Computational Models of High-Definition Electroconvulsive Therapy for Focal or Multitargeting Treatment

Abstract

Attempts to dissociate electroconvulsive therapy (ECT) therapeutic efficacy from cognitive side effects of ECT include modifying electrode placement, but traditional electrode placements employing 2 large electrodes are inherently nonfocal, limiting the ability to selectively engage targets associated with clinical benefit while avoiding nontargets associated with adverse side effects. Limited focality represents a technical limitation of conventional ECT, and there is growing evidence that the spatial distribution of the ECT electric fields induced in the brain drives efficacy and side effects. Computational models can be used to predict brain current flow patterns for existing and novel ECT montages. Using finite element method simulations (under quasi-static, nonadaptive assumptions, 800-mA total current), the electric fields generated in the superficial cortex and subcortical structures were predicted for the following traditional ECT montages (bilateral temporal, bifrontal, right unilateral) and experimental montages (focal electrically administered seizure therapy, lateralized high-definition [HD]-ECT, unilateral 4 × 1-ring HD-ECT, bilateral 4 × 1-ring HD-ECT, and a multipolar HD-ECT). Peak brain current density in regions of interest was quantified. Conventional montages (bilateral bifrontal, right unilateral) each produce distinct but diffuse and deep current flow. Focal electrically administered seizure therapy and lateralized HD-ECT produce unique, lateralized current flow, also impacting specific deep regions. A 4 × 1-ring HD-ECT restricts current flow to 1 (unilateral) or 2 (bilateral) cortical regions. Multipolar HD-ECT shows optimization to a specific target set. Future clinical trials are needed to determine whether enhanced control over current distribution is achieved with these experimental montages, and the resultant seizures, improve the risk/benefit ratio of ECT.

Electroconvulsive therapy (ECT) is an established treatment for severe mood and psychotic disorders, with its use supported by evidence of efficacy and safety.^1–6 Ongoing improvements in outcomes and side effects have been attributed to refinements in ECT dose,^7,8 namely, individualizing total charge based on seizure threshold, reducing pulse width, and using more focal electrode placement.^9–13 Existing ECT electrode montages in clinical use today vary electrode placement and shape (pad or disk) but are all bipolar (ie, 2 electrodes) and relatively large in size and interelectrode distance.

Across forms of neuromodulation, computational models of current flow support the understanding of individual differences in the dose that each patient actually receives and can suggest means of dose optimization.^14–18 Models of ECT have focused on conventional bipolar montages including bifrontal (BF), bilateral (BL), and right unilateral (RUL),^19–25 most of which use either pad (~21 cm²) or disk (~20 cm²) electrodes. Bipolar montage variations have been simulated.²⁶ Focal electrically administered seizure therapy (FEAST) ECT uses a bipolar electrode montage with a ~3.2 cm² frontal electrode and ~16 cm² return electrode.^11,27 Lee et al (2012) demonstrated that FEAST has a distinct spatial distribution with higher E-field strength in the right prefrontal cortex than conventional montages.²³ Lee et al (2016) further demonstrated that a novel frontomedial electrode placement was better than FEAST at targeting prefrontal structures while avoiding the medial temporal lobe, thought to be associated with cognitive side effects.²⁸ Conventional and experimental montages like FEAST and frontomedial placement influence distinct but significant fractions of cortex and varied deep brain structures. Prior modeling studies found that differences in the brain current flow patterns were associated with individual differences in ECT outcomes.^29–31

Reduced electrode size such as high-definition (HD) electrodes³² were developed for transcranial electrical stimulation (HD-tES), and their use in arrays for brain targeting was demonstrated first in computational models^33,34 and then experimental validation.^35–39 Small electrodes enable placing electrodes closer together (eg, proximal bipolar HD^40,41) and in multichannel arrays (eg, 4 × 1 ring HD³³). By variation in the number, position, and current-per HD electrodes, HD-tES can be customized for a range of applications varying from maximizing focality at superficial^33,42 or deep regions,^43–45 current flow direction at a target gyrus,^40,41,46 or maximizing intensity to one³⁴ or more targets.⁴⁷ HD transcranial direct current stimulation and HD transcranial alternating current stimulation are now tested expansively, including in clinical trials^48–52 and in leveraging active controls to dissect brain-behavior relationships.^53–55

Recently, a version of HD electrodes was developed supporting high-intensity current pulse trains, as used in ECT. Could novel ECT electrode montages using HD electrodes (HD-ECT) be developed to leverage the varied targeting advantages of HD-tES approaches? To address this question, we applied finite element method (FEM) stimulations (under quasi-static, nonadaptive assumptions, fixed total current) of the electric fields induced on the head and brain by traditional (BF, bitemporal [BT], RUL) and experimental focal (FEAST, lateralized HD-ECT, unilateral 4 × 1-ring HD-ECT, BL 4 × 1-ring HD-ECT, and a multipolar HD-ECT) ECT montages. We characterized the comparative brain anatomy engaged by the electric field of each ECT protocol and quantify the effects on specific nodes relevant to the pathophysiology of mood disorders (ie, putative targets) and the adaptive processing of memory (putative antitargets): subgenual cingulate cortex, insula, amygdala, hippocampus, nucleus accumbens, and midbrain (including the ventral tegmental area and substantial nigra). Such analyses predict only the current flow patterns (electric fields) in the brain for each electrode montage and not functional or behavioral consequences. Nonetheless, consideration of relative superficial and deep brain targeting by conventional ECT and novel HD-ECT montages is a step in rigorous consideration of candidate interventions prior to their selection for validation.

METHODS

We adapted an anatomical model of human head based on an optimized high-resolution (1 mm³) MRI scans and segmented the head into compartments representing scalp/skin, fat, skull, CSF, gray-matter, white-matter, air, cingulate gyrus, basal ganglia, corpus callosum, thalamus, subthalamic nucleus, fornix, nucleus accumbens, hippocampus, amygdala, midbrain, mammillary bodies, pons, medulla oblongata, and insula. Computer-aided design models of montage-specific ECT electrodes of various shapes and dimensions were modeled in Solidworks (Dassault Systemes Corp., MA) and later imported into the Simpleware (Synopsys, CA) to position them over the targeted brain region. The volumetric mesh generated from a voxel-based tetrahedral adaptive meshing algorithm was eventually exported to COMSOL Multiphysics 5.5 (COMSOL Inc., MA) to computationally solve the electric field distribution (Fig. 1). The resulting mesh across all simulated electrode montages comprised of > 12 M tetrahedral elements.

FIGURE 1.

Modeling pipeline with an exemplary ECT montage. (A) Shows segmented deeper brain structures (color coded) in an MRI. (B) Shows an exemplary conventional ECT BF montage with color coded anode (red) and cathode (blue). (C) Illustration of current flow via current density streamlines during ECT. (D) Predicted electric field profile at the cortex. (E) Sagittal, coronal, and axial cut of human head showing electric field at the subcortical structures.

An isotropic electrical conductivity (S/m) of the brain tissues were assigned as: scalp/skin: 0.465, fat: 0.025, skull: 0.01, CSF/meninges: 0.8, gray-matter: 0.276, white-matter: 0.126, air: 1 × 10⁻¹⁵, with remaining tissues, cingulate gyrus, basal ganglia, corpus callosum, thalamus, subthalamic nucleus, fornix, nucleus accumbens, hippocampus, amygdala, insula, mammillary bodies: 0.02, pons, medulla oblongata, midbrain: 0.04.^56,57

We modeled three conventional bipolar montages (BF, BL, and RUL), 2 lateralized montages (FEAST, lateralized HD with a large HD electrode on the ipsilateral and 4 standard HD electrodes on the contralateral hemisphere), unilateral and BL 4 × 1-ring high-definition (4 × 1 HD) each ring consisting of 1 center large HD electrode and 4 standard HD electrodes, and a multipolar HD montage (with 9 standard HD electrodes). Across all montages, electrodes were overlayed on a thin (~2-mm thickness) conductive gel (0.018 S/m^25,32). All electrodes were modeled with 1-mm thickness and assigned a metal conductivity of 5.9 × 10⁷ S/m.⁵⁸ A circular electrode (Ø = 5 cm) was modeled and positioned across all the conventional ECT montages. HD montages used “standard” HD (Ø = 1 cm) or “large” HD (Ø = 2 cm) electrodes. Complete electrode montage detailed follows:

Conventional ECT Montages

Bifrontal

Electrodes were positioned bilaterally at 5 cm above the outer angle of the orbit on a line parallel to the sagittal place (Fig. 1).^19,28

Bilateral

Electrodes were positioned frontotemporally at 2.5 cm above the midpoint of the line connecting the external canthus and tragus (Fig. 1)^19,28

Right Unilateral

One electrode was positioned at 2.5 cm to the right vertex and the second electrode was positioned over the right frontotemporal (Fig. 1).^19,28

Lateralized ECT Montages

Feast ECT

A circular anterior electrode (Ø = 2 cm) was positioned with lower boundary just above the center of the right eyebrow, whereas the posterior rectangular electrode (2.5 cm × 6.3 cm) was positioned over the right motor strip (Fig. 2).²⁸

FIGURE 2.

Predicted electric field profile using a conventional ECT BF (top), Bilateral (middle), and Right unilateral (bottom) montage. (A1) BF montage current flow from anode (left, red) to cathode (right, blue) via current density streamlines. (B11, B12, B13, B14) shows predicted electric field across the cortex. (C1, D1, E1) is a sagittal, axial, and coronal view of predicted electric field, respectively. (A2) Bilateral montage current flow from anode (left, red) to cathode (right, blue) via current density streamlines. (B21, B22, B23, B24) shows predicted electric field across the cortex. (C2, D2, E2) is a sagittal, axial, and coronal view of predicted electric field, respectively. (A3) Right unliteral montage current flow from anode (top, red) to cathode (right, blue) via current density streamlines. (B31, B32, B33, B34) shows predicted electric field across the cortex. (C3, D3, E3) is a sagittal, axial, and coronal view of predicted electric field, respectively.

Lateralized HD-ECT

In this montage, a large HD electrode (anode) was positioned on the ipsilateral and 4 standard HD electrodes (cathode) were positioned on the contralateral hemisphere (Fig. 2).

4 × 1-Ring HD-ECT Montages

Unilateral 4 × 1-Ring HD-ECT

A concentric-ring configuration using 4 standard HD electrodes (cathode) and 1 large center HD electrode (anode) was deployed in a circular fashion³³ over the right motor cortex. The interelectrode distance between the large HD center electrode (5.1 cm diameter) and surrounding cathode HD electrodes was 3 cm (Fig. 3).

FIGURE 3.

Predicted electric field profile using lateralized ECT montages (FEAST) (top), lateralized HD-ECT (bottom). (A1) FEAST current flow from anode (circular, red) to cathode (blue, rectangular) via current density streamlines for FEAST montage. (B11, B12, B13, B14) shows predicted electric field across the cortex. (C1, D1, E1) is a sagittal, axial, and coronal view of predicted electric field, respectively. (A2) Lateralized HD-ECT current flow from a large HD anode (right, red) from ipsilateral hemisphere to 4 standard HD cathode electrodes (left, blue) on the contralateral hemisphere via current density streamlines. (B21, B22, B23, B24) shows predicted electric field across the cortex. (C2, D2, E2) is a sagittal, axial, and coronal view of predicted electric field, respectively.

Bilateral 4 × 1-Ring HD-ECT

In this montage, two 4 × 1-ring HD-electrodes (each ring consisting of 1 large center HD electrode and 4 standard HD electrodes) were positioned over each brain hemisphere (Fig. 3).

Multipolar HD-ECT Montage

We further investigated the possibility of introducing a multitarget optimized montage with multiple targets of the hippocampus, amygdala, and subgenual cingulate. This optimized HD-ECT montage comprised 9 standard HD-electrodes positioned at the FP1, FC4, FPZ, F8, P8, C3, TP7, EX5, and EC2 according to 10–10 EEG system (Fig. 4).

FIGURE 4.

Predicted electric field profile using a unilateral and BL 4 × 1-ring HD-ECT. Each 4 × 1-ring consisting of a large center HD electrode and 4 standard HD electrode. (A1) Unilateral 4 × 1-ring HD current flow from a large center HD anode (red) to 4 standard HD electrodes (blue) via current density streamlines in a unilateral 4 × 1-ring HD-ECT montage. (B11, B12, B13, B14) shows predicted electric field across the cortex. (C1, D1, E1) is a sagittal, axial, and coronal view of predicted electric field, respectively. (A2) Bilateral 4 × 1-rings current flow, where applied current was evenly distributed through a large center anode (red) HD electrode to 4 standard HD electrodes on each hemisphere, shown via current density streamlines. (B21, B22, B23, B24) shows predicted electric field across the cortex. (C2, D2, E2) is a sagittal, axial, and coronal view of predicted electric field, respectively.

The Laplace equation (∇ (σ∇ V) = 0, where V is an electric potential and σ is conductivity) was solved under steady state assumption to predict Electric field magnitude at different brain regions. For boundary conditions, a static inward normal current density (J_norm) corresponding to 800 mA total was applied to the exposed surface of the anode electrode(s) whereas ground was applied to the exposed surface of the cathode electrode. Note that input current was divided evenly through anode electrodes (400 mA each) in the dual 4 × 1 HD-ECT Bilateral montage with larger center electrode. In the case of an optimized ROI-specific HD-ECT, the current was injected via anode electrode positioned at FP1 (289.88 mA), P8 (350.44 mA), and FPZ (159.68 mA) whereas C3 (−262.40 mA), F8 (−75.08 mA), FC2 (−801.08 mA), FC4 (−197.2 mA), TP7 (−55.72 mA), and EX5 (−128.52 mA) were grounded (cathode).

The remaining external boundaries of the model were electrically insulated. The relative tolerance was set to 1 × 10⁻⁶ to improve solution accuracy and solve the finite element method (FEM) model in COMSOL using the linear system solver of conjugate gradients. Note for lateralized montages, we considered electric fields in subcortical structures for each hemisphere (Table 1). Our goal was limited to simulating the relative focality of the considered montages under assumptions of tissue isotopy,^59,60 quasi-static nonadaptive tissue properties^24,25 including no dispersion,^61,62 quasi-uniform assumption of neuromodulation by electric fields,^56,63,64 and so considering only instant peak current flow during a pulse (anode/cathode distinctions are only for specifying how intensity controlled). In representing current flow patterns for each class of montages (conventional, lateralized, 4 × 1-ring, multipolar), we plotted electric fields up to >150 V/m or >300 V/m to support visual comparisons, although absolute peaks are noted in the text and tables. Unless otherwise stated, the peak electric field represents 99th percentile of electric field produced in a region of interest (ROI). Peak electric fields exclude magnitudes observed in <5% of an ROI to minimize the contributed edge effects.

TABLE 1.

Predicted Electric Field at Different Brain Structures by the Conventional, Lateralized, 4 × 1-Ring HD, and a Multipolar HD-ECT Montages

	Brain Regions Peak E-Field (V/m, % Cortex)
		ROI
Montages	Cortex	Subgenual Cingulate	Hippocampus	Amygdala	Insula	Midbrain	Accumbens
Conventional
BF	311	327, 105%	204, 66%	165, 53%	600, 193%	190, 61%	579, 186%
BT	273	317, 116%	380, 139%	280, 103%	996, 365%	338, 124%	998, 366%
RUL	275	194, 71%	250, 91%	130, 47%	566, 206%	139, 51%	400, 145%
Experimental
FEAST	342	R:207, 60%; L:186, 54%	R:94, 27%; L:50, 15%	R:127, 37%; L: 97, 28%	R:335, 98%; L:145, 42%	R:200, 58%; L:123, 36%	R:274, 80%; L: 267, 78%
HD-ECT
Lateralized HD-ECT	320	R:189, 59%; L:182, 57%	R:232, 73%; L:177, 55%	R:141, 44%; L: 130, 41%	R:556, 174%; L:512, 160%	R:198, 62%; L:110, 34%	R:363, 113%; L:350, 109%
4 × 1-ring HD-ECT
Unilateral 4 × 1-ring HD-ECT	184	7, 3%	22, 12%	10, 5%	75, 41%	30, 16%	23, 13%
Bilateral 4 × 1-ring HD-ECT	90	5, 5%	13, 14%	6, 7%	38, 42%	19, 21%	13, 14%
Optimized multipolar
Anatomically targeted multipolar HD-ECT	153	89, 58%	129, 84%	79, 52%	242, 158%	67, 44%	153, 100%

Electrical fields in subcortical structures are considered for each hemisphere for lateralized montages. Peak electric field is also reported as % peak cortex for each subcortical structure. L and R refer to the left and right hemisphere.

RESULTS

Computational models of current flow are standard and validated tools in neuromodulation analysis and therapy design. Here we predict brain current flow patterns produced by conventional ECT montages (BF, BL, RUL), lateralized ECT montages (FEAST, lateralized HD-ECT with 1 large HD electrode on ipsilateral and 4 standard HD electrodes on the contralateral hemisphere), 4 × 1-ring HD-ECT with 1 large center HD electrode and 4 standard HD electrodes (unilateral, BL with two 4 × 1 rings), and a multipolar HD-ECT. In each case, we predict the electric field across the cortex and subcortical structures such as hippocampus, amygdala, anterior and subgenual cingulate gyrus, insula, midbrain (ventral thalamus and substantia nigra), and nucleus accumbens. With the limited goal of comparing current flow patterns across ECT montages (under quasi-static and quasi-uniform assumptions; see Methods), comparisons are normalized by considering 800-mA total current for each montage.

The predicted peak electric field using the conventional ECT montages at the cortex and subcortical structures as % cortex (subgenual cingulate, hippocampus, amygdala, insula, midbrain, and accumbens) were 311 V/m, 327 V/m (105%), 204 V/m (66%), 165 V/m (53%), 600 V/m (193%), 190 V/m (61%), and 579 V/m (186%), respectively for the BF montage; 273 V/m, 317 V/m (116%), 380 V/m (139%), 280 V/m (103%), 996 V/m (365%), 338 V/m (124%), and 998 V/m (366%), respectively for the BL montage; and 275 V/m, 194 V/m (71%), 250 V/m (91%), 130 V/m (47%), 566 V/m (206%), 139 V/m (51%), and 400 V/m (145%), respectively for the RUL montage (Fig. 2, Table 1). Among the conventional montages, BL produced a higher peak electric field across cortical and subcortical structures, with the lowest relative sparing of deep structures (Table 1).

For simulated hemisphere, the lateralized ECT montage FEAST predicted peak electric field and % cortex were 342 V/m in cortex, 207 V/m (R), 60% and 186 V/m (L), 54% in subgenual cingulate, 94 V/m (R), 27% and 50 V/m (L), 15% in hippocampus, 127 V/m (R), 37% and 97 V/m (L), 28% in amygdala, 335 V/m (R), 98% and 145 V/m (L), 42% in insula, 200 V/m (R), 58% and 123 V/m (L), 36% in midbrain, and 274 V/m (R), 80% and 267 V/m (L), 78% in accumbens (Fig. 3). For simulated hemisphere, the lateralized HD-ECT montage predicted peak electric fields were 320 V/m in cortex, 189 V/m (R), 59% and 182 V/m (L), 57% in subgenual cingulate, 232 V/m (R), 73% and 177 V/m (L), 55% in hippocampus, 141 V/m (R), 44% and 130 V/m (L), 41% in amygdala, 556 V/m (R), 174% and 512 V/m (L), 160% in insula, 198 V/m (R), 62% and 110 V/m (L), 34% in midbrain, and 363 V/m (R), 113% and 350 V/m (L), 109% in accumbens (Fig. 3). Both lateralized ECT montages produced lateralized electric field; however, the intensity was higher in subcortical structures for the lateralized HD-ECT montage compared with the FEAST montage. Peak electric field at the cortex was comparable to the conventional ECT montages (Fig. 3, Table 1).

The peak electric field magnitude predicted by the unilateral 4 × 1-ring HD-ECT montage consisting of a large center HD electrode and 4 standard HD electrodes was ~2× higher in the cortex and subcortical ROIs compared to the BL 4 × 1-ring HD-ECT (Fig. 4, Table 1).

With the goal of maximizing electric field focality at specific subcortical ROIs that are clinically assessed for different psychiatric disorders, we developed an optimized multipolar HD-ECT montage by deploying 9 HD-electrodes (Fig. 5A1). The peak predicted electric field localized at the subgenual cingulate, hippocampus, and amygdala were 89 V/m, 129 V/m, and 79 V/m, respectively (Fig. 5, Table 1).

FIGURE 5.

Predicted electric field profile using a multipolar HD-ECT. (A1) shows current flow from standard HD-electrodes positioned over FP1, P8, and FPZ (anodes, red) to C3, F8, FC2, FC4, TP7, and EX5 (cathodes, blue) via current density streamlines. (B11, B12, B13, B14) shows predicted electric field across the cortex. (C1, D1, E1) is a sagittal, axial, and coronal view of predicted electric field, respectively.

DISCUSSION

In this modeling study, we explored the feasibility of altering brain current flow patterns during ECT using conventional and experimental montages. Our findings demonstrate that predicted electric field varies substantially across the montages assessed. Relative to the other montages evaluated, the unilateral 4 × 1-ring HD-ECT montage yielded the most focal electric field distribution and resulted in the least electric field in the hippocampus. The clinical implications of this relative gain in focality and sparing of hippocampus will need to be evaluated in future clinical trials (for review, see the study by Deng et al⁶⁵).

For safety and tolerability, we considered large (Ø = 2 cm) HD electrodes where current was centered (800 mA), and large or small (Ø = 1 cm) electrodes where current was split (200 mA per electrode with 4 ring electrodes). A prior tES study used standard (Ø = 1 cm) HD electrodes in a single 50-μs pulse applications for motor threshold determination.³⁶ Maximum current up to 1817 mA was safely administered to an individual without complications. This tES current amplitude is more than double the maximum ECT pulse current amplitude. Individual ECT pulses are longer (> 500 μs) and applied in trains. Using large HD electrodes (Ø = 2 cm) to delivery typically high-dose treatment with ECT (800 mA, 0.3-ms pulse width, 120-Hz frequency, 8-second train duration), the resultant charge density on the electrodes is 73 mC/cm². To provide a benchmark, conventional transcranial direct current stimulation procedures, with 5 × 7-cm² electrodes at 2 mA applied for 20 minutes, results in a charge density of 69 mC/cm².

Our analysis, though quantitative, is theoretical and subject to technical assumptions (including nonadaptive quasi-static current flow physics, and regional quasi-uniform representation of resulting electric fields). Notwithstanding a general heuristic that regions with minimal electric field are unlikely to be influenced and increasing electric fields increase likelihood of neuromodulation^66–68—the relation between brain current flow patterns and clinical outcomes is complex and may reflects local electric field or network changes, including seizure initiation and propagation.^{20,30,31,69,70} There is emerging evidence in support of the importance of the distribution and overall properties of generated electric fields (including as modified by individual anatomy) in determining the biological and clinical effects.^{7,29,30,67,71,72} Predictions of brain current flow patterns are a tool, with recognized limitations, in the design and interpretation of ECT^{7,22,24,26,62,73,74}—just as they have played a central role in innovation across neuromodulation.⁷⁵

A modeling approach as taken here does not account for factors such as differences in susceptibility to electric fields including seizure threshold (eg, charge / number of pulses), across brain regions. There are established differences in seizure threshold between the conventional montages considered here, and there are no such data with the novel montages. We also did not consider individual anatomical differences^{25,31,66,68,76} as our focus was showing broad differences in current flow patterns between ECT montages. Evidently, clinical outcomes must be established empirically. Modeling approaches should inform candidate ECT montages and details of how they should be evaluated (eg, dose titration).

Conventional ECT has limited anatomical specificity because the induced electric fields are diffuse, with associated generalized seizures; recent studies challenge whether such diffuse changes are necessary to achieve clinical benefit. ECT outcomes have been associated with specific brain regions such as the limbic temporal lobe (amygdala, hippocampus), the anterior cingulate cortex, or the ventral striatum.⁷⁷ In addition, biological changes seem to follow the laterality of stimulation, that is, unilateral ECT leads to unilateral effects while BL ECT leads to BL effects.^78,79 Structural studies have identified volumetric increases specific to nodes of the reward circuitry that correlate with the improvement in anhedonia and motivation after ECT.⁸⁰ Other studies have identified volumetric increases in limbic regions relevant to the pathophysiology of mood disorders and the adaptive processing of memory, though not always correlating with clinical outcomes.⁸¹ Similarly, studies using functional MRI,^82,83 diffusion MRI,^84,85 and MRI spectroscopy^78,86 have identified functional, tractographic and neurochemical changes respectively in concrete disease-relevant brain regions. While research on the region-specific biological consequences on ECT is ongoing, these studies—along with the generally recognized importance montage—support the importance of considering ECT regional electric fields.

The montages selected here do not imply their endorsement for clinical practice, nor do they represent a comprehensive set of novel approaches. Indeed, no given ECT montage is proposed as necessarily superior to any other—rather we show in principle how HD-ECT opens new prospects for testing. In regard to conventional montages there are subvariations in placement technique and electrode shape (pad vs the disks considered here). In regard to novel montages, the 4 × 1-ring approach can be displaced over any targeted cortical region, and peak electric field and area of stimulation titrated through electrode ring size⁴⁶. The HD multi target approach presented, while designed with consideration of specific deep brain targets (hippocampus, amygdala, and subgenual cingulate), represented one of many possible multitarget variations.^34,87 The models presented, while not definitive, represent a rational step toward the design of novel ECT montages to be evaluated in subsequent clinical trials. Our analysis demonstrates the principles of single (4 × 1) and double (two 4 × 1) cortical target ECT, as well as multitarget ECT, with specific implementations depending on clinical trials goals.

Data availability statement:

All data that support the findings of this study are included within the article (and any supplementary files).