diffusion structure and function. To address this knowledge gap, this dissertation presents three interrelated studies that aim to explore the impact of glioma on brain connectivity from different perspectives. At first, this thesis aims to develop a new methodology to identify SC alterations, to later link WM impairments to the local brain metabolism of glucose, derived from [18F]fluorodeoxyglucose (FDG) PET data. This research represents a significant preliminary step in exploring brain function, structure, and glioma pathology. Subsequently, we extend the analysis of glioma at the whole brain level by integrating both SC and FC, to provide a more comprehensive view of the impact of glioma on brain networks. WM tract integrity, WM microstructural properties, and functional coupling features are integrated to enhance their complementary information. This novel integrated approach highlights widespread connectivity alterations, providing more comprehensive insights than single modality analyses, especially in seemingly healthy tissue areas. Finally, we employ advanced deep learning techniques, utilizing variational autoencoder based methods, to further explore and quantify the SC-FC connectivity perturbations. This approach offers new insights into the relationship between structural, functional, and combined structural-functional connectivity changes in glioma patients. We also focus on the glioma SC-FC glioma fingerprinting characterization. To conclude, this work provides new frameworks for understanding how the brain responds at the global level to tumor-induced perturbations.