MedElevator Войти в исследования
← Поиск публикаций

DOI 10.1109/tnsre.2021.3114367

Медицинская библиотека · IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society · 2021

Finding Kinematics-Driven Latent Neural States From Neuronal Population Activity for Motor Decoding.

Kim Min-Ki, Sohn Jeong-Woo, Kim Sung-Phil

СтатьяПоддержка исследования: вне правительства СШАПеревод пока не подготовлен

Аннотация

While intracortical brain-machine interfaces (BMIs) demonstrate feasibility to restore mobility to people with paralysis, it is still challenging to maintain high-performance decoding in clinical BMIs. One of the main obstacles for high-performance BMI is the noise-prone nature of traditional decoding methods that connect neural response explicitly with physical quantity, such as velocity. In contrast, the recent development of latent neural state model enables a robust readout of large-scale neuronal population activity contents. However, these latent neural states do not necessarily contain kinematic information useful for decoding. Therefore, this study proposes a new approach to finding kinematics-dependent latent factors by extracting latent factors' kinematics-dependent components using linear regression. We estimated these components from the population activity through nonlinear mapping. The proposed kinematics-dependent latent factors generate neural trajectories that discriminate latent neural states before and after the motion onset. We compared the decoding performance of the proposed analysis model with the results from other popular models. They are factor analysis (FA), Gaussian process factor analysis (GPFA), latent factor analysis via dynamical systems (LFADS), preferential subspace identification (PSID), and neuronal population firing rates. The proposed analysis model results in higher decoding accuracy than do the others ( % improvement on average). Our approach may pave a new way to extract latent neural states specific to kinematic information from motor cortices, potentially improving decoding performance for online intracortical BMIs.

Для цитирования: Kim Min-Ki, Sohn Jeong-Woo, Kim Sung-Phil. Finding Kinematics-Driven Latent Neural States From Neuronal Population Activity for Motor Decoding.. IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society. 2021. DOI: 10.1109/tnsre.2021.3114367.

Предметные рубрики MeSH

Подготовка и перевод публикации
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…

История · было → стало

Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.

Тип публикации указан по данным источника. Уровень доказательности автоматически не присваивается.