{ "id": 468771, "title": "Papers & Model Architectures", "url": "/competitions/hms-harmful-brain-activity-classification/discussion/468771", "author_name": "", "post_date": "2024-01-17T22:15:40.038295500Z", "votes": 39, "comment_count": 2, "views": 0, "content": "

Note : Only from article abstract, recent papers and skipped if it mention similar model architecture

\n
\n

Discussion Motivated to explore different architectures in this domain
\n Haha, yeah. I think mega ensembles will do good in this competition. Also a multimodal model that can take all the data (spectrogram and eeg waveform and stacked over GBT output) and have the benefits of CNN, RNN, self attention will make very strong \"single\" models. by @cdeotte

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Automatic epileptic seizure detection based on EEG using a moth-flame optimization of one-dimensional convolutional neural networks-1D CNNEGG
\n
\n

This study proposes an automatic detection model for an EEG based on moth-flame optimization (MFO) optimized one-dimensional convolutional neural networks (1D-CNN). First, according to the characteristics and need for early epileptic seizure detection, a data augmentation method for dividing an EEG into small samples is proposed. Second, the hyperparameters are tuned based on MFO and trained for an EEG. Finally, the softmax classifier is used to output EEG classification from a small-sample and single channel.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
SMARTSeiz: Deep Learning with Attention Mechanism for Accurate Seizure Recognition in IoT Healthcare Devices-CNN + RNNEGG
\n
\n

Currently, doctors invest significant manual effort in inspecting Electroencephalograph (EEG) signals to identify seizure activity. However, EEG-based seizure detection algorithms face challenges in real-world scenarios due to non-stationary EEG data and variable seizure patterns among patients and recording sessions. Therefore, a sophisticated computer-based approach is necessary to analyze complex EEG records. In this work, the authors proposed a hybrid approach by combining traditional convolution neural (CN) and recurrent neural networks (RNN) along with an attention mechanism for the automatic recognition of epileptic seizures through EEG signal analysis. This attention mechanism focuses on significant subsets of EEG data for class recognition, resulting in improved model performance.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Epileptic Seizure Detection with an End-to-End Temporal Convolutional Network and Bidirectional Long Short-Term Memory Model-CNN + Bi-LSTMEEG0.5-45 Hz band-pass-
\n
\n

Automatic seizure detection plays a key role in assisting clinicians for rapid diagnosis and treatment of epilepsy. In view of the parallelism of temporal convolutional network (TCN) and the capability of bidirectional long short-term memory (BiLSTM) in mining the long-range dependency of multi-channel time-series, we propose an automatic seizure detection method with a novel end-to-end TCN-BiLSTM model in this work. First, raw EEG is filtered with a 0.5-45 Hz band-pass filter, and the filtered data are input into the proposed TCN-BiLSTM network for feature extraction and classification. Post-processing process including moving average filtering, thresholding and collar technique is then employed to further improve the detection performance. The method was evaluated on two EEG database.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Epileptic Seizure Detection Based on Path Signature and Bi-LSTM Network With Attention Mechanismpath signature algorithmBi-LSTM + ATTEEG-path signature algorithm ?
\n
\n

Automatic seizure detection using electroen-cephalogram (EEG) can significantly expedite the diagnosis of epilepsy, thereby facilitating prompt treatment and reducing the risk of future seizures and associated complications. While most existing EEG-based epilepsy detection studies employ deep learning models, they often ignore the chronological relationships between different EEG channels. To tackle this limitation, a novel automatic epilepsy detection method is proposed, which leverages path signature and Bidirectional Long Short-Term Memory (Bi-LSTM) neural network with an attention mechanism. The path signature algorithm is used to extract discriminative features for capturing the dynamic dependencies between different channels of EEG, while Bi-LSTM with attention further analyzes the inherent temporal dependencies hidden in EEG signal features.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Automatic Seizure Detection Based on Stockwell Transform and TransformerGo through paperTransformerEGG => Spectrogram-interesting to see post processing approach, how we can be use for KL metric?
\n
\n

pilepsy is a chronic neurological disease associated with abnormal neuronal activity in the brain. Seizure detection algorithms are essential in reducing the workload of medical staff reviewing electroencephalogram (EEG) records. In this work, we propose a novel automatic epileptic EEG detection method based on Stockwell transform and Transformer. First, the S-transform is applied to the original EEG segments, acquiring accurate time-frequency representations. Subsequently, the obtained time-frequency matrices are grouped into different EEG rhythm blocks and compressed as vectors in these EEG sub-bands. After that, these feature vectors are fed into the Transformer network for feature selection and classification.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
An improved GBSO-TAENN-based EEG signal classification model for epileptic seizure detectionFinite Linear Haar wavelet-based Filtering-EGG-difficult to understand this paper => non-ml approach
\n
\n

Detection and classification of epileptic seizures from the EEG signals have gained significant attention in recent decades. Among other signals, EEG signals are extensively used by medical experts for diagnosing purposes. So, most of the existing research works developed automated mechanisms for designing an EEG-based epileptic seizure detection system. Machine learning techniques are highly used for reduced time consumption, high accuracy, and optimal performance. Still, it limits by the issues of high complexity in algorithm design, increased error value, and reduced detection efficacy. Thus, the proposed work intends to develop an automated epileptic seizure detection system with an improved performance rate. Here, the Finite Linear Haar wavelet-based Filtering (FLHF) technique is used to filter the input signals and the relevant set of features are extracted from the normalized output with the help of Fractal Dimension (FD) analysis. Then, the Grasshopper Bio-Inspired Swarm Optimization (GBSO) technique is employed to select the optimal features by computing the best fitness value and the Temporal Activation Expansive Neural Network (TAENN) mechanism is used for classifying the EEG signals to determine whether normal or seizure affected.

\n
\n

Last but not least

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Mixture of Experts for EEG-Based Seizure Subtype Classification--MoE-never imagined MoE already used in this area => choice ensemble? or MoE?
\n
\n

This paper proposes two novel Mixture of Experts (MoE) models, Seizure-MoE and Mix-MoE, for EEG-based seizure subtype classification. Particularly, Mix-MoE adequately addresses the above two challenges: 1) it introduces a novel imbalanced sampler to address significant class imbalance; and 2) it incorporates a priori knowledge of manual EEG features into the deep neural network to improve the classification performance. Experiments on two public datasets demonstrated that the proposed Seizure-MoE and Mix-MoE outperformed multiple existing approaches in cross-subject EEG-based seizure subtype classification. Our proposed MoE models may also be easily extended to other EEG classification problems with severe class imbalance, e.g., sleep stage classification.

\n
\n

Preprocessing

\n\n

Domain Questions
\n-- what is blind source separation?
\n-- what is path signature algorithm?
\n-- not many recent papers used egg spectrogram?
\n-- no wavnet used?

\n

ML Questions
\n-- how we can use threshold for KL Metric?
\n-- MoE vs Ensemble ?

\n

Will start my notebook series for each paper, is good others also join in this series
\nShare new papers if you come across interesting and i miss the respective architecture

", "messages": [ { "id": "2606850", "postDate": "01/17/2024 22:15:40", "content": "

Note : Only from article abstract, recent papers and skipped if it mention similar model architecture

\n
\n

Discussion Motivated to explore different architectures in this domain
\n Haha, yeah. I think mega ensembles will do good in this competition. Also a multimodal model that can take all the data (spectrogram and eeg waveform and stacked over GBT output) and have the benefits of CNN, RNN, self attention will make very strong \"single\" models. by @cdeotte

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Automatic epileptic seizure detection based on EEG using a moth-flame optimization of one-dimensional convolutional neural networks-1D CNNEGG
\n
\n

This study proposes an automatic detection model for an EEG based on moth-flame optimization (MFO) optimized one-dimensional convolutional neural networks (1D-CNN). First, according to the characteristics and need for early epileptic seizure detection, a data augmentation method for dividing an EEG into small samples is proposed. Second, the hyperparameters are tuned based on MFO and trained for an EEG. Finally, the softmax classifier is used to output EEG classification from a small-sample and single channel.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
SMARTSeiz: Deep Learning with Attention Mechanism for Accurate Seizure Recognition in IoT Healthcare Devices-CNN + RNNEGG
\n
\n

Currently, doctors invest significant manual effort in inspecting Electroencephalograph (EEG) signals to identify seizure activity. However, EEG-based seizure detection algorithms face challenges in real-world scenarios due to non-stationary EEG data and variable seizure patterns among patients and recording sessions. Therefore, a sophisticated computer-based approach is necessary to analyze complex EEG records. In this work, the authors proposed a hybrid approach by combining traditional convolution neural (CN) and recurrent neural networks (RNN) along with an attention mechanism for the automatic recognition of epileptic seizures through EEG signal analysis. This attention mechanism focuses on significant subsets of EEG data for class recognition, resulting in improved model performance.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Epileptic Seizure Detection with an End-to-End Temporal Convolutional Network and Bidirectional Long Short-Term Memory Model-CNN + Bi-LSTMEEG0.5-45 Hz band-pass-
\n
\n

Automatic seizure detection plays a key role in assisting clinicians for rapid diagnosis and treatment of epilepsy. In view of the parallelism of temporal convolutional network (TCN) and the capability of bidirectional long short-term memory (BiLSTM) in mining the long-range dependency of multi-channel time-series, we propose an automatic seizure detection method with a novel end-to-end TCN-BiLSTM model in this work. First, raw EEG is filtered with a 0.5-45 Hz band-pass filter, and the filtered data are input into the proposed TCN-BiLSTM network for feature extraction and classification. Post-processing process including moving average filtering, thresholding and collar technique is then employed to further improve the detection performance. The method was evaluated on two EEG database.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Epileptic Seizure Detection Based on Path Signature and Bi-LSTM Network With Attention Mechanismpath signature algorithmBi-LSTM + ATTEEG-path signature algorithm ?
\n
\n

Automatic seizure detection using electroen-cephalogram (EEG) can significantly expedite the diagnosis of epilepsy, thereby facilitating prompt treatment and reducing the risk of future seizures and associated complications. While most existing EEG-based epilepsy detection studies employ deep learning models, they often ignore the chronological relationships between different EEG channels. To tackle this limitation, a novel automatic epilepsy detection method is proposed, which leverages path signature and Bidirectional Long Short-Term Memory (Bi-LSTM) neural network with an attention mechanism. The path signature algorithm is used to extract discriminative features for capturing the dynamic dependencies between different channels of EEG, while Bi-LSTM with attention further analyzes the inherent temporal dependencies hidden in EEG signal features.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Automatic Seizure Detection Based on Stockwell Transform and TransformerGo through paperTransformerEGG => Spectrogram-interesting to see post processing approach, how we can be use for KL metric?
\n
\n

pilepsy is a chronic neurological disease associated with abnormal neuronal activity in the brain. Seizure detection algorithms are essential in reducing the workload of medical staff reviewing electroencephalogram (EEG) records. In this work, we propose a novel automatic epileptic EEG detection method based on Stockwell transform and Transformer. First, the S-transform is applied to the original EEG segments, acquiring accurate time-frequency representations. Subsequently, the obtained time-frequency matrices are grouped into different EEG rhythm blocks and compressed as vectors in these EEG sub-bands. After that, these feature vectors are fed into the Transformer network for feature selection and classification.

\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
An improved GBSO-TAENN-based EEG signal classification model for epileptic seizure detectionFinite Linear Haar wavelet-based Filtering-EGG-difficult to understand this paper => non-ml approach
\n
\n

Detection and classification of epileptic seizures from the EEG signals have gained significant attention in recent decades. Among other signals, EEG signals are extensively used by medical experts for diagnosing purposes. So, most of the existing research works developed automated mechanisms for designing an EEG-based epileptic seizure detection system. Machine learning techniques are highly used for reduced time consumption, high accuracy, and optimal performance. Still, it limits by the issues of high complexity in algorithm design, increased error value, and reduced detection efficacy. Thus, the proposed work intends to develop an automated epileptic seizure detection system with an improved performance rate. Here, the Finite Linear Haar wavelet-based Filtering (FLHF) technique is used to filter the input signals and the relevant set of features are extracted from the normalized output with the help of Fractal Dimension (FD) analysis. Then, the Grasshopper Bio-Inspired Swarm Optimization (GBSO) technique is employed to select the optimal features by computing the best fitness value and the Temporal Activation Expansive Neural Network (TAENN) mechanism is used for classifying the EEG signals to determine whether normal or seizure affected.

\n
\n

Last but not least

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
PaperFeaturesModelSequenceFiltersNotes
Mixture of Experts for EEG-Based Seizure Subtype Classification--MoE-never imagined MoE already used in this area => choice ensemble? or MoE?
\n
\n

This paper proposes two novel Mixture of Experts (MoE) models, Seizure-MoE and Mix-MoE, for EEG-based seizure subtype classification. Particularly, Mix-MoE adequately addresses the above two challenges: 1) it introduces a novel imbalanced sampler to address significant class imbalance; and 2) it incorporates a priori knowledge of manual EEG features into the deep neural network to improve the classification performance. Experiments on two public datasets demonstrated that the proposed Seizure-MoE and Mix-MoE outperformed multiple existing approaches in cross-subject EEG-based seizure subtype classification. Our proposed MoE models may also be easily extended to other EEG classification problems with severe class imbalance, e.g., sleep stage classification.

\n
\n

Preprocessing

\n\n

Domain Questions
\n-- what is blind source separation?
\n-- what is path signature algorithm?
\n-- not many recent papers used egg spectrogram?
\n-- no wavnet used?

\n

ML Questions
\n-- how we can use threshold for KL Metric?
\n-- MoE vs Ensemble ?

\n

Will start my notebook series for each paper, is good others also join in this series
\nShare new papers if you come across interesting and i miss the respective architecture

", "rawMarkdown": "*Note* : **Only from article abstract, recent papers and skipped if it mention similar model architecture**\n> **[Discussion](https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468684#2606504)** Motivated to explore different architectures in this domain \n> Haha, yeah. I think mega ensembles will do good in this competition. Also a multimodal model that can take all the data (spectrogram and eeg waveform and stacked over GBT output) and have the benefits of CNN, RNN, self attention will make very strong \"single\" models. by @cdeotte\n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Automatic epileptic seizure detection based on EEG using a moth-flame optimization of one-dimensional convolutional neural networks](https://pubmed.ncbi.nlm.nih.gov/38161793/) | - | 1D CNN | EGG | | | \n> This study proposes an automatic detection model for an EEG based on moth-flame optimization (MFO) optimized one-dimensional convolutional neural networks (1D-CNN). First, according to the characteristics and need for early epileptic seizure detection, a data augmentation method for dividing an EEG into small samples is proposed. Second, the hyperparameters are tuned based on MFO and trained for an EEG. Finally, the softmax classifier is used to output EEG classification from a small-sample and single channel.\n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [SMARTSeiz: Deep Learning with Attention Mechanism for Accurate Seizure Recognition in IoT Healthcare Devices](https://pubmed.ncbi.nlm.nih.gov/38055360/) | - | CNN + RNN | EGG | | | \n> Currently, doctors invest significant manual effort in inspecting Electroencephalograph (EEG) signals to identify seizure activity. However, EEG-based seizure detection algorithms face challenges in real-world scenarios due to non-stationary EEG data and variable seizure patterns among patients and recording sessions. Therefore, a sophisticated computer-based approach is necessary to analyze complex EEG records. In this work, the authors proposed a hybrid approach by combining traditional convolution neural (CN) and recurrent neural networks (RNN) along with an attention mechanism for the automatic recognition of epileptic seizures through EEG signal analysis. This attention mechanism focuses on significant subsets of EEG data for class recognition, resulting in improved model performance. \n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Epileptic Seizure Detection with an End-to-End Temporal Convolutional Network and Bidirectional Long Short-Term Memory Model](https://pubmed.ncbi.nlm.nih.gov/38230571/) | - | CNN + Bi-LSTM | EEG | 0.5-45 Hz band-pass | - | \n> Automatic seizure detection plays a key role in assisting clinicians for rapid diagnosis and treatment of epilepsy. In view of the parallelism of temporal convolutional network (TCN) and the capability of bidirectional long short-term memory (BiLSTM) in mining the long-range dependency of multi-channel time-series, we propose an automatic seizure detection method with a novel end-to-end TCN-BiLSTM model in this work. First, raw EEG is filtered with a 0.5-45 Hz band-pass filter, and the filtered data are input into the proposed TCN-BiLSTM network for feature extraction and classification. Post-processing process including moving average filtering, thresholding and collar technique is then employed to further improve the detection performance. The method was evaluated on two EEG database. \n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Epileptic Seizure Detection Based on Path Signature and Bi-LSTM Network With Attention Mechanism](https://pubmed.ncbi.nlm.nih.gov/38224524/) | path signature algorithm | Bi-LSTM + ATT | EEG | - | path signature algorithm ? |\n> Automatic seizure detection using electroen-cephalogram (EEG) can significantly expedite the diagnosis of epilepsy, thereby facilitating prompt treatment and reducing the risk of future seizures and associated complications. While most existing EEG-based epilepsy detection studies employ deep learning models, they often ignore the chronological relationships between different EEG channels. To tackle this limitation, a novel automatic epilepsy detection method is proposed, which leverages path signature and Bidirectional Long Short-Term Memory (Bi-LSTM) neural network with an attention mechanism. The path signature algorithm is used to extract discriminative features for capturing the dynamic dependencies between different channels of EEG, while Bi-LSTM with attention further analyzes the inherent temporal dependencies hidden in EEG signal features. \n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Automatic Seizure Detection Based on Stockwell Transform and Transformer](https://pubmed.ncbi.nlm.nih.gov/38202939/) | Go through paper | Transformer | EGG => Spectrogram | - | interesting to see post processing approach, how we can be use for KL metric? |\n> pilepsy is a chronic neurological disease associated with abnormal neuronal activity in the brain. Seizure detection algorithms are essential in reducing the workload of medical staff reviewing electroencephalogram (EEG) records. In this work, we propose a novel automatic epileptic EEG detection method based on Stockwell transform and Transformer. First, the S-transform is applied to the original EEG segments, acquiring accurate time-frequency representations. Subsequently, the obtained time-frequency matrices are grouped into different EEG rhythm blocks and compressed as vectors in these EEG sub-bands. After that, these feature vectors are fed into the Transformer network for feature selection and classification.\n\n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [An improved GBSO-TAENN-based EEG signal classification model for epileptic seizure detection](https://pubmed.ncbi.nlm.nih.gov/38191643/) | Finite Linear Haar wavelet-based Filtering | - | EGG | - | difficult to understand this paper => non-ml approach |\n> Detection and classification of epileptic seizures from the EEG signals have gained significant attention in recent decades. Among other signals, EEG signals are extensively used by medical experts for diagnosing purposes. So, most of the existing research works developed automated mechanisms for designing an EEG-based epileptic seizure detection system. Machine learning techniques are highly used for reduced time consumption, high accuracy, and optimal performance. Still, it limits by the issues of high complexity in algorithm design, increased error value, and reduced detection efficacy. Thus, the proposed work intends to develop an automated epileptic seizure detection system with an improved performance rate. Here, the Finite Linear Haar wavelet-based Filtering (FLHF) technique is used to filter the input signals and the relevant set of features are extracted from the normalized output with the help of Fractal Dimension (FD) analysis. Then, the Grasshopper Bio-Inspired Swarm Optimization (GBSO) technique is employed to select the optimal features by computing the best fitness value and the Temporal Activation Expansive Neural Network (TAENN) mechanism is used for classifying the EEG signals to determine whether normal or seizure affected.\n\n**Last but not least**\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Mixture of Experts for EEG-Based Seizure Subtype Classification](https://pubmed.ncbi.nlm.nih.gov/38032784/) | - | - | MoE | - | never imagined MoE already used in this area => choice ensemble? or MoE? |\n> This paper proposes two novel Mixture of Experts (MoE) models, Seizure-MoE and Mix-MoE, for EEG-based seizure subtype classification. Particularly, Mix-MoE adequately addresses the above two challenges: 1) it introduces a novel imbalanced sampler to address significant class imbalance; and 2) it incorporates a priori knowledge of manual EEG features into the deep neural network to improve the classification performance. Experiments on two public datasets demonstrated that the proposed Seizure-MoE and Mix-MoE outperformed multiple existing approaches in cross-subject EEG-based seizure subtype classification. Our proposed MoE models may also be easily extended to other EEG classification problems with severe class imbalance, e.g., sleep stage classification.\n\n**Preprocessing**\n- [Ongoing EEG artifact correction using blind source separation](https://pubmed.ncbi.nlm.nih.gov/38219404/) -- anyone from this area can explain about it? -- is this noise removal type of preprocessing?\n\n\n**Domain Questions**\n-- what is blind source separation?\n-- what is path signature algorithm?\n-- not many recent papers used egg spectrogram?\n-- no wavnet used?\n\n**ML Questions**\n-- how we can use threshold for KL Metric?\n-- MoE vs Ensemble ?\n \n**Will start my notebook series for each paper, is good others also join in this series**\n**Share new papers if you come across interesting and i miss the respective architecture**", "votes": null }, { "id": "2620309", "postDate": "01/26/2024 03:56:17", "content": "

Thanks! But, wow, it's overwhelming to me to see all the work (listed here and in other posts) that has already been done on EEG signals and events detection and classification.

", "rawMarkdown": "Thanks! But, wow, it's overwhelming to me to see all the work (listed here and in other posts) that has already been done on EEG signals and events detection and classification.", "votes": null }, { "id": "2620364", "postDate": "01/26/2024 05:13:47", "content": "

Thank you @dan3dewey. I will working on papers to code module soon will release 1st paper notebook.

", "rawMarkdown": "Thank you @dan3dewey. I will working on papers to code module soon will release 1st paper notebook.", "votes": null } ], "comments": [ { "id": 2620309, "author_name": "dan3dewey", "author_url": "", "post_date": "01/26/2024 03:56:17", "content": "

Thanks! But, wow, it's overwhelming to me to see all the work (listed here and in other posts) that has already been done on EEG signals and events detection and classification.

", "votes": null, "replies": [ { "id": 2620364, "author_name": "seshurajup", "author_url": "", "post_date": "01/26/2024 05:13:47", "content": "

Thank you @dan3dewey. I will working on papers to code module soon will release 1st paper notebook.

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2606850": "*Note* : **Only from article abstract, recent papers and skipped if it mention similar model architecture**\n> **[Discussion](https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468684#2606504)** Motivated to explore different architectures in this domain \n> Haha, yeah. I think mega ensembles will do good in this competition. Also a multimodal model that can take all the data (spectrogram and eeg waveform and stacked over GBT output) and have the benefits of CNN, RNN, self attention will make very strong \"single\" models. by @cdeotte\n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Automatic epileptic seizure detection based on EEG using a moth-flame optimization of one-dimensional convolutional neural networks](https://pubmed.ncbi.nlm.nih.gov/38161793/) | - | 1D CNN | EGG | | | \n> This study proposes an automatic detection model for an EEG based on moth-flame optimization (MFO) optimized one-dimensional convolutional neural networks (1D-CNN). First, according to the characteristics and need for early epileptic seizure detection, a data augmentation method for dividing an EEG into small samples is proposed. Second, the hyperparameters are tuned based on MFO and trained for an EEG. Finally, the softmax classifier is used to output EEG classification from a small-sample and single channel.\n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [SMARTSeiz: Deep Learning with Attention Mechanism for Accurate Seizure Recognition in IoT Healthcare Devices](https://pubmed.ncbi.nlm.nih.gov/38055360/) | - | CNN + RNN | EGG | | | \n> Currently, doctors invest significant manual effort in inspecting Electroencephalograph (EEG) signals to identify seizure activity. However, EEG-based seizure detection algorithms face challenges in real-world scenarios due to non-stationary EEG data and variable seizure patterns among patients and recording sessions. Therefore, a sophisticated computer-based approach is necessary to analyze complex EEG records. In this work, the authors proposed a hybrid approach by combining traditional convolution neural (CN) and recurrent neural networks (RNN) along with an attention mechanism for the automatic recognition of epileptic seizures through EEG signal analysis. This attention mechanism focuses on significant subsets of EEG data for class recognition, resulting in improved model performance. \n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Epileptic Seizure Detection with an End-to-End Temporal Convolutional Network and Bidirectional Long Short-Term Memory Model](https://pubmed.ncbi.nlm.nih.gov/38230571/) | - | CNN + Bi-LSTM | EEG | 0.5-45 Hz band-pass | - | \n> Automatic seizure detection plays a key role in assisting clinicians for rapid diagnosis and treatment of epilepsy. In view of the parallelism of temporal convolutional network (TCN) and the capability of bidirectional long short-term memory (BiLSTM) in mining the long-range dependency of multi-channel time-series, we propose an automatic seizure detection method with a novel end-to-end TCN-BiLSTM model in this work. First, raw EEG is filtered with a 0.5-45 Hz band-pass filter, and the filtered data are input into the proposed TCN-BiLSTM network for feature extraction and classification. Post-processing process including moving average filtering, thresholding and collar technique is then employed to further improve the detection performance. The method was evaluated on two EEG database. \n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Epileptic Seizure Detection Based on Path Signature and Bi-LSTM Network With Attention Mechanism](https://pubmed.ncbi.nlm.nih.gov/38224524/) | path signature algorithm | Bi-LSTM + ATT | EEG | - | path signature algorithm ? |\n> Automatic seizure detection using electroen-cephalogram (EEG) can significantly expedite the diagnosis of epilepsy, thereby facilitating prompt treatment and reducing the risk of future seizures and associated complications. While most existing EEG-based epilepsy detection studies employ deep learning models, they often ignore the chronological relationships between different EEG channels. To tackle this limitation, a novel automatic epilepsy detection method is proposed, which leverages path signature and Bidirectional Long Short-Term Memory (Bi-LSTM) neural network with an attention mechanism. The path signature algorithm is used to extract discriminative features for capturing the dynamic dependencies between different channels of EEG, while Bi-LSTM with attention further analyzes the inherent temporal dependencies hidden in EEG signal features. \n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Automatic Seizure Detection Based on Stockwell Transform and Transformer](https://pubmed.ncbi.nlm.nih.gov/38202939/) | Go through paper | Transformer | EGG => Spectrogram | - | interesting to see post processing approach, how we can be use for KL metric? |\n> pilepsy is a chronic neurological disease associated with abnormal neuronal activity in the brain. Seizure detection algorithms are essential in reducing the workload of medical staff reviewing electroencephalogram (EEG) records. In this work, we propose a novel automatic epileptic EEG detection method based on Stockwell transform and Transformer. First, the S-transform is applied to the original EEG segments, acquiring accurate time-frequency representations. Subsequently, the obtained time-frequency matrices are grouped into different EEG rhythm blocks and compressed as vectors in these EEG sub-bands. After that, these feature vectors are fed into the Transformer network for feature selection and classification.\n\n\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [An improved GBSO-TAENN-based EEG signal classification model for epileptic seizure detection](https://pubmed.ncbi.nlm.nih.gov/38191643/) | Finite Linear Haar wavelet-based Filtering | - | EGG | - | difficult to understand this paper => non-ml approach |\n> Detection and classification of epileptic seizures from the EEG signals have gained significant attention in recent decades. Among other signals, EEG signals are extensively used by medical experts for diagnosing purposes. So, most of the existing research works developed automated mechanisms for designing an EEG-based epileptic seizure detection system. Machine learning techniques are highly used for reduced time consumption, high accuracy, and optimal performance. Still, it limits by the issues of high complexity in algorithm design, increased error value, and reduced detection efficacy. Thus, the proposed work intends to develop an automated epileptic seizure detection system with an improved performance rate. Here, the Finite Linear Haar wavelet-based Filtering (FLHF) technique is used to filter the input signals and the relevant set of features are extracted from the normalized output with the help of Fractal Dimension (FD) analysis. Then, the Grasshopper Bio-Inspired Swarm Optimization (GBSO) technique is employed to select the optimal features by computing the best fitness value and the Temporal Activation Expansive Neural Network (TAENN) mechanism is used for classifying the EEG signals to determine whether normal or seizure affected.\n\n**Last but not least**\n| Paper | Features | Model | Sequence | Filters | Notes |\n| --- | --- | --- | --- | --- | -- |\n| [Mixture of Experts for EEG-Based Seizure Subtype Classification](https://pubmed.ncbi.nlm.nih.gov/38032784/) | - | - | MoE | - | never imagined MoE already used in this area => choice ensemble? or MoE? |\n> This paper proposes two novel Mixture of Experts (MoE) models, Seizure-MoE and Mix-MoE, for EEG-based seizure subtype classification. Particularly, Mix-MoE adequately addresses the above two challenges: 1) it introduces a novel imbalanced sampler to address significant class imbalance; and 2) it incorporates a priori knowledge of manual EEG features into the deep neural network to improve the classification performance. Experiments on two public datasets demonstrated that the proposed Seizure-MoE and Mix-MoE outperformed multiple existing approaches in cross-subject EEG-based seizure subtype classification. Our proposed MoE models may also be easily extended to other EEG classification problems with severe class imbalance, e.g., sleep stage classification.\n\n**Preprocessing**\n- [Ongoing EEG artifact correction using blind source separation](https://pubmed.ncbi.nlm.nih.gov/38219404/) -- anyone from this area can explain about it? -- is this noise removal type of preprocessing?\n\n\n**Domain Questions**\n-- what is blind source separation?\n-- what is path signature algorithm?\n-- not many recent papers used egg spectrogram?\n-- no wavnet used?\n\n**ML Questions**\n-- how we can use threshold for KL Metric?\n-- MoE vs Ensemble ?\n \n**Will start my notebook series for each paper, is good others also join in this series**\n**Share new papers if you come across interesting and i miss the respective architecture**", "2620309": "Thanks! But, wow, it's overwhelming to me to see all the work (listed here and in other posts) that has already been done on EEG signals and events detection and classification.", "2620364": "Thank you @dan3dewey. I will working on papers to code module soon will release 1st paper notebook." }, "source": "meta" }