UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic image
Abstract Two-photon microscopy is indispensable in cell and molecular biology for its high-resolution visualization of cellular and molecular dynamics. However, the inevitable low signal-to-noise conditions significantly degrade image quality, obscuring essential details and complicating morphologic...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2024-11-01
|
| Series: | Complex & Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s40747-024-01633-7 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832571204379082752 |
|---|---|
| author | Yuer Lu Yongfa Ying Chen Lin Yan Wang Jun Jin Xiaoming Jiang Jianwei Shuai Xiang Li Jinjin Zhong |
| author_facet | Yuer Lu Yongfa Ying Chen Lin Yan Wang Jun Jin Xiaoming Jiang Jianwei Shuai Xiang Li Jinjin Zhong |
| author_sort | Yuer Lu |
| collection | DOAJ |
| description | Abstract Two-photon microscopy is indispensable in cell and molecular biology for its high-resolution visualization of cellular and molecular dynamics. However, the inevitable low signal-to-noise conditions significantly degrade image quality, obscuring essential details and complicating morphological analysis. While existing denoising methods such as CNNs, Noise2Noise, and DeepCAD serve broad applications in imaging, they still have limitations in preserving texture structures and fine details in two-photon microscopic images affected by complex noise, particularly in sophisticated structures like neuronal synapses. To improve two-photon microscopy image denoising effectiveness, by experimenting on real two-photon microscopy images, we propose a novel deep learning framework, the UNet-Att model, which integrates a specifically tailored UNet++ architecture with attention mechanisms. Specifically, this approach consists of a sophisticated downsampling module for extracting hierarchical features at varied scales, and an innovative attention module that strategically emphasizes salient features during the integration process. The architecture is completed by an ingenious upsampling pathway that reconstructs the image with high fidelity, ensuring the retention of textural integrity. Additionally, the model supports end-to-end training, optimizing its denoising efficacy. The UNet-Att model proves to surpass mainstream algorithms in the dual objectives of denoising and preserving the textural intricacies of original images, which is evidenced by an increase of 9.42 dB in the high Peak Signal-to-Noise Ratio (PSNR) coupled with an improvement of 0.1131 in the Structural Similarity Index Measurement (SSIM). The ablation experiments reveal the effectiveness of each module. The associated Python packages and datasets of UNet-Att are freely available at https://github.com/ZjjDh/UNet-Att . |
| format | Article |
| id | doaj-art-e32be553036c4af8a83b4d70f851a8ac |
| institution | Kabale University |
| issn | 2199-4536 2198-6053 |
| language | English |
| publishDate | 2024-11-01 |
| publisher | Springer |
| record_format | Article |
| series | Complex & Intelligent Systems |
| spelling | doaj-art-e32be553036c4af8a83b4d70f851a8ac2025-02-02T12:50:18ZengSpringerComplex & Intelligent Systems2199-45362198-60532024-11-0111111710.1007/s40747-024-01633-7UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic imageYuer Lu0Yongfa Ying1Chen Lin2Yan Wang3Jun Jin4Xiaoming Jiang5Jianwei Shuai6Xiang Li7Jinjin Zhong8Wenzhou Key Laboratory of Biophysics, Wenzhou Institute, University of Chinese Academy of SciencesDepartment of Physics, and Fujian Provincial Key Laboratory for Soft Functional Materials Research, Xiamen UniversityDepartment of Physics, and Fujian Provincial Key Laboratory for Soft Functional Materials Research, Xiamen UniversityMcCormick School of Engineering, Northwestern UniversityDepartment of Physics, and Fujian Provincial Key Laboratory for Soft Functional Materials Research, Xiamen UniversityWenzhou Key Laboratory of Biophysics, Wenzhou Institute, University of Chinese Academy of SciencesWenzhou Key Laboratory of Biophysics, Wenzhou Institute, University of Chinese Academy of SciencesDepartment of Physics, and Fujian Provincial Key Laboratory for Soft Functional Materials Research, Xiamen UniversityWenzhou Key Laboratory of Biophysics, Wenzhou Institute, University of Chinese Academy of SciencesAbstract Two-photon microscopy is indispensable in cell and molecular biology for its high-resolution visualization of cellular and molecular dynamics. However, the inevitable low signal-to-noise conditions significantly degrade image quality, obscuring essential details and complicating morphological analysis. While existing denoising methods such as CNNs, Noise2Noise, and DeepCAD serve broad applications in imaging, they still have limitations in preserving texture structures and fine details in two-photon microscopic images affected by complex noise, particularly in sophisticated structures like neuronal synapses. To improve two-photon microscopy image denoising effectiveness, by experimenting on real two-photon microscopy images, we propose a novel deep learning framework, the UNet-Att model, which integrates a specifically tailored UNet++ architecture with attention mechanisms. Specifically, this approach consists of a sophisticated downsampling module for extracting hierarchical features at varied scales, and an innovative attention module that strategically emphasizes salient features during the integration process. The architecture is completed by an ingenious upsampling pathway that reconstructs the image with high fidelity, ensuring the retention of textural integrity. Additionally, the model supports end-to-end training, optimizing its denoising efficacy. The UNet-Att model proves to surpass mainstream algorithms in the dual objectives of denoising and preserving the textural intricacies of original images, which is evidenced by an increase of 9.42 dB in the high Peak Signal-to-Noise Ratio (PSNR) coupled with an improvement of 0.1131 in the Structural Similarity Index Measurement (SSIM). The ablation experiments reveal the effectiveness of each module. The associated Python packages and datasets of UNet-Att are freely available at https://github.com/ZjjDh/UNet-Att .https://doi.org/10.1007/s40747-024-01633-7Two-photon microscopy imageImage denoisingDeep learningAttentionUnet++ |
| spellingShingle | Yuer Lu Yongfa Ying Chen Lin Yan Wang Jun Jin Xiaoming Jiang Jianwei Shuai Xiang Li Jinjin Zhong UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic image Complex & Intelligent Systems Two-photon microscopy image Image denoising Deep learning Attention Unet++ |
| title | UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic image |
| title_full | UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic image |
| title_fullStr | UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic image |
| title_full_unstemmed | UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic image |
| title_short | UNet-Att: a self-supervised denoising and recovery model for two-photon microscopic image |
| title_sort | unet att a self supervised denoising and recovery model for two photon microscopic image |
| topic | Two-photon microscopy image Image denoising Deep learning Attention Unet++ |
| url | https://doi.org/10.1007/s40747-024-01633-7 |
| work_keys_str_mv | AT yuerlu unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT yongfaying unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT chenlin unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT yanwang unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT junjin unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT xiaomingjiang unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT jianweishuai unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT xiangli unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage AT jinjinzhong unetattaselfsuperviseddenoisingandrecoverymodelfortwophotonmicroscopicimage |