Authors :
Presenting Author: Omar Chishti, BS – Yale School of Medicine
Evan Collins, BS, MS – Department of Biological Engineering – Massachusetts Institute of Technology; Hari McGrath, MBBS – Graduate Research Associate, Department of Neurosurgery, Yale School of Medicine; Tiffany Zhang, BS – Undergraduate Research Assistant, Department of Computer Science, Yale College; Imran Quraishi, MD, PhD – Assistant Professor, Department of Neurology, Yale School of Medicine; Lawrence Hirsch, MD – Professor, Department of Neurology, Yale School of Medicine; Christopher Benjamin, PhD – Associate Professor, Department of Neurology, Yale School of Medicine; Eyiyemisi Damisah, MD – Assistant Professor, Department of Neurosurgery, Yale School of Medicine; Hitten Zaveri, PhD – Associate Professor, Department of Neurology, Yale School of Medicine; Dennis Spencer, MD – Harvey and Kate Cushing Professor Emeritus, Department of Neurosurgery, Yale School of Medicine; Adithya Sivaraju, MD, MHA – Associate Professor, Comprehensive Epilepsy Center, Department of Neurology, Yale School of Medicine
Rationale: Electrical stimulation mapping (ESM) has long been regarded as the gold standard for delineating language-relevant cortex.
1 Despite its long history, language mapping using ESM has marked practice variation with both extent of language tasks administered & stimulation parameters.
1,2 Language tasks tend to be heavily weighted towards identification of naming sites, and task selection hasn’t received much attention.
3,4Methods: This is a single center retrospective cohort study done at the Yale Comprehensive Epilepsy Center. 15 patients with focal drug resistant epilepsy requiring intracranial EEG monitoring, and subsequent extra-operative language mapping were part of this study.
Stimulation protocol: Language mapping was bipolar, biphasic, 50Hz, 1 to 10 mA & 0.3 ms pulse-width. Six different language tasks (visual naming, auditory naming, reading, repetition, auditory comprehension, & visual comprehension) were administered at each electrode contact pair. ICEEG recordings are observed for afterdischarges or seizures.
Data processing: We transformed the language mapping data for each patient into the Yale Brain Atlas (YBA) space, which consists of 690 parcels, each 1 cm2 built around conserved anatomical features to help communicate anatomically unambiguous localization.5 Task map intensity for each parcel was adjusted for sampling frequency across parcels, accounting for cohort wide robustness of task involvement, and normalized for stimulation amperage. We computed task correlation (across six language tasks) using pairwise Kendall rank coefficient using atlas space vectors. We quantified language involvement, and a variety of other descriptive statistics. We also investigated language presence in the absence of visual naming disruption, and incorporated structural connectivity within the language network.
Results:
Some key results are depicted in Fig. 1 and 2: language task maps, inter-task correlation, relative density of tasks across lobes, language involvement at the gyral level, a language core (parcels involved in all language tasks), and the distribution of cortical areas dominated by different tasks.
Conclusions:
Multitask ESM with precise anatomical correlation is required to visualize the true spatial extent of language cortex. Application of multitask language mapping reveals cortical areas subserving language function despite visual naming being clear. Repetition and reading were the tasks for which visual naming had the lowest co-localization predictive power. A high degree of correlation exists between different language tasks with visual and auditory naming showing the highest. We do not report on post-operative outcomes as most sites with language were spared from resection, which is an inherent limitation of current literature aside from the removal of basal temporal language area.1 However, visualization of the eloquent cortex (with polyfunctional parcels subserving multiple language tasks) is crucial to address the knowledge gap of the spatial extent of different language faculties.
Funding: Swebilius Foundation; NIH R01 NS109062
References:
1 Hamberger et al. 2007.
2 Hamberger et al. 2014.
3 Hermann et al. 2015.
4 Bartha-Doering et al. 2014
5 McGrath et al. 2022.