In 31 patients, heartbeat-locked SEEG residuals did not increase with CSF or gray matter probability

Three prior studies found heartbeat-locked bipolar SEEG activity 200–600 ms after the ECG R peak that remained after linear ECG regression, but did not exclude pulse artifact. This preregistered analysis tested whether the residual increased with template CSF probability (pulse account) or gray matter probability (neural account). It used archived per-pair statistics from 31 patients with ds004100 SEEG data (3,095 pairs) and pseudo-beat controls. Tissue probabilities were sampled within 4 mm of each contact, and per-patient Spearman correlations were evaluated with two-sided sign-flip tests and Holm correction for two primary tests. Correlations were negative: residual versus CSF, mean rho = −0.081 (8/31 positive; p = 0.010, Holm p = 0.021); residual versus gray matter, mean rho = −0.110 (7/31 positive; p = 0.021, Holm p = 0.021). These signs opposed both preregistered predictions. The correlations did not differ significantly from pseudo-beat controls (p = 0.10 and 0.21), and neither remained significant after a post hoc exclusion of 254 pairs outside the template brain (p = 0.062 and 0.18; Holm p = 0.12 and 0.18). Thus, template tissue class at this spatial resolution did not distinguish the pulse and neural accounts. A secondary association showed larger residuals in superior pairs in 25/31 patients (rho = +0.132, p = 0.00025); the control showed no such association (rho = +0.019, p = 0.59). This was not a primary test, and vessel distance was not tested. Coordinates were template-space rather than patient anatomy, with probable errors of several millimeters; each bipolar pair spans about 5.6 mm. The analysis used 300 s per patient and 200 surrogates per pair (minimum p = 0.005), in epilepsy patients with one ECG lead, no patient MRI, and no vessel atlas.

stereo-EEG · heartbeat-evoked response · pulse artifact · intracranial EEG · tissue probability · epilepsy