EEG and seizure detection
Neonatal EEG seizure classification with compact networks, validated patient by patient across the 79 newborns of the Helsinki corpus, aimed at running on embedded hardware at the bedside.
Research conducted in partnership with the Brain Institute at UFRN and partner institutions: neuroscience, biomedical engineering, neonatology.
Neonatal EEG seizure classification with compact networks, validated patient by patient across the 79 newborns of the Helsinki corpus, aimed at running on embedded hardware at the bedside.
Respiratory rate estimated from video through thoracoabdominal segmentation and optical-flow analysis, with no electrode or belt on the infant.
Pose estimation on NICU video to describe activity and posture through the stay, the basis for reading spontaneous behaviour.
Whole-body and body-part segmentation with promptable foundation models, delimiting the region of interest the other signals come from.
The fronts above share the same corpus and build on one another: segmentation delimits the region of interest the respiration and movement signals come from. The work is conducted with researchers at the Brain Institute, UFRN, focused on the neonatal ICU, and feeds directly into the platform’s models and clinical validation.
Peer-reviewed articles produced with the NeuroMate team and partner institutions.
Ribeiro SNS, Fernandes AER, Vargas MERR, Velame RC, Tavares Filho MA, Leão RN, Maia H, Pereira SA, Rodrigues-Machado MG
2026 · 185(8):617 · doi:10.1007/s00431-026-07267-w
Leão RN, Fernandes AER, Tavares Filho MA, Maia H
2026 · in press
Maia H, Fernandes AER, Tavares Filho MA, Ribeiro SNS, Pereira SA, Rodrigues-Machado MG, Leão RN
2026 · in press
Fernandes AER, Silva TG, Ribeiro AJV, Freitas JWR, Barreto ACNG, Cobucci RN, Tavares Filho MA, Soares AWA, Sequerra EB, Maia H, Leão RN
2026 · in press