How to Use MRI to Optimize Brain Stimulation (tES/TMS/tFUS)

Non-invasive brain stimulation (NIBS) technologies, including transcranial electrical (tES), focused ultrasound (tFUS) and magnetic (TMS) stimulation, have emerged as promising interventions in human brain mapping. There is a growing effort to develop individualized interventions using MRI data, reinforced recently by FDA approval of a new TMS protocol for depression that includes individualized fMRI-based targeting along with other modifications with higher reported effect size compared to previous “one size fits all” protocols. This talk gives an overview on how neuroimaging can inform NIBS. We will also discuss the dimensions for individualizing tES/tFUS/TMS protocols to enhance efficacy. We propose a multifaceted approach of personalizing NIBS using neuroimaging, considering four levels: (1) context, (2) target, (3) dose, and (4) timing. We will introduce and share a single subject multimodal MRI database (structural and functional MRI and DTI) for hands-on experiences and discuss analytic pathways to use this data for individualized tES/tFUS/TMS using a polling system. By addressing inter- and intra-individual variability, neuroimaging informed brain stimulation encourages broader and more systematic adoption of personalized NIBS techniques to improve interventional outcomes in human brain mapping. 

Using image guided transcranial magnetic stimulation, it is possible to directly target functional networks associated with specific behavioral processes. The contributions of image-guided TMS and other neuromodulatory techniques to cognitive neuroscience stem from their roots in psychology and neuroscience, which are aimed at understanding brain-behavior relationships. However, this causal intervention approach builds on correlational methods in cognitive psychology and neuroscience and allows researchers to directly test the brain-behavior hypotheses generated by these fields. The purpose of this talk is to highlight the strengths and challenges of using image-guided TMS to test brain-behavior hypotheses in cognitive neuroscience using motor learning, working memory, and emotional processing as example systems. The talk will focus on understanding how the large parameter space afforded by image-guided TMS allows for robust experimental control, but also requires careful consideration of the experimental design characteristics.
 

Transcranial magnetic stimulation (TMS) is a form of noninvasive brain stimulation that is FDA-cleared in the United States for treatment of major depression, obsessive-compulsive disorder, and nicotine dependence. TMS uses electromagnetic fields to temporarily alter patterns of neuronal activity. Importantly, TMS exerts both local effects at the stimulation site and downstream effects across functionally connected networks. The overwhelming majority of TMS protocols used for clinical treatment use targets based on scalp landmarks, which are imprecise and may not be optimal. The use of MRI to identify TMS targets allows for greater precision and a wider array of targets. This talk will provide an overview of various MRI targeting methods, including both structural (T1, DTI) and functional (resting-state and task) methods. This talk will also discuss the use of personalized rTMS targets and will show examples of personalized network-targeting and traditional MNI-based targeting using neuronavigation software. We will also discuss how MRI can be used to measure TMS effects, including timing of post-TMS neuroimaging and the appropriate MRI modality. Participants will engage in small group discussions about their own study designs and potential neuroimaging-based targeting approaches. 

Forms of non-invasive transcranial electrical stimulation using low-intensity currents are widely applied to investigate brain-behaviour relationships in humans and as therapeutic applications for brain disorders. Advantages of these techniques include their good side-effect profile, relatively low cost, opportunities for home usage and possibility for a wide range of stimulation patterns (by varying for example waveforms, location and timings). However, because their effects on brain function tend to be more nuanced and associated with modulations of the timings or levels of brain activity, their combination with neuroimaging is key to improve their applicability, efficiency and basic understanding of mechanism of action in the human brain.
During this talk I will illustrate how brain imaging techniques, particularly fMRI and EEG can shape our understanding of the mechanisms by which transcranial electrical stimulation acts on brain function and how they can be used to inform target engagement and the selection of more effective stimulation parameters.
 

Transcranial low-intensity focused ultrasound (FUS) is an emerging noninvasive neuromodulation tool that enables focal and highly precise stimulation of brain targets, particularly deep brain structures. Ultrasound neuromodulation can be achieved using either a wearable single ultrasound transducer or more sophisticated transducer arrays, each offering distinct technical advantages depending on the location, volume, and precision of the targeted brain area. Its clinical applications, such as in the management of chronic pain and mental disorders, are still in their early stages. Since ultrasound energy attenuates as it penetrates the skull, target planning typically relies on simulations. However, due to the inhomogeneity of skull thickness and bone density, it is desirable in clinical practice to monitor the real-time location of the FUS beam in the brain using MR-ARFI (magnetic resonance acoustic radiation force imaging) and to assess the functional engagement of the brain target with functional MRI. During my talk, I will review our recent technical developments in MR-guided FUS systems and the use of MR-ARFI for real-time tracking of the FUS beam at various brain targets. I will also discuss the exploration of FUS parameters for modulating pain circuits and their associated functional pathways in a large animal model. These preclinical studies provide a unique opportunity to identify optimal targets and refine dosage and stimulation paradigms before progressing to clinical trials. 

Individualized fMRI-based, precision targeting has been highlighted in recent studies and FDA-approvals, higher reported effect sizes compared to previous approaches. In addition to neuroimaging-based precision approaches, broader and deeper stimulation profiles have also shown promise, particularly in the area of addiction. These broader and deeper stimulation techniques may increase and accelerate clinical applications for a wider range of patients. Additionally, more recent studies have highlighted the importance of context and neural circuit engagement during stimulation. This is an ideal bridge for the NIBS field to transition from neuroscience mechanisms to clinical application of neuroimaging informed brain stimulation. In this talk I will provide a step by step guide on how to develop provocation protocols to engage the neural circuit of interest for neuromodulation procedures, and those that may be of most relevance to clinical applications. In addition, I will highlight methods to verify neural target engagement and modulation and how that relates to clinical outcomes, such as relapse, in the context of addiction.
 

Validating target circuit engagement is essential for enhancing the precision and efficacy of brain stimulation. Concurrent TMS-MRI combines the spatial resolution of MRI with the real-time neurophysiological insights of TMS, allowing for direct measurement of brain response and validation of target engagement. Using examples from basic research in cognition, I will demonstrate how this approach can reveal network dynamics underlying key brain functions, such as memory and language. Additionally, I will highlight clinical applications, showing how target engagement can be linked to treatment response in depression. Practical considerations, including optimizing scanner parameters and experimental protocols to improve data quality, will also be discussed. Through interactive discussions and real-world examples, participants will gain strategies for effectively integrating TMS-MRI into their own research. 

Functional MRI (fMRI) is frequently used to assess modulation of brain activity by transcranial electrical stimulation (tES). However, concurrent tES-fMRI is technically challenging: first, it requires MRI-compatible tES equipment; second, the use of electric currents inside the MR scanner can lead to artifacts on the fMRI data; third, the observed brain activity can be due to direct effects of the currents on neural activity, but also be influenced by side effects. In this course, we will give an overview of methodological and experimental aspects of concurrent tES-fMRI with a focus on the requirements for an unambiguous interpretation of the observed BOLD activity patterns. We set specific emphasis on the impact of the electric current flow on echo-planar imaging (EPI) that is the standard sequence for fMRI due to its high temporal resolution, despite the well-known susceptibility to field inhomogeneities which cause geometric distortions mainly in the phase-encoding direction. These are well-known to cause severe distortions in regions of air-tissue interfaces. However, we show that subtle effects caused by the tES current-induced magnetic fields can also be seen. This issue has been demonstrated in subjects post-mortem, but is otherwise largely neglected. We show that the small in-vivo brain currents indeed induce negligible magnetic fields, but that the magnetic fields induced by the currents in the electrode cables can in fact cause artifacts comparable to the blood-oxygen-level-dependent (BOLD) contrast. We will explain how these artifacts arise during acquisition and processing of the fMRI data and discuss how to design more robust experiments. In particular, examples of artifact-prone cable configurations and susceptible fMRI paradigms will be presented, followed by an open discussion on best-practices for mitigating their impact on the fMRI analysis. 

A variety of approaches to interleave functional MRI with TMS have been accomplished and yield complementary evidence for circuit engagement and modulation. The include gaps between TRs to prevent corruption of images, throwing out corrupted fMRI volumes if acquired concurrently with TMS, and even inter-slice TMS delivery. Triggering the scanner and the TMS box can be managed with a single host computer running a variety of software (psychopy, eprime, matlab, etc.). Controlling for non-specific effects of stimulation is typically carried out by comparing other active stimulation sites given that there is no commercially available sham TMS compatible with the MRI environment. Using fMRI to guide TMS positioning extends beyond the TMS/fMRI environment but is an important consideration in deciding how to do a valid TMS/fMRI experiment. Considerations including the strength of the evidence for the target, the amount and type of imaging data, weighting individualized targets based on atlases or coordinates from the imaging literature, as well as practical considerations such as the likelihood of causing scalp pain at a particular location and the distance of the brain from the scalp. The audience will be shown several fMRI ‘blobs’ from a real participant and will be polled on which target they would choose. This will be compared with the target we actually selected from the experiment along with a rationale for the choice. 

Emerging research indicates that many psychiatric disorders are characterized by dysfunctional brain circuits. In this talk I will briefly reiterate how this circuit-based framework has reshaped and improved the capacity to identify optimal brain stimulation targets for psychiatric disorders. I will illustrate how these circuits can be targeted on a personalized basis and briefly overview the evidence for this approach from basic and clinical research. I will then detail practical considerations around scanner parameters and neuronavigation accuracy that have implications for personalized and circuit-based targeting. Lastly, I will briefly highlight hardware solutions that provide a cheaper alternative to neuronavigation where desired.