Where you aim brain stimulation matters, and a gentler alternative to ECT met its bar

Depression treatment by device has been available for two decades, and the arguments about it have largely been about whether it works. Two randomised trials published in 2026 move past that question to sharper ones: where the stimulation should be aimed, and what it costs the patient afterwards.

Aiming

Transcranial magnetic stimulation for depression is conventionally targeted by measuring across the scalp — a rule of thumb that places the coil over roughly the right part of the prefrontal cortex in roughly the right patient. Functional imaging offers an alternative: find the individual's own connectivity target and aim at that.

A randomised trial tested the two approaches head to head in accelerated TMS. Forty participants — 22 female (55%), mean age 45.7 years (SD 15.2) — were randomised and treated [s1].

Median reduction on the Montgomery-Åsberg Depression Rating Scale was 24 points (IQR 19–28) with connectivity-based targeting against 18 points (IQR 10–23) with scalp-based targeting (P = .02) [s1]. The effect size was 0.8 as a Cohen d analog, with a 95% CI of 0.26 to 1.54, and the authors express the practical implication as a number needed to scan of 5 individuals [s1].

"Number needed to scan" is an unusual and useful framing: scan five patients to gain one additional meaningful outcome over the cheaper targeting method. Whether that is worth an MRI is a health-economics question the trial does not answer, and the authors position their results as something that could facilitate cost-effectiveness analyses and help power a confirmatory trial rather than as a settled case [s1].

The validation buried in the middle of the paper is the part likeliest to survive. Individualised targets were reproducible within an individual, with a split-half distance of 4.47 mm, and different between individuals, at 12.97 mm (P < .001) [s1].

That is the finding that makes the rest coherent. If individual targets were not stable within a person, connectivity targeting would be measuring noise. If they were not different between people, there would be nothing for individualisation to add and a scalp measurement would do. Both conditions hold, at roughly a threefold separation between within-person and between-person distances [s1].

Forty participants is small, the confidence interval on the effect size runs from 0.26 to 1.54, and this is one trial [s1]. It is a strong signal in need of confirmation, which is how the authors frame it.

What it costs

The second trial addresses the oldest problem in this field. Electroconvulsive therapy remains among the most effective treatments in psychiatry for severe depression, and the reason patients refuse it is memory.

Magnetic seizure therapy induces a seizure using magnetic fields rather than direct electrical current, with the aim of a more focal effect and less cognitive damage. This trial tested it against right unilateral ultrabrief pulse ECT — already the gentler mainstream option, not the older bilateral technique.

Between June 2018 and March 2024, 292 participants were enrolled, of whom 239 were randomly assigned, with three withdrawing before treatment [s2]. The ECT group had a mean age of 45.5 years (SD 14.8, range 18–78), 80% White, with 65 female (55%) and 54 male (45%); the magnetic seizure therapy group had a mean age of 44.5 years (SD 13.8, range 18–75), 100 (85%) White, with 70 female (60%) and 47 male (40%) [s2].

Enrolment concluded before reaching the intended sample size [s2]. That is a real limitation and it is disclosed in the findings rather than buried.

On efficacy, the difference in remission rates was 5.3% in favour of ECT — 27.8% against 22.5% for magnetic seizure therapy — which established non-inferiority on the trial's prespecified test (Z-test p = 0.048, 95% CI −4.4 to 14.9) [s2].

Two things are true about that result at once. Non-inferiority was met on the prespecified criterion. And the confidence interval runs from −4.4 to 14.9, with a p-value of 0.048 against a conventional 0.05 threshold, in a trial that stopped short of its target sample [s2]. This is non-inferiority demonstrated at the margin, not comfortably.

The safety difference is where the trial is most convincing. Worsening autobiographical memory occurred in 17.3% of ECT participants against 2.7% of magnetic seizure therapy participants (p = 0.0003) [s2]. Twelve participants in the ECT group and three in the magnetic seizure therapy group withdrew from treatment because of non-serious adverse events [s2].

A roughly sixfold difference in the specific harm that drives refusal of this treatment, at a p-value two orders of magnitude below the efficacy result, is a more robust finding than the remission comparison.

The authors' interpretation is that the overall risk-benefit profile supports considering magnetic seizure therapy as a first-line convulsive therapy in major depressive disorder, particularly for patients who refuse right unilateral ultrabrief ECT [s2]. That last clause is the honest scope. For a patient who has already declined ECT, an option with comparable remission and far less memory disturbance is a genuine addition. It is a weaker argument for displacing a treatment someone is willing to have, given remission was numerically lower.

The trial was funded by the National Institutes of Mental Health rather than a manufacturer [s2].

What both trials suggest

Neither is about a new mechanism. Both are about refinement — aim the existing coil better, or induce the existing seizure more focally — and in both cases the refinement produced a measurable gain.

Also worth noting: both trials are small by the standards of drug licensing. Forty participants in one, 239 randomised in the other, one of which did not reach its planned sample [s1][s2]. Device trials in psychiatry run at a scale that would be considered preliminary in cardiology, which is a structural feature of the field rather than a criticism of either study.

What to watch

Whether a confirmatory, adequately powered trial reproduces the connectivity-targeting advantage, and whether anyone publishes the cost-effectiveness analysis that decides if a scan per patient is affordable [s1]. And whether magnetic seizure therapy's memory advantage holds in a trial that reaches its intended sample size [s2].

This article describes published trial results. It is not medical advice, and decisions about treatment for severe depression belong with a clinician.

Sources

• [s1] Connectivity- vs Scalp-Based Targeting of Accelerated Transcranial Magnetic Stimulation for Depression: A Randomized Clinical Trial, JAMA Psychiatry, 2026;83(8):807–817.
• [s2] Confirmatory efficacy and safety trial of magnetic seizure therapy versus right unilateral ultrabrief pulse electroconvulsive therapy, The Lancet Psychiatry, 2026;13(5):376–386.