RESUMEN
Signal processing is a very useful field of study in the interpretation of signals in many everyday applications. In the case of applications with time-varying signals, one possibility is to consider them as graphs, so graph theory arises, which extends classical methods to the non-Euclidean domain. In addition, machine learning techniques have been widely used in pattern recognition activities in a wide variety of tasks, including health sciences. The objective of this work is to identify and analyze the papers in the literature that address the use of machine learning applied to graph signal processing in health sciences. A search was performed in four databases (Science Direct, IEEE Xplore, ACM, and MDPI), using search strings to identify papers that are in the scope of this review. Finally, 45 papers were included in the analysis, the first being published in 2015, which indicates an emerging area. Among the gaps found, we can mention the need for better clinical interpretability of the results obtained in the papers, that is not to restrict the results or conclusions simply to performance metrics. In addition, a possible research direction is the use of new transforms. It is also important to make new public datasets available that can be used to train the models.
RESUMEN
PURPOSE: Glioma is associated with pathologically high (peri)tumoral brain activity, which relates to faster progression. Functional connectivity is disturbed locally and throughout the entire brain, associating with symptomatology. We, therefore, investigated how local activity and network measures relate to better understand how the intricate relationship between the tumor and the rest of the brain may impact disease and symptom progression. METHODS: We obtained magnetoencephalography in 84 de novo glioma patients and 61 matched healthy controls. The offset of the power spectrum, a proxy of neuronal activity, was calculated for 210 cortical regions. We calculated patients' regional deviations in delta, theta and lower alpha network connectivity as compared to controls, using two network measures: clustering coefficient (local connectivity) and eigenvector centrality (integrative connectivity). We then tested group differences in activity and connectivity between (peri)tumoral, contralateral homologue regions, and the rest of the brain. We also correlated regional offset to connectivity. RESULTS: As expected, patients' (peri)tumoral activity was pathologically high, and patients showed higher clustering and lower centrality than controls. At the group-level, regionally high activity related to high clustering in controls and patients alike. However, within-patient analyses revealed negative associations between regional deviations in brain activity and clustering, such that pathologically high activity coincided with low network clustering, while regions with 'normal' activity levels showed high network clustering. CONCLUSION: Our results indicate that pathological activity and connectivity co-localize in a complex manner in glioma. This insight is relevant to our understanding of disease progression and cognitive symptomatology.
