Median Nerve Stimulation to Detect Awake and Anesthetized States with Riemannian Classifiers - INRIA Bordeaux - Sud-Ouest Accéder directement au contenu
Communication Dans Un Congrès Année : 2024

Median Nerve Stimulation to Detect Awake and Anesthetized States with Riemannian Classifiers

Résumé

Among all the operations with general anesthesia worldwide, some patients have had the terrible experience of Accidental Awareness during General Anesthesia (AAGA), an unexpected awakening during surgery [1], potentially leading to pain, recall of surgical events and post-traumatic stress disorders [2]. The inability to predict AAGA using only clinical signs, leads to the use of brain activity monitors [3]. Given AAGA patients’ first reflex to move, impeded by neuromuscular-blocking agents, we propose using a new Brain Computer Interface (BCI) with Median Nerve Stimulation (MNS) to detect their movement intentions (MI). Combining MI and MNS significantly impacts MNS-generated patterns, leading to better MI detection accuracy. MNS also provides a trigger, resulting in better classification (+18%) compared to asynchronous BCI [4]. To continue in this direction, we analyzed the MNS patterns of 13 patients receiving propofol, an intravenous anesthetic, during surgery at CHU Brugmann, Belgium [5]. Two conditions were examined: EEG activities of preoperative MNS while the patient is awake (MNS-awake), and intraoperative MNS during various stages of anesthesia (MNS-anesthesia). Then, a Riemannian Minimum Distance to the Mean classifier achieved 97% test balanced accuracy in distinguishing awake and anesthetized states. We also observed the classifier’s response evolution with anesthesia depth, in terms of distance between the MNS-awake class centroid and covariance matrices of other trials throughout the surgery (Fig. 1). This distance appears to track the patients’ awareness level: it is greater during the maintenance phase, at deeper anesthetic concentrations, than during the induction or emergence phases. This holds potential for creating a one class classifier using this distance, only relying on MNS-awake data, as MNS-anesthesia data are usually unavailable for calibration during regular surgeries.
Fichier principal
Vignette du fichier
AbstractCORTICO24_MarissensCueva.pdf (244.76 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04570975 , version 1 (03-06-2024)

Identifiants

  • HAL Id : hal-04570975 , version 1

Citer

Valérie Marissens Cueva, Sébastien Rimbert, Ana Maria Cebolla Alvarez, Mathieu Petieau, Viktoriya Vitkova, et al.. Median Nerve Stimulation to Detect Awake and Anesthetized States with Riemannian Classifiers. Journées CORTICO 2024 - COllectif pour la Recherche Transdisciplinaire sur les Interfaces Cerveau-Ordinateur, https://cortico.fr/, May 2024, Nancy, France. ⟨hal-04570975⟩
30 Consultations
23 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More