Development of iNeurofeedback instantaneous neurofeedback technology and creation of a hardware-software system for increasing and restoring cognitive function and therapy of neurological disorders

Neurofeedback (NF) is a neuromodulation technology that allows a person to learn to consciously or unconsciously control their own brain activity (Figure 1). This approach analyzes real-time brain activity signals and calculates a specific parameter characterizing brain activity, such as instantaneous alpha rhythm power, information about which is presented to the user as feedback.


Fig. 1. The principle of the neurofeedback method. (From the article: Smetanin N, Belinskaya A, Lebedev M, Ossadtchi A. Digital filters for low-latency quantification of brain rhythms in real time. J Neural Eng. 2020 Aug 4;17(4):046022. doi: 10.1088/1741-2552/ab890f. PMID: 32289760).

NOS therapy is currently used to treat conditions such as hyperactivity and attention deficit, stroke sequelae, anxiety, depression, epilepsy, etc., as well as to improve cognitive abilities and in training athletes on national level teams. Despite the proven effectiveness of the method for the treatment of a number of diseases, the use of currently available NOS systems is associated with a number of limitations: 
normalization of brain activity and reduction of clinical manifestations of neurological disorders as a result of NOS therapy is not always long-term; 
the problem of individual insensitivity to neurobiological therapy is relevant (there is a large cohort (up to 40%) of patients for whom the use of existing NOS systems proves ineffective); 
the use of a conscious explicit behavioral strategy (e.g., relaxation and attempting to disconnect from sensory channels during alpha rhythm training) increases the effectiveness of training. In the absence of such a strategy, normalization of EEG patterns with the help of modern NOS therapies proves to be significantly difficult.

Our researches have shown that one of the reasons of existing NOS limitations is inadequately long delay between the moment of change of brain activity and the moment of presentation of the feedback signal reflecting this change. It turns out that the changing parameter of brain activity during training of the occipital alpha rhythm is the number of short bursts of brain activity (200-300 milliseconds in duration) per unit time, while the duration and amplitude of these bursts remain unchanged (Ossadtchi et al., 2017). Thus, the essence of training is to teach a person to enter such target states and increase the number of bursts. Consequently, timely positive reinforcement of these transitions should play a key role in effective training. At the same time, in most modern systems, the feedback signal of current brain activity is presented with a delay of more than 500 ms (Fig. 2). Under such conditions, it is difficult to correlate the feedback with the event in response to which it was presented. Reducing the latency of the feedback signal will significantly increase the probability of such feedback and initiate the mechanisms of cortical plasticity necessary to achieve the long-term effect of training (Belinskaya et al., 2020).


Fig. 2. Optimal for effective learning and typically used by modern systems, the time window of feedback presentation. 

We have tested a mathematical algorithm for envelope estimation with a delay of about 100 ms (more than two times less than in standard NOS systems) and received a patent for utility model RU207767U1 - low-latency neurofeedback device. This study showed an improvement in the results of neurobiocontrol when using our developed method of feedback with minimal delay (Smetanin et al., 2020, Belinskaya et al., 2020) (Fig. 3).

Figure 3. Reducing the delay of feedback leads to a stronger post-training effect of NOS (from the article: Belinskaya, A., Smetanin, N., Lebedev, M. A., & Ossadtchi, A. Short-delay neurofeedback facilitates training of the parietal alpha rhythm. Journal of Neural Engineering, 2020, 12/16/20,https://doi.org/10.1088/1741-2552/abc8d7)

The algorithm for fast evaluation of the envelope of a narrowband rhythm is a complex-valued filter with a finite impulse response (CFIR), for which the parameters are selected so as to achieve the minimum filtering delay while maintaining the most plausible appearance of the real envelope of the narrowband signal. The peculiarity of the filter is that it simultaneously performs narrowband filtering and conversion into a complex-valued analytical signal, allowing then to isolate the envelope and the rhythm phase, from which a stimulus can then be formed for neurofeedback or real-time TMC stimulation.
Despite the clear advantages of the CFIR filter, it remains a window transformation with a fixed, and rather high, delay. Our further studies have shown that using the Kalman filtering methodology in combination with models of rhythmic activity as a frequency-modulated signal can reduce the algorithmic delay of burst detection to almost zero with little noise. This is achieved by using a realistic model of rhythm behavior in the Kalman filter, which allows very fast adaptation to instantaneous changes in the amplitude or phase of the rhythm (Fig. 4).

Fig. 4. Algorithmic feedback delay using CFIR filter and Kalman filter.

The work is supported by a Start grant from the Foundation for the Promotion of Innovation and is conducted jointly with "Brainstart.

Published articles on the project:

1. Ossadtchi, A., Shamaeva, T., Okorokova, E., Moiseeva, V., & Lebedev, M. A. (2017). Neurofeedback learning modifies the incidence rate of alpha spindles, but not their duration and amplitude. Scientific reports , 7 (1), 1-12,https://link.springer.com/content/pdf/10.1038/s41598-017-04012-0.pdf

2. Smetanin, N., Belinskaya, A., Lebedev, M. A., & Ossadtchi, A. Digital filters for low-latency quantification of brain rhythms in real-time. Journal of Neural Engineering, 2020,  29.07.20, 2, https://iopscience.iop.org/article/10.1088/1741-2552/ab890f

3. Belinskaya, A., Smetanin, N., Lebedev, M. A., & Ossadtchi, A. Short-delay neurofeedback facilitates training of the parietal alpha rhythm. Journal of Neural Engineering, 2020, 16.12.20,https://doi.org/10.1088/1741-2552/abc8d7