Dilated 1-D CNN for Identifying ALS from Raw EMG Signal

This paper presents ALSNET, a Dilated 1-D CNN model developed to identify Amyotrophic Lateral Sclerosis (ALS) from raw Electromyography (EMG) signals. The introduction explores ALS, its impact on motor neurons, and the importance of early diagnosis. EMG signals and various features used in ALS identification are discussed. The paper compares traditional approaches with the proposed CNN model and highlights the performance evaluation of ALSNet.

• ALS identification
• EMG signals
• CNN model
• Motor neuron diseases
• Medical research

patt854 Follow

Uploaded on Feb 22, 2025 | 0 Views

Download Presentation

Please find below an Image/Link to download the presentation.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.

You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.

Presentation Transcript

1. ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL Paper ID: 8762 K. M. Naimul Hassan, Md. Shamiul Alam Hridoy, Naima Tasnim, Atia Faria Chowdhury, Tanvir Alam Roni, Sheikh Tabrez, Arik Subhana, Celia Shahnaz Department of Electrical and Electronic Engineering (EEE), Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh

2. OUTLINE Introduction Existing Research Proposed Methodology ALSNet: A Dilated 1-D CNN Dataset ALSNet Training Performance Evaluation Summary ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

3. INTRODUCTION Amyotrophic Lateral Sclerosis (ALS)- oOne of the most common neuromuscular diseases oAffects both lower and upper motor neurons oDevelopment of symptoms over a long period of time oEarly diagnosis needed for prevention of the disease and improve the quality of life for ALS patients Image source ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

4. INTRODUCTION Electromyography (EMG)- oBio-signal consisting of several motor unit action potentials (MUAPs) oVarious time and frequency domain features explored to identify ALS from EMG signals Image source ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

5. EXISTING RESEARCH Traditional approaches involving manual feature extraction Feature Extraction Method Classifier Mel-frequency cepstral coefficient (MFCC) K-nearest neighbors (KNN) Hand-crafted feature extraction Discrete cosine transform (DCT) KNN Spectral feature extraction from dominant MUAP of EMG KNN Intrinsic mode functions (IMFs) using empirical mode decom position (EMD) Least square support vector machine (LS-SVM) Short time Fourier transform (STFT) Convolutional neural network (CNN) Time-frequency (T-F) representation of EMG signal Spectrogram Continuous wavelet transform (CWT) CNN CNN Smoothed pseudo-Wigner-Ville distribution (SPWVD) CNN ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

6. MOTIVATION A classifier taking raw EMG signal as input not considered in previous research works Motivation- oReduction in computational cost by eliminating manual feature extraction step oFaster and better practical implementation ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

7. PROPOSED METHODOLOGY Supervised Binary classification Output: ALS (1) or Normal (0) Raw EMG Signal ALSNet ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

8. ALSNet: A DILATED 1-D CNN Increase in dilation rate: Widens the gap between two kernels and helps to integrate more information from a wider context o Previously applied successfully in biomedical image segmentation, speech synthesis and sound source localization ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

9. DATASET N2001 EMGLAB 10 normal subjects (4 females and 6 males) aged between 21-37 years 8 normal subjects (4 females and 4 males) aged between 35-67 years All the EMG signals were recorded under usual conditions for MUAP analysis ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

10. DATASET Sampled at 24 kHz frequency and recorded for almost 11 seconds A total number of 302 EMG signals recorded from the brachial biceps and medial vastus muscles were used (151-Normal, 151- ALS) Segmented into 1s windows Train: Validation: Test= 80:20:25 ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

11. ALSNet TRAINING GPU Accelerator Loss function Optimizer Initial learning rate No. of epochs Batch Size Execution time Binary Cross-entropy Adam 0.001 67 48 721.7s ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

12. PERFORMANCE EVALUATION ??+?? ??????? ????????: ??+??+??+?? ?? ??+?? ?? ??+?? ??+?? 2 ??????????? (??): ???????????(??): ???????? ????????: ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

13. PERFORMANCE EVALUATION Existing Method MFCC + KNN DCT + KNN EMD + LS-SVM Overall Accuracy (%) 92.50 95.00 95.00 Sensitivity (%) Specificity (%) Balanced Accuracy (%) 87.00 92.00 92.75 76.00 86.00 93.00 98.00 98.00 92.54 MUAP + KNN T-F + CNN 2D-1 96.5 96.69 88 99.33 97.59 93.67 95.92 94.24 T-F + CNN 2D-2 96.80 94.8 98.8 96.8 ALSNet 97.74 96.77 98.59 97.68 ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

14. SUMMARY A 1D dilated convolutional neural network is proposed for identifying ALS from raw EMG signal Performance: Showing good promise Can be useful in early diagnosis of ALS Practically implementable ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN

15. THANK YOU K. M. Naimul Hassan Md. Shamiul Alam Hridoy Naima Tasnim Atia Faria Chowdhury Tanvir Alam Roni Sheikh Tabrez Arik Subhana Celia Shahnaz Department of Electrical and Electronic Engineering (EEE) Bangladesh University of Engineering and Technology (BUET) West Palashi Campus, Dhaka-1205, Bangladesh ALSNET: A DILATED 1-D CNN FOR IDENTIFYING ALS FROM RAW EMG SIGNAL, PRESENTER: K. M. NAIMUL HASSAN