Ensuring that novel resting‐state <scp>fMRI</scp> metrics are physiologically grounded, interpretable and meaningful (A commentary on Canna <i>et al</i>., 2017)

Resting-state functional magnetic resonance imaging (rs-fMRI) is a popular modality for studying human brain function, with a profusion of user-friendly software tools available for data analysis (Buckner et al., 2013).Conventional approaches measure correlations in spontaneous blood-oxygen-level-dependent (BOLD) signals across networks of brain regions; more recently, graph theoretical measures, including betweenness-and degree-centrality, have helped to further characterize the functional architecture of the brain (Bullmore & Sporns, 2009).Newer metrics have also been developed to assess dynamic aspects of rs-fMRI network activity; examples include BOLD signal variability (Fox & Raichle, 2007), 'sliding-window' correlations (Chang & Glover, 2010), and amplitude of low-frequency fluctuations (ALFF) (Zou et al., 2008).The fast-proliferating set of techniques also includes regional homogeneity (Zang et al., 2004), multi-voxel pattern analysis (Norman et al., 2006), brain-wide association studies (Cheng et al., 2016) and numerous others.Such tools, properly applied, have the potential to advance our understanding of human brain function in health and disease.In this paper, Interhemispheric Functional Connectivity in Anorexia and Bulimia Nervosa, Canna et al., (2017) apply one such novel analysis to rs-fMRI in eating disorders.The authors used voxel-mirrored homotopic connectivity (VMHC) to identify possible interhemispheric differences in brain activity between patients with anorexia nervosa (AN), bulimia nervosa (BN) and healthy controls.VMHC measures BOLD signal correlations between a given voxel in one hemisphere and the contralateral voxel at the mirror-image location, under the assumption that these two voxels contain 'homotopic' volumes of brain tissue with related functions.Relative to controls, AN participants displayed lower VMHC in the insula, while BN participants displayed lower VMHC in the dorsolateral prefrontal and orbitofrontal cortex.The authors also used a technique called interhemispheric spectral coherence analysis (IHSC) to examine the BOLD power spectra in regions with VMHC differences.The authors attribute the observed differences to abnormal cognitive-and reward-based behaviours conventionally observed in these eating disorders.However, the reader may be left wondering whether mirror-image voxels necessarily contain brain regions with homotopic functions, or how best to interpret high vs. low VMHC values, or how to interpret signal coherence between different bands of the rs-fMRI signal.Techniques like VMHC add to an already-crowded toolbox of analytical techniques for rs-fMRI data, and raise several questions regarding their use and interpretation.First, how should one select the appropriate rs-fMRI technique to address the research question?Why is a given technique preferable to others for a given question?Are there any assumptions inherent to the technique that could impact the validity of findings?Finally, how does the metric relate to the underlying brain activity or physiology (already measured indirectly via the BOLD effect)?For the paper by Canna and colleagues specifically, one might ask, what is VMHC expected to reveal regarding the underlying pathophysiology of AN/BN?Why select VMHC over some other method to study AN/BN?Studies employing novel rs-fMRI techniques should provide clear rationales for their use in the context of the study population or research aims.Interpretability presents a critical issue for novel rs-fMRI analysis techniques.For many metrics, there is limited information available regarding their a priori relationship to neurophysiology, phenotype or normative distribution.To be properly interpretable, findings arising from VMHC, IHSC or other rs-fMRI techniques require theoretical and empirical benchmarks in other modalities, such as behavioural, clinical or neurophysiological measures.For example, where the study by Canna and colleagues found lower insular VMHC in AN, the authors helpfully reviewed parallels between this finding and previous research involving resting-state and task-based fMRI in AN.However, the result would be more interpretable with a more neurophysiologically grounded account of how deficits in left-right insula connectivity might contribute to the pathophysiology of AN.Interpretability would be further bolstered by making reference to other modalities, like electrophysiology, and by relating abnormal connectivity metrics to a specific phenotype, using psychometric or clinical measures, or behavioural tasks.The same applies to spectral analyses, where the physiological significance of different sub-spectra of the rs-fMRI signal is not wellestablished.In the absence of external benchmarks, the significance of low VMHC or any particular cross-spectral coherence may be difficult to interpret in any illuminating way.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.