An in vitro simulation method of cognitive function based on microelectrode array and brain organoids and its application

Abstract

The present invention provides an in vitro simulation method for cognitive function based on a microelectrode array (MEA) and brain organoids, and its use, which belongs to the interdisciplinary fields of in vitro model construction of brain organoids, neural engineering, and brain-like intelligence. In an in vitro environment, the present invention uses MEA electrical stimulation to convert external signal information (such as sound) into electrical stimulation input that can be perceived by brain organoids, allowing brain organoids to gradually learn and recognize external signals through repeated electrical stimulation; after a period of washout stimulation, the reduction in recognition accuracy is observed, thereby simulating the important memory and forgetting processes in cognitive function. This method can quantitatively evaluate the electrophysiological responses of brain organoids under external electrical stimulation and changes in their biological mechanisms, providing new possibilities for using brain organoids to explore advanced cognitive functions.

Description

Technical Field

[0001] The present invention belongs to the field of in vitro model construction of brain organoids, the intersection of neural engineering and brain-like intelligence, and specifically relates to an in vitro simulation method of cognitive functions based on microelectrode arrays and brain organoids and its use. Background Art

[0002] In existing neuroscience research, in vitro environments make it difficult to simulate the brain's dynamic processing of cognitive information. Traditional microelectrode arrays (MEAs) are primarily used to record the electrical activity of neurons under physiological or pathological conditions. Previous studies have focused on recording the spontaneous discharge activity of cultured brain slices, neuronal monolayers, or brain organoids to evaluate functional changes in neural networks under developmental, drug, or pathological conditions. A Chinese patent (CN118620840A) discloses a method for culturing and detecting brain organoids. By physically stimulating brain organoids in the early stages of culture, this method accelerates their maturation and enables the maturation and activation of their spontaneous neural networks. However, existing technologies have yet to achieve a closed-loop system that actively encodes electrical stimulation to induce neural tissue to perform cognitive functions (such as emotion recognition, learning and memory).

[0003] Brain organoids, three-dimensional brain-like tissue models derived from pluripotent stem cells, possess certain neural network structural and functional characteristics. However, effective technical means remain for studying higher-level brain functions, such as responses to external inputs and assessment of cognitive function. The lack of methods for decoding active input and response information in brain organoids makes it impossible to assess their ability to encode and decode complex cognitive signals.

[0004] Therefore, by converting external information of multiple modalities (such as sound, pictures, and text) into electrical stimulation parameters, the cognitive function processing of brain organoids can be precisely controlled, which is of great value to the application of brain organoid models in cognitive science. Summary of the Invention

[0005] In order to solve the above-mentioned problems existing in the prior art, the purpose of the present invention is to provide an in vitro simulation method of cognitive function based on microelectrode arrays and brain organs and its use.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides an in vitro method for simulating cognitive function based on a microelectrode array and brain organoids, the method comprising the following steps:

[0008] (1) Preparation of mature brain organoids or other neural tissues with functional neural networks;

[0009] (2) Coupling mature brain organoids or other neural tissues with microelectrode arrays to achieve bidirectional signal interaction;

[0010] (3) Construct a multi-dimensional electrical stimulation input set for cognitive function-related processes based on the neuroelectrophysiological characteristics of the human brain;

[0011] (4) using a microelectrode array to input the electrical stimulation obtained in step (3) into the mature brain organoid or other neural tissue in step (2), record the electrophysiological response signals of the mature brain organoid or other neural tissue, analyze the electrophysiological response signals, and evaluate the performance of the mature brain organoid or other neural tissue in cognitive-related tasks;

[0012] (5) repeated electrical stimulation training;

[0013] (6) Stop electrical stimulation training and evaluate the performance of mature brain organoids or other neural tissues on cognitive-related tasks.

[0014] Furthermore, in step (1), the method for preparing mature brain organoids comprises the following steps: culturing embryonic stem cells or induced pluripotent stem cells and differentiating them into mature brain organoids; the culture is a three-dimensional suspension culture, and the culture time is 50-500 days.

[0015] Furthermore, the culture time is 120 days.

[0016] In step (2), the coupling method is to transfer the mature brain organoids or other neural tissues to a microelectrode array culture plate for continued culture; the microelectrode array culture plate is a microelectrode array culture plate whose culture plate surface is coated with Matrigel solution or Laminin solution; and the continued culture time is 7-450 days.

[0017] Furthermore, the continued culture time is 14 days.

[0018] Furthermore, step (3) includes the following steps: characterizing and decomposing the cognition-related multimodal signals, extracting key bio-perception features, and then encoding the key bio-perception features into electrical stimulation parameters that conform to the laws of neural electrophysiological conduction, and encoding them into microelectrode array electrical stimulation patterns to obtain an electrical stimulation input set for cognitive function-related processes; the cognition-related multimodal signals are cognition-related sounds, texts, or pictures.

[0019] Furthermore, the cognitive-related multimodal signal in step (3) is an emotion-related sound, and step (3) includes the following steps: taking an emotion-related sound sample, removing environmental noise through bandpass filtering, extracting Mel-frequency cepstral coefficients using the Python-based Librosa audio processing library, each Mel-frequency cepstral coefficient corresponds to the energy distribution of a specific frequency interval, and extracting 8-64 dimensional Mel-frequency cepstral coefficient feature vectors, generating a binary coding sequence through threshold judgment after normalization processing, and obtaining an electrical stimulation input set.

[0020] Furthermore, the emotion-related sounds in step (3) are two types of emotion audio samples of "happy" and "sad" selected from a standardized speech database.

[0021] Furthermore, the electrical stimulation in step (4) has a stimulation intensity of ≥400 mV and a stimulation frequency of 60-80 Hz;

[0022] In step (4) and step (6), the performance of the cognitive-related task is the accuracy of emotion recognition;

[0023] In step (4), the electrophysiological response signal is a discharge grid diagram recorded after electrical stimulation, and the discharge grid diagram is a heat map of the number of discharges of each electrode within a specific time window; the analysis is to perform principal component dimensionality reduction on the discharge events within 100ms after electrical stimulation, and further perform unsupervised clustering analysis or machine learning classification analysis; the accuracy of emotion recognition is obtained by classifying and calculating the response characteristics of different emotion types after analyzing the electrophysiological response signal through an unsupervised clustering algorithm or a machine learning and deep learning classification algorithm; the unsupervised clustering algorithm is kmeans, and the machine learning and deep learning classification algorithms are XGBoost and CNN.

[0024] Furthermore, the repeated electrical stimulation training in step (5) is repeated electrical stimulation training for 1 to 6 consecutive days, 1 to 5 times a day.

[0025] Furthermore, the repeated electrical stimulation training in step (5) is repeated electrical stimulation training once a day for three consecutive days.

[0026] Furthermore, in step (6), the time for stopping the electrical stimulation training is 3-10 days.

[0027] Furthermore, the time for stopping the electrical stimulation training is 7 days.

[0028] The present invention also provides a method for constructing an in vitro model for performing cognitive functions, the method comprising the following steps:

[0029] (1) Preparation of mature brain organoids or other neural tissues with functional neural networks;

[0030] (2) Coupling mature brain organoids or other neural tissues with microelectrode arrays to achieve bidirectional signal interaction;

[0031] (3) Construct a multi-dimensional electrical stimulation input set for cognitive function-related processes based on the neuroelectrophysiological characteristics of the human brain;

[0032] (4) using a microelectrode array to input the electrical stimulation obtained in step (3) into the mature brain organoid or other neural tissue, recording the electrophysiological response signals of the mature brain organoid or other neural tissue, analyzing the electrophysiological response signals, and evaluating the performance of the mature brain organoid or other neural tissue in cognitive-related tasks;

[0033] (5) Repeated electrical stimulation training to obtain an in vitro model for performing cognitive functions;

[0034] Wherein, steps (1)-(5) are as described above.

[0035] Furthermore, the in vitro model for performing cognitive functions is an in vitro model for emotion recognition.

[0036] The present invention also provides the use of the above-mentioned in vitro simulation method of cognitive function based on microelectrode arrays and brain organoids, and the method of constructing an in vitro model for performing cognitive function in neural injury repair, drug screening, research on the mechanism of mental illness, and development of intelligent neural interfaces.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] This study, using active electrical stimulation of MEAs, replicates the processes of emotion recognition learning, memory, and forgetting in brain organoids, successfully simulating the cognitive function execution of active brain tissue in vitro. This study can quantitatively assess the electrophysiological responses and biological mechanism changes of brain organoids under external electrical stimulation, providing a new tool for in-depth research on the advanced cognitive functions and brain-like properties of brain organoids.

[0039] This invention creates a novel, feasible, and innovative dynamic simulation system for cognitive function. It provides a new paradigm for in vitro simulation of complex cognitive processes (exemplified by emotion recognition, memory, and forgetting) in neuroscience, brain-inspired intelligence, and neural engineering. It is expected to provide theoretical foundations and technical support for subsequent research on emotion-related neural damage repair, drug screening, research on the mechanisms of psychiatric disorders, and the development of intelligent neural interfaces. Furthermore, a key advantage of this invention lies in its scalability and multidimensionality. By electrically encoding multimodal external information (such as sound, images, and text) and adjusting electrical stimulation parameters, the cognitive processing of brain organoids can be precisely controlled, and the effects of different modal information types and content on cognitive function can be explored. By continuously optimizing the frequency, amplitude, and pulse width of electrical stimulation, more precise simulations of different cognitive processes (such as emotion recognition, memory formation, and forgetting) can be achieved. With further research, this approach can be extended to other higher-level brain functions, such as decision-making, language comprehension, and emotion regulation, greatly broadening the application of brain organoid models in cognitive science.

[0040] Obviously, based on the above contents of the present invention, according to common technical knowledge and customary means in this field, without departing from the above basic technical ideas of the present invention, other various forms of modifications, replacements or changes can be made.

[0041] The following is a further detailed description of the present invention through specific embodiments in the form of examples. However, this should not be construed as limiting the scope of the present invention to the following examples. All technologies implemented based on the above-mentioned content of the present invention fall within the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The black double-strike line represents a broken axis and is used to compress the coordinate range.

[0043] Figure 2 The results of the in vitro electrical stimulation experiment to simulate cognitive function using the method of the present invention are shown in Figure 2. The solid red line box indicates the first discharge peak; the dashed red line box indicates the second discharge peak.

[0044] Figure 3 Results of an experiment to determine the electrical stimulation frequency for simulating cognitive function in vitro using the method of the present invention. Red solid line area: Output signals at frequencies of 60-80 Hz have higher discrimination.

[0045] Figure 4 A diagram showing the accuracy of brain organoids performing an emotion recognition task. The solid red boxes represent the accuracy (mean ± SD) on days 0, 3, 4, and 10, respectively.

[0046] Figure 5 The PCA clustering results of brain organoids before and after training are shown. The emotion recognition accuracy on day 0 was 0.6571, and on day 3 it was 0.7036. DETAILED DESCRIPTION

[0047] The raw materials and equipment used in the present invention are all known products and are obtained by purchasing commercially available products.

[0048] The overall process of the method and device of the present invention is as follows Figure 1 shown.

[0049] Example 1: Simulating the cognitive function execution process of emotion recognition, memory formation and forgetting in vitro by MEA electrical stimulation training and data analysis of brain organoids

[0050] (1) Generation of brain organoids

[0051] Human embryonic stem cells (ESCs, H9) were used ( E8, Gibco ) culture medium to maintain dry culture, when the cell state is appropriate, follow Lancaster et al., performed neural differentiation and three-dimensional suspension culture using a standardized brain organoid induction medium ( STEMdiff™ Cerebral Organoid Kit, STEMCELL Technologies ) were cultured long-term according to the manufacturer's instructions until maturity. After 120 days of culture, initially mature brain organoids with diverse neural cell types were obtained. These brain organoids exhibited a certain degree of electrophysiological activity in both morphology and function, and were able to spontaneously generate discharge patterns similar to those of early neural networks.

[0052] (2) Integration of brain organoids and MEAs

[0053] The above-mentioned cultured and initially mature brain organoids were transferred to the MEA system ( Maestro Pro, Axion BioSystems ) on a matching chip. MEA's high-throughput microelectrodes can simultaneously collect electrical signals and deliver electrical stimulation to brain organoids. To ensure full contact between brain organoids and electrodes and obtain a stable signal coupling relationship, the surface of the MEA culture plate is coated with 5% Matrigel ( Matrigel, Corning ) solution. Organoids were then cultured in MEA plates at 37°C and 5% CO2 to allow them to gradually adapt and stabilize in their new environment. After the organoids had adapted to the MEA plate for 14 days and their spontaneous discharges, as evidenced by the establishment of tight connections between the microelectrode chips, they were then subjected to electrical stimulation using emotional sound signals to simulate cognitive function.

[0054] (3) Screening of electrical stimulation parameters

[0055] ① Screen out the appropriate electrical stimulation intensity:

[0056] To achieve accurate and efficient simulation of the cognitive functions of brain organoids, the present invention conducted a series of experiments to optimize electrical stimulation parameters based on neurophysiological characteristics. First, the electrical stimulation intensity was selected for brain organoids at different stages (such as Day 40 of the juvenile stage and Day 120 of the mature stage), ranging from 100mV to 800mV. The experimental results showed that ( Figure 2 ): In immature brain organoids (D40), lower stimulation intensities (100 mV to 400 mV) can only induce short-term peak discharges (about 10 ms), indicating that the neural network is not yet perfect and cannot form sustained neural synchronous activity. When the stimulation intensity reaches 400 mV and above, mature brain organoids (D120) are able to form sustained discharges after stimulation and trigger a second discharge peak, indicating that the neural network already has good synaptic connection efficiency and broader neural synchronization. This phenomenon shows that the stimulation intensity has an important influence on the activation and information processing ability of the neural network. Immature brain organoids lacking functional neural networks show a transient single-peak discharge pattern in response to low-intensity stimulation, while mature brain organoids with functional neural networks can induce sustained multi-peak discharges under high-intensity stimulation.

[0057] ②Screen out the appropriate electrical stimulation frequency:

[0058] Stimulation frequency is an important factor affecting the information processing and cognitive function of brain organoids. Based on the use of 400 mV stimulation intensity, the present invention tested the effects of different frequencies from 20 Hz to 100 Hz on mature brain organoids (D120). The experimental results showed that ( Figure 3 ): The discharge frequency of brain organoids reached a peak within the frequency range of 60-80 Hz. This frequency range coincides with the gamma wave rhythm (30-80 Hz) when the brain performs higher-order cognitive functions, indicating that appropriate frequencies can effectively enhance neuronal synchronization and synaptic plasticity. Further dimensionality reduction using principal component analysis (PCA) revealed that the output signals at a frequency of 60-80 Hz had higher discrimination, indicating that this frequency can better support complex cognitive tasks such as emotion recognition, memory formation, and forgetting.

[0059] The above screening experiments demonstrate that appropriate stimulation intensity and frequency play a crucial role in simulating the electrophysiological activity and cognitive functions of brain organoids. By optimizing these electrical stimulation parameters, the present invention not only improves the learning and memory capabilities of brain organoids but also better simulates the dynamic changes in cognitive functions such as emotion recognition, memory formation, and forgetting. These screening parameters provide an important experimental basis for further research on the performance of brain organoids in cognitive tasks and can provide strong technical support for subsequent neuroscience research and brain-inspired intelligence applications.

[0060] (4) Feature coding of emotional sound signals

[0061] Audio samples of two emotional categories, "happy" and "sad," were selected from a standardized speech database. 100 emotional sound samples were randomly selected from the database for feature extraction and electrical encoding conversion. After bandpass filtering to remove ambient noise, Mel-Frequency Cepstral Coefficients (MFCCs) were extracted using the Python-based Librosa audio processing library. Each MFCC coefficient corresponds to the energy distribution of a specific frequency range. MFCC feature vectors of 8-64 dimensions (selected based on the MEA chip used) were extracted and normalized, then thresholded (threshold = 15) to generate a binary encoding sequence. Segments with feature values ​​above a preset threshold were converted into biphasic square wave electrical stimulation, while segments below the threshold corresponded to silence. Fifty encoded stimulus signals were generated for each emotional speech segment to construct a multidimensional emotional electrical stimulation input set.

[0062] (5) In vitro simulation of the cognitive function execution process of emotion recognition and memory

[0063] Training plan: Day 0 (pre-training) starts with training for three consecutive days (Days 1, 2, and 3), followed by a break of one week (7 days). Emotion recognition accuracy is then retested 7 days after training (Day 10).

[0064] In this cognitive simulation task, testing was performed before training (Day 0), after three training sessions (Day 3), one day after training (Day 4), and seven days after training (Day 10). Before the organoids were subjected to the cognitive task, a 5-minute baseline of spontaneous discharge activity was collected. During three consecutive days of training, the organoids were administered the aforementioned 100 randomized emotional electrical stimulation sequences (stimulation parameters: 400 mV intensity, 60 Hz frequency) at the same time each day (3:00 PM, at least 6 hours after the organoid medium was replaced). During the stimulation, the MEA recorded the organoids' electrophysiological responses in real time, capturing discharge frequency, inter-channel synchrony, and network dynamics. Discharge events were recorded for each electrode within a 100-ms time window after each stimulation (discharge was 1 for each electrode, and no discharge was 0 for each electrode). Discharge events within this 100-ms window were exported as a two-dimensional matrix of 0s and 1s for subsequent cognitive task accuracy analysis and visualized as a fence plot using Matlab software. Python-based code was used to analyze the discharge event matrix after electrical stimulation. Principal component analysis was first used to reduce the high-dimensional data, and the first two principal components were extracted to construct a two-dimensional distribution map. Furthermore, an unsupervised clustering algorithm (kmeans) (alternatively, machine learning and deep learning classification algorithms (XGBoost, CNN)) was used to distinguish the response distributions of the two emotional categories and evaluate the classification accuracy. This was used to calculate the accuracy of the brain organoids in the emotion recognition cognitive task.

[0065] The results are as follows Figure 4 As shown in the figure, before the training started (day 0), the response differentiation of brain organoids to electrical stimulation of different emotional types was about 50%, and their emotion recognition accuracy was low. However, with the completion of three consecutive electrical stimulation training sessions (day 3), the response characteristics of brain organoids gradually became more characterized, and the discharge patterns to different emotional stimuli showed more obvious differences, thus showing a higher emotion recognition accuracy in cluster analysis ( Figure 5 ). This demonstrates that brain organoids have plasticity and a tendency to learn and remember in response to repeated emotional electrical stimulation inputs, and that their internal networks adapt and integrate specific features under conditions of continuous stimulation.

[0066] (6) In vitro simulation of emotion recognition and forgetting processes

[0067] After the continuous training is completed, the electrical stimulation input to the brain organoids is temporarily stopped, and the brain organoids are allowed to rest in the MEA culture plate for one week. This interval is equivalent to providing the system with a "no reinforcement input" memory forgetting window. After one week, on the 10th day, the same emotional electrical stimulation sequence as in the previous training was input to the brain organoids again, and the electrophysiological response signals were analyzed. The results are as follows Figure 4As shown, the recognition accuracy at this time is significantly lower than the level before the stimulation is stopped, reflecting that the brain organoids' memory of emotional characteristics gradually decays, thus simulating the natural forgetting process experienced by the real brain under no reinforcement conditions.

[0068] In summary, the present invention provides a method for simulating cognitive function execution processes (exemplified by emotion recognition, memory, and forgetting) in brain organoid models using microelectrode array (MEA) technology. This method lies at the intersection of in vitro brain organoid model construction, neuroengineering, and brain-inspired intelligence. In an in vitro environment, MEA electrical stimulation is used to convert external signals (such as sound) into perceptible electrical stimulation inputs for the brain organoid, allowing the organoid to gradually learn and recognize external signals through repeated stimulation. A period of stimulation cessation is then followed to observe the decline in recognition accuracy, thereby simulating the important cognitive processes of memory and forgetting. This method can quantitatively assess the electrophysiological responses of brain organoids under external electrical stimulation and the changes in their biological mechanisms, providing a new tool for studying the higher-level cognitive functions and brain-like properties of brain organoids. It provides a new paradigm for simulating complex cognitive processes (emotion recognition, memory, and forgetting) in vitro for research in neuroscience, brain-inspired intelligence, and neuroengineering, and is expected to provide theoretical basis and technical support for subsequent research on emotion-related neural damage repair, drug screening, psychiatric disease mechanisms, and the development of intelligent neural interfaces.

Claims

1. A method for simulating cognitive function in vitro based on microelectrode arrays and brain organoids, characterized in that: The method comprises the following steps: (1) Preparation of mature brain organoids or other neural tissues with functional neural networks; (2) Coupling mature brain organoids or other neural tissues with microelectrode arrays to achieve bidirectional signal interaction; (3) Construct a multi-dimensional electrical stimulation input set for cognitive function-related processes based on the neuroelectrophysiological characteristics of the human brain; (4) using a microelectrode array to input the electrical stimulation obtained in step (3) into the mature brain organoid or other neural tissue in step (2), recording the electrophysiological response signals of the mature brain organoid or other neural tissue, analyzing the electrophysiological response signals, and evaluating the accuracy of emotion recognition of the mature brain organoid or other neural tissue; (5) repeated electrical stimulation training; (6) Stop electrical stimulation training and evaluate the accuracy of emotion recognition in mature brain organoids or other neural tissues; Wherein, the coupling method in step (2) is to transfer the mature brain organoids or other neural tissues to a microelectrode array culture plate for further culture; Step (3) includes the following steps: taking emotion-related sound samples, removing environmental noise through bandpass filtering, extracting Mel-frequency cepstral coefficients using the Python-based Librosa audio processing library, each Mel-frequency cepstral coefficient corresponds to the energy distribution of a specific frequency interval, and extracting 8-64 dimensional Mel-frequency cepstral coefficient feature vectors, generating a binary coding sequence through threshold determination after normalization processing, obtaining an electrical stimulation input set, and obtaining an electrical stimulation input set for cognitive function-related processes; The electrical stimulation in step (4) has a stimulation intensity of ≥400 mV and a stimulation frequency of 60-80 Hz.

2. The method according to claim 1, characterized in that In step (1), the method for preparing mature brain organoids comprises the following steps: culturing embryonic stem cells or induced pluripotent stem cells and differentiating them into mature brain organoids; the culturing is a three-dimensional suspension culture, and the culturing time is 50-500 days; The microelectrode array culture plate in step (2) is a microelectrode array culture plate whose surface is coated with Matrigel solution or Laminin solution; and the continued culture time is 7-450 days.

3. The method according to claim 1, characterized in that The emotion-related sounds in step (3) are two types of emotion audio samples, "happy" and "sad", selected from a standardized speech database.

4. The method according to claim 1, wherein In step (4), the electrophysiological response signal is a discharge grid diagram recorded after electrical stimulation; the analysis is to perform principal component dimensionality reduction on the discharge events within 100ms after electrical stimulation, and further perform unsupervised cluster analysis or machine learning classification analysis; the accuracy of emotion recognition is obtained by classifying and calculating the response characteristics of different emotion types after analyzing the electrophysiological response signal through an unsupervised clustering algorithm or a machine learning and deep learning classification algorithm; the unsupervised cluster analysis is kmeans, and the machine learning and deep learning classification algorithms are XGBoost and CNN.

5. The method according to any one of claims 1 to 4, characterized in that The repeated electrical stimulation training in step (5) is repeated electrical stimulation training for 1 to 6 consecutive days, 1 to 5 times a day.

6. The method according to any one of claims 1 to 4, characterized in that In step (6), the time for stopping the electrical stimulation training is 3-10 days.

7. Use of the in vitro simulation method of cognitive function based on microelectrode arrays and brain organoids as described in any one of claims 1 to 6 in drug screening, research on the mechanisms of mental illness, and development of intelligent neural interfaces.

8. Use of the in vitro simulation method of cognitive function based on microelectrode arrays and brain organoids as described in any one of claims 1 to 6 for quantitatively evaluating the electrophysiological responses and biological mechanism changes of brain organoids under external electrical stimulation.

Images