[论文解读] Implementation of Deep Neural Networks to Classify EEG Signals using Gramian Angular Summation Field for Epilepsy Diagnosis

Palani Thanaraj Krishnan, B. Parvathavarthini+4 more|arXiv (Cornell University)|Mar 8, 2020

EEG and Brain-Computer Interfaces参考文献 42被引用 42

一句话总结

该论文将 EEG 片段转换为 Gramian Angular Summation Field 图像，并评估深度 CNN（自定义 CNN 以及预训练的 AlexNet、VGG16、VGG19）在癫痫检测中的迁移学习和基于特征的 ANN 的表现，在 GASF 表征上取得了出色的指标。

ABSTRACT

This paper evaluates the approach of imaging timeseries data such as EEG in the diagnosis of epilepsy through Deep Neural Network (DNN). EEG signal is transformed into an RGB image using Gramian Angular Summation Field (GASF). Many such EEG epochs are transformed into GASF images for the normal and focal EEG signals. Then, some of the widely used Deep Neural Networks for image classification problems are used here to detect the focal GASF images. Three pre-trained DNN such as the AlexNet, VGG16, and VGG19 are validated for epilepsy detection based on the transfer learning approach. Furthermore, the textural features are extracted from GASF images, and prominent features are selected for a multilayer Artificial Neural Network (ANN) classifier. Lastly, a Custom Convolutional Neural Network (CNN) with three CNN layers, Batch Normalization, Max-pooling layer, and Dense layers, is proposed for epilepsy diagnosis from GASF images. The results of this paper show that the Custom CNN model was able to discriminate against the focal and normal GASF images with an average peak Precision of 0.885, Recall of 0.92, and F1-score of 0.90. Moreover, the Area Under the Curve (AUC) value of the Receiver Operating Characteristic (ROC) curve is 0.92 for the Custom CNN model. This paper suggests that Deep Learning methods widely used in image classification problems can be an alternative approach for epilepsy detection from EEG signals through GASF images.

研究动机与目标

• 利用 Gramian Angular Field 表征来激发用于癫痫诊断的时序 EEG 数据的成像应用。
• 在 GASF EEG 图像上评估基于预训练 DNN（AlexNet、VGG16、VGG19）的迁移学习。
• 为基于 GASF 图像的癫痫诊断开发自定义 CNN 架构。
• 从 GASF 图像提取纹理特征并在选定特征上使用多层前馈神经网络分类器。

提出的方法

• 将 EEG 片段转换为用于正常和癫痫焦点信号的 RGB GASF 图像。
• 应用 AlexNet、VGG16、VGG19 进行 GASF 图像的分类的迁移学习。
• 从 GASF 图像提取纹理特征，并选择显著特征用于多层 ANN 分类器。
• 提出一个具有三层卷积、批归一化、最大池化和全连接层的自定义 CNN 进行分类。
• 使用精确度、召回率、F1 分数和 AUC-ROC 等指标评估性能。

实验结果

研究问题

• RQ1 Can GASF representations of EEG signals enable effective epilepsy detection using image-based deep learning models?
• RQ2 How do pre-trained CNNs (AlexNet, VGG16, VGG19) perform on GASF-based epilepsy detection compared to a custom CNN?
• RQ3 Do textural features from GASF images improve epilepsy classification via a neural classifier?
• RQ4 What is the diagnostic performance (precision, recall, F1, AUC) of the proposed methods on GASF EEG images?

主要发现

• The custom CNN achieved average precision 0.885, recall 0.92, and F1-score 0.90.
• The custom CNN achieved an AUC of 0.92 on the ROC curve.
• Pre-trained CNNs and textural feature-based ANN were evaluated alongside the custom CNN, demonstrating feasibility of GASF-based epilepsy detection.
• Nine-page paper includes 8 figures and 2 tables supporting the results.