135 Etiology in Lesion-Symptom Mapping: Tumor vs. Stroke small to perform this analysis with enough statistical power. Moreover, patients selected for awake surgery differ from patient under general anesthesia in various baseline characteristics.8 Additionally, the use of anti-epileptic medication and/ or dexamethasone may add noise to the cognitive data of our tumor sample. Nevertheless, this will not lead to gross violations of the internal validity of the lesion-symptom associations. Stroke data was collected in a multi-site research project. Patients with larger infarction and typically also more severe clinical stroke symptoms, participate less frequently in research. Thus, LSM studies that include research participants may be systematically biased towards smaller strokes occurring in less eloquent areas and consequently less severe cognitive deficits. Although we have a large variation of stroke volume in our sample, the lefthemispheric strokes are considerably smaller than those in the right hemisphere. This is probably a direct result of the exclusion of patients who were too aphasic to participate. Still, the lesion-symptom associations that are found with research samples, such as our own, remain valid. Moreover, with a large enough sample size, small lesions will allow for better specificity in lesion-symptom associations than large lesions. Tumor delineation We used the hyperintensities on the T2-Flair images to delineate tumor tissue. These hyperintensities are independent of the enhancement of the lesion, and thereby tumor grade, and thus form a widely usable representation of the extent of brain volume potentially hampered in its function. Our tumor lesion masks include tumor core, necrosis, and edema. We implicitly assumed that all visually abnormal tissue on the T2-image is damaged and functionally affected. However, the effect of edema and even tumor infiltration on function may vary within these areas and some function could be preserved. This is probably one explanation for the inconsistent lesion-symptom associations across etiologies in this study. Future research may be aimed at determining the separate effects of edema, infiltration and enhancing tumor core on cognition. Lesion-network mapping Recent insights from both structural and functional network studies, have shown the importance of connections between areas when investigating cognitive effects from lesions. The relevance of white matter damage to cognitive outcome is also supported by the current findings even when using an atlas with only 34 fiber bundles.77 Future work could use both stroke and tumor lesions as seeds in well-established structural or functional connectomes which can then be used as input for lesion network mapping analysis.53,73,78,79 Such studies could elucidate 5
RkJQdWJsaXNoZXIy MTk4NDMw