USSD: Unsupervised Sleep Spindle Detector
Sleep spindles (SSs) appear in electroencephalogram (EEG) recordings during sleep stage N2, and they are usually detected through visual inspection by an expert. Labeling SSs in large datasets is time-consuming and depends on the expert criteria. In this work, we propose an unsupervised SS detector...
Saved in:
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2025-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/10848108/ |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
_version_ | 1832576759987437568 |
---|---|
author | Edgardo Ramirez Pablo A. Estevez Martin D. Adams Claudio A. Perez Marcelo Garrido Gonzalez Patricio Peirano |
author_facet | Edgardo Ramirez Pablo A. Estevez Martin D. Adams Claudio A. Perez Marcelo Garrido Gonzalez Patricio Peirano |
author_sort | Edgardo Ramirez |
collection | DOAJ |
description | Sleep spindles (SSs) appear in electroencephalogram (EEG) recordings during sleep stage N2, and they are usually detected through visual inspection by an expert. Labeling SSs in large datasets is time-consuming and depends on the expert criteria. In this work, we propose an unsupervised SS detector based on dictionary learning called the Unsupervised Sleep Spindle Detector (USSD). The proposed detector learns prototype SSs of different lengths (called atoms). An unsupervised adaptive threshold method based on the distribution of the automatically detected SS lengths is developed, which allows the adaptation of the USSD algorithm to different datasets in an unsupervised way. For each detection, the USSD estimates the probability of being an SS. The USSD performances on the labeled MASS-SS2 and INTA-UCH datasets yield F1-scores of <inline-formula> <tex-math notation="LaTeX">$0.72 \pm 0.02$ </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">$0.72 \pm 0.04$ </tex-math></inline-formula>, respectively. The USSD outperforms the A7 and LUNA detectors, which are traditional unsupervised models. Next, we fine-tune the resulting USSD model with 20% of the labeled MASS-SS2 and INTA-UCH datasets, achieving F1 scores of <inline-formula> <tex-math notation="LaTeX">$0.78 \pm 0.06$ </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">$0.75 \pm 0.05$ </tex-math></inline-formula>, respectively. In addition, the SSs detected by USSD on the unlabeled CAP dataset are used to pre-train a supervised deep learning method, which after fine-tuning with 20% of the MODA dataset, reaches an F1-score of <inline-formula> <tex-math notation="LaTeX">$0.81 \pm 0.02$ </tex-math></inline-formula>. |
format | Article |
id | doaj-art-660ba8cd69e74e88afdf5130348d3771 |
institution | Kabale University |
issn | 2169-3536 |
language | English |
publishDate | 2025-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Access |
spelling | doaj-art-660ba8cd69e74e88afdf5130348d37712025-01-31T00:01:30ZengIEEEIEEE Access2169-35362025-01-0113186441865910.1109/ACCESS.2025.353253610848108USSD: Unsupervised Sleep Spindle DetectorEdgardo Ramirez0https://orcid.org/0009-0004-2862-180XPablo A. Estevez1https://orcid.org/0000-0001-9164-4722Martin D. Adams2https://orcid.org/0000-0002-1085-0506Claudio A. Perez3https://orcid.org/0000-0002-5484-4159Marcelo Garrido Gonzalez4https://orcid.org/0000-0002-6027-3735Patricio Peirano5https://orcid.org/0000-0002-4387-3982Department of Electrical Engineering, University of Chile, Santiago, ChileDepartment of Electrical Engineering, University of Chile, Santiago, ChileDepartment of Electrical Engineering, University of Chile, Santiago, ChileDepartment of Electrical Engineering, University of Chile, Santiago, ChileSleep Laboratory, Institute of Nutrition and Food Technology, University of Chile, Santiago, ChileSleep Laboratory, Institute of Nutrition and Food Technology, University of Chile, Santiago, ChileSleep spindles (SSs) appear in electroencephalogram (EEG) recordings during sleep stage N2, and they are usually detected through visual inspection by an expert. Labeling SSs in large datasets is time-consuming and depends on the expert criteria. In this work, we propose an unsupervised SS detector based on dictionary learning called the Unsupervised Sleep Spindle Detector (USSD). The proposed detector learns prototype SSs of different lengths (called atoms). An unsupervised adaptive threshold method based on the distribution of the automatically detected SS lengths is developed, which allows the adaptation of the USSD algorithm to different datasets in an unsupervised way. For each detection, the USSD estimates the probability of being an SS. The USSD performances on the labeled MASS-SS2 and INTA-UCH datasets yield F1-scores of <inline-formula> <tex-math notation="LaTeX">$0.72 \pm 0.02$ </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">$0.72 \pm 0.04$ </tex-math></inline-formula>, respectively. The USSD outperforms the A7 and LUNA detectors, which are traditional unsupervised models. Next, we fine-tune the resulting USSD model with 20% of the labeled MASS-SS2 and INTA-UCH datasets, achieving F1 scores of <inline-formula> <tex-math notation="LaTeX">$0.78 \pm 0.06$ </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">$0.75 \pm 0.05$ </tex-math></inline-formula>, respectively. In addition, the SSs detected by USSD on the unlabeled CAP dataset are used to pre-train a supervised deep learning method, which after fine-tuning with 20% of the MODA dataset, reaches an F1-score of <inline-formula> <tex-math notation="LaTeX">$0.81 \pm 0.02$ </tex-math></inline-formula>.https://ieeexplore.ieee.org/document/10848108/EEGsleep spindleunsupervised learningdictionary learning |
spellingShingle | Edgardo Ramirez Pablo A. Estevez Martin D. Adams Claudio A. Perez Marcelo Garrido Gonzalez Patricio Peirano USSD: Unsupervised Sleep Spindle Detector IEEE Access EEG sleep spindle unsupervised learning dictionary learning |
title | USSD: Unsupervised Sleep Spindle Detector |
title_full | USSD: Unsupervised Sleep Spindle Detector |
title_fullStr | USSD: Unsupervised Sleep Spindle Detector |
title_full_unstemmed | USSD: Unsupervised Sleep Spindle Detector |
title_short | USSD: Unsupervised Sleep Spindle Detector |
title_sort | ussd unsupervised sleep spindle detector |
topic | EEG sleep spindle unsupervised learning dictionary learning |
url | https://ieeexplore.ieee.org/document/10848108/ |
work_keys_str_mv | AT edgardoramirez ussdunsupervisedsleepspindledetector AT pabloaestevez ussdunsupervisedsleepspindledetector AT martindadams ussdunsupervisedsleepspindledetector AT claudioaperez ussdunsupervisedsleepspindledetector AT marcelogarridogonzalez ussdunsupervisedsleepspindledetector AT patriciopeirano ussdunsupervisedsleepspindledetector |