Personalized functional imaging identifies brain stimulation target for a patient with trauma-induced functional disruption

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


Dear editor
Non-invasive neuromodulatory techniques such as transcranial magnetic stimulation (TMS) have been increasingly used to treat a wide range of psychiatric and neurological disorders [1]. Recent studies have shown that personalized functional MRI (fMRI)guided neuromodulation may outperform conventional anatomyguided stimulation for patients with treatment-resistant major depressive disorder (MDD) [2,3]. In these studies, subject-specific targets identified in the dorsolateral prefrontal region by their robust functional anti-correlations with the subgenual anterior cingulate cortex (sgACC) achieved better treatment outcomes than those target regions showing scant sgACC anti-correlations. This empirically-driven approach to TMS target identification is advantageous in MDD, but for a variety of brain disorders without an empirically defined circuit, identification of personalized TMS targets is challenging. We designed a strategy for personalized brain stimulation based on abnormal RSFC assessed by resting-state fMRI (rsfMRI) and applied it to a 21-year-old patient suffering from severe motor, cognitive and mood impairments due to brain trauma. We stimulated the personalized target using TMS and evaluated the RSFC changes along with symptom improvement following treatment. Here, we leveraged fMRI and TMS to provide proof of concept for personalized fMRI-guided closed-loop stimulation that includes functional brain assessment for target identification, neuromodulation, and reassessment of functional connectivity.

Case report
The patient was a 21-year-old male snowboard athlete who sustained severe trauma during training that resulted in an intracranial venous sinus thrombosis, cerebellar hematoma, and obstructive hydrocephalus. A structural MRI revealed a large lesion along his cerebellar vermis that extended into both cerebellar hemispheres (Nov 20th, 2019, Fig. 1A). The patient suffered from a severe disorder of consciousness and underwent craniotomy with hematoma evacuation twice within a week. After stabilization of vital signs, the patient was transferred to our hospital (Feb 19th, 2020) and underwent a rigorous rehabilitation regimen to Jun 8th, 2020 (6.5 months following the initial injury, see supplementary methods). However, his recovery began to plateau and he was still not able to complete ambulation exercises without assistance (see Video. S1 for walking performance). Neuropsychological tests indicated significant cognitive deficits and mood dysfunction (Tables S1 and S2 and Fig. 1C).
To map the patient's functional network organization and assess the integrity of functional connectivity, we collected 24-min resting state fMRI data and parcellated the patient's cerebral cortex into 18 individual-specific functional networks (see supplementary methods) [4]. A functional abnormality map was generated (see supplementary methods) by comparing the patient's fMRI with 1000 healthy subjects' data. High abnormality was observed in the left DMPFC region (Fig. 1B), which is functionally associated with multiple cognitive and emotional symptoms [5e7] and is anatomically connected to the cerebellum via cerbello-thalamocortical pathways [8,9]. Stimulation to the DMPFC might impact the functional networks and cortico-cerebellar circuits associated with aforementioned symptoms. On Jun 8th, 2020, treatment began with an efficient form of high-frequency TMS known as intermittent theta burst stimulation (iTBS) using a Magstim Rapid magnetic stimulator (Magstim, Whitland, U.K.) with a figure-ofeight coil placed over the left DMPFC (Fig. 1B). Each iTBS session contained a total of 600 pulses (50 Hz triplets delivered at 5 Hz) at 90% resting-state motor threshold (see supplementary methods).
After 20 sessions of iTBS extending across a period of 4-weeks, the patient was administered the same battery of neuropsychiatric tests and fMRI examinations. We found remarkable improvements in his general cognitive functioning and mood, with 38.3e84.6% increases in testing scores ( Fig. 1C and Tables S1 and S2). In addition, the patient was able to ambulate independently. Remarkably, over the course of three months following iTBS treatment, the patient's motor abilities continued to improve to the point that he was able to run without assistance (see Video. S2 for running performance). At the one-year follow-up visit (June of 2021), the patient's cognitive functions showed steady or continued improvement compared to the baseline, and the mood scores showed a modest increase, with 46.2%e97.5% increases in testing scores (Fig. 1C).
To determine how the above-mentioned symptom improvements related to changes in the RSFC, we placed seeds in the PCC, canonical parietal nodes of the default mode network (DMN). PCC showed significant improvement in long-distance RSFC and anticorrelations with attention networks (Fig. 1D, S1A-B; p < 0.001, Wilcoxon signed rank tests). The absence of anti-correlations between DMN and attention networks has previously been linked with symptoms associated with various brain disorders [10]. The appearance of long-distance correlations and anti-correlations within the DMN and attention networks after iTBS might be  associated with the patient's improvements in cognitive functioning and mood. Moreover, the RSFC between the primary motor cortex and contralateral cortical and cerebellar sensorimotor areas emerged after iTBS treatment (Fig. 1E, S1C-D; p < 0.0001, Wilcoxon signed rank tests), which temporally coincided with patient's improvements in motor function. Although the mechanisms of neural recovery are not yet understood, it may be that the tailored stimulation led to cortical and cerebellar neuroplastic changes above and beyond the threshold that could be achieved from conventional rehabilitation, thereby overcoming the local optimal state (i.e., the plateau in recovery) to reorganize functional networks on a global scale, which might explain the continued improvements observed in both motor functions after treatment.
In summary, we present a novel pipeline to detect functional abnormalities using fMRI which can guide neuromodulation, and to assess normalization of the brain networks following the treatment. When applying this pipeline to a case with severe functional disruption, our approach shows sensitivity in its ability to detect functional abnormalities in RSFC prior to treatment and normalization of these network abnormalities following fMRI-guided TMS modulation, which temporally coincided with improvements in cognition, mood and motor function. The practical application of our approach extends RSFC from a research tool for retrospective studies to a practical tool for threading the pipeline. Furthermore, the post-treatment assessments based on RSFC closes the loop along the continuum from initial diagnosis, to treatment, and final assessment. The conceptual framework that combines rsfMRI and TMS interconnects each of these components into a personalized, non-invasive, closed-loop brain stimulation system.

Declaration of competing interest
HL is on the chief scientific advisory board for Neural Galaxy LLC and is listed as an inventor in issued patents on brain imaging that are unrelated to this work.