Authors :
Presenting Author: Anna Maslarova, MD PhD – New York University, Grossman School of Medicine
Jiyun Shin, PhD – New York University Langone Health
Andrea Navas-Olive, PhD – Institute of Science and Technology, Austria
Mihály Vöröslakos, MD PhD – New York University, Grossman School of Medicine
Hajo Hamer, MD – University Hospital Erlangen
Arnd Doerfler, MD – University Hospital Erlangen
Simon Henin, Ph.D. – New York University Langone Health
György Buzsáki, MD PhD – New York University, Grossman School of Medicine
Anli Liu, M.D. – New York University Langone Health
Rationale:
Hippocampal sharp-wave ripples (SPW-Rs) are high-frequency oscillations critical for memory consolidation. Despite extensive characterization in rodents, their detection in humans is limited by coarse spatial sampling, interictal epileptiform discharges (IEDs), and a lack of consensus on their localization and morphology. We identified distinctive features of SPW-Rs and IEDs in mice and applied these to optimize ripple detection in humans.
Methods:
We analyzed hippocampal recordings from 5 APP/PS1 transgenic mice and 13 surgical epilepsy patients (6 temporal, 7 extratemporal). The APP/PS1 model of Alzheimer’s disease reliably expresses hippocampal IEDs but lacks frequent seizures or hippocampal sclerosis. We used wide-coverage 1024-channel probes (SiNAPS, Neuronexus), to characterize the spectrotemporal features of rodent SPW-Rs and IEDs across hippocampal subfields. In patients implanted with hybrid macro-/microwire hippocampal electrodes (Behnke-Fried, AdTech), hippocampal subfield locations were identified via CT-MRI co-registration and ASHS segmentation. Automatic event detection was performed on band-pass filtered signals (20-80 Hz for IEDs, 130-200 Hz for rodent SPW-Rs, 80-250 Hz for human ripples). We used the spectral, spatial, and morphological features of rodent SPW-Rs and IEDs to guide detection of SPW-Rs and IEDs in human recordings.
Results:
In mice, SPW-Rs were characterized by 1) localization to the CA1 pyramidal layer with rapid decay of ripple power with distance; 2) multiple consecutive ripple cycles visible on the raw LFP, and 3) narrowband spectral peaks around 160 Hz. IEDs, by contrast, showed: 1) widespread propagation across all hippocampal subfields 2) broad spectral power (20–400 Hz), and 3) large-amplitude sharp transients lacking ripple cycles (Fig.1).
These findings revealed that channel location, false positives from filtered IEDs and inspection of the raw LFP waveforms may be leveraged to improve ripple detection in humans. We implemented following detection pipeline: 1) IED snippet removal prior to ripple detection 2) identification of ripple-positive channels based on subfield location (CA1); 3) presence of narrowband peaks > 60 Hz, and 4) unsupervised clustering of ripple/IED waveforms (LFP) using uniform manifold approximation analysis (UMAP). This method was implemented in our open-source toolbox, ripmap (https://github.com/acnavasolive/ripmap). Event classification by ripmap was independently validated by three raters.
The pipeline identified ~40% of channels as ripple-negative. On the remaining channels, ripmap eliminated an additional ~15% of false positives (Fig.2), significantly reducing the need for manual curation.
Conclusions:
Human ripple detection must account for anatomical and pathological confounds. By leveraging rodent SPW-R features, we developed a pipeline that improves the specificity of ripple detection in humans. This approach provides a foundation for future studies on memory-related ripple activity in the human brain.
Funding:
DFG MA 10301/1- 1 (AM), NYU FACES (AM, AL)
NOMIS fellowship (AN-O)
R01 NS127954 (AL), R01 NS1007806 (AL), 1K23NS104252 (AL)
NYU Department of Neurology (AL)
NIH MH122391 (GB)
U19 NS107616 (GB)