CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 Estimation
Video-based peripheral oxygen saturation (SpO2) estimation, utilizing solely RGB cameras, offers a non-contact approach to measuring blood oxygen levels. Previous studies set a stable and unchanging environment as the premise for non-contact blood oxygen estimation. Additionally, they utilized a sma...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-01-01
|
| Series: | Bioengineering |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2306-5354/11/2/113 |
| _version_ | 1797298868786823168 |
|---|---|
| author | Jiahe Peng Weihua Su Haiyong Chen Jingsheng Sun Zandong Tian |
| author_facet | Jiahe Peng Weihua Su Haiyong Chen Jingsheng Sun Zandong Tian |
| author_sort | Jiahe Peng |
| collection | DOAJ |
| description | Video-based peripheral oxygen saturation (SpO2) estimation, utilizing solely RGB cameras, offers a non-contact approach to measuring blood oxygen levels. Previous studies set a stable and unchanging environment as the premise for non-contact blood oxygen estimation. Additionally, they utilized a small amount of labeled data for system training and learning. However, it is challenging to train optimal model parameters with a small dataset. The accuracy of blood oxygen detection is easily affected by ambient light and subject movement. To address these issues, this paper proposes a contrastive learning spatiotemporal attention network (CL-SPO2Net), an innovative semi-supervised network for video-based SpO2 estimation. Spatiotemporal similarities in remote photoplethysmography (rPPG) signals were found in video segments containing facial or hand regions. Subsequently, integrating deep neural networks with machine learning expertise enabled the estimation of SpO2. The method had good feasibility in the case of small-scale labeled datasets, with the mean absolute error between the camera and the reference pulse oximeter of 0.85% in the stable environment, 1.13% with lighting fluctuations, and 1.20% in the facial rotation situation. |
| first_indexed | 2024-03-07T22:41:21Z |
| format | Article |
| id | doaj.art-1ada87bdcbb64740ba35f84111eb1ed6 |
| institution | Directory Open Access Journal |
| issn | 2306-5354 |
| language | English |
| last_indexed | 2024-03-07T22:41:21Z |
| publishDate | 2024-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Bioengineering |
| spelling | doaj.art-1ada87bdcbb64740ba35f84111eb1ed62024-02-23T15:07:50ZengMDPI AGBioengineering2306-53542024-01-0111211310.3390/bioengineering11020113CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 EstimationJiahe Peng0Weihua Su1Haiyong Chen2Jingsheng Sun3Zandong Tian4School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, ChinaSchool of Mechanical Engineering, Hebei University of Technology, Tianjin 300401, ChinaSchool of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, ChinaSchool of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, ChinaSchool of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, ChinaVideo-based peripheral oxygen saturation (SpO2) estimation, utilizing solely RGB cameras, offers a non-contact approach to measuring blood oxygen levels. Previous studies set a stable and unchanging environment as the premise for non-contact blood oxygen estimation. Additionally, they utilized a small amount of labeled data for system training and learning. However, it is challenging to train optimal model parameters with a small dataset. The accuracy of blood oxygen detection is easily affected by ambient light and subject movement. To address these issues, this paper proposes a contrastive learning spatiotemporal attention network (CL-SPO2Net), an innovative semi-supervised network for video-based SpO2 estimation. Spatiotemporal similarities in remote photoplethysmography (rPPG) signals were found in video segments containing facial or hand regions. Subsequently, integrating deep neural networks with machine learning expertise enabled the estimation of SpO2. The method had good feasibility in the case of small-scale labeled datasets, with the mean absolute error between the camera and the reference pulse oximeter of 0.85% in the stable environment, 1.13% with lighting fluctuations, and 1.20% in the facial rotation situation.https://www.mdpi.com/2306-5354/11/2/113contrastive learningperipheral oxygen saturation (SpO2)computer visiondeep learningremote photoplethysmography (rPPG)attention mechanism |
| spellingShingle | Jiahe Peng Weihua Su Haiyong Chen Jingsheng Sun Zandong Tian CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 Estimation Bioengineering contrastive learning peripheral oxygen saturation (SpO2) computer vision deep learning remote photoplethysmography (rPPG) attention mechanism |
| title | CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 Estimation |
| title_full | CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 Estimation |
| title_fullStr | CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 Estimation |
| title_full_unstemmed | CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 Estimation |
| title_short | CL-SPO2Net: Contrastive Learning Spatiotemporal Attention Network for Non-Contact Video-Based SpO2 Estimation |
| title_sort | cl spo2net contrastive learning spatiotemporal attention network for non contact video based spo2 estimation |
| topic | contrastive learning peripheral oxygen saturation (SpO2) computer vision deep learning remote photoplethysmography (rPPG) attention mechanism |
| url | https://www.mdpi.com/2306-5354/11/2/113 |
| work_keys_str_mv | AT jiahepeng clspo2netcontrastivelearningspatiotemporalattentionnetworkfornoncontactvideobasedspo2estimation AT weihuasu clspo2netcontrastivelearningspatiotemporalattentionnetworkfornoncontactvideobasedspo2estimation AT haiyongchen clspo2netcontrastivelearningspatiotemporalattentionnetworkfornoncontactvideobasedspo2estimation AT jingshengsun clspo2netcontrastivelearningspatiotemporalattentionnetworkfornoncontactvideobasedspo2estimation AT zandongtian clspo2netcontrastivelearningspatiotemporalattentionnetworkfornoncontactvideobasedspo2estimation |