MLP-UNet: Glomerulus Segmentation
Glomerulus segmentation in kidney tissue segments is a crucial nephropathology process used to diagnose renal diseases effectively. This study proposes a novel and robust application of MLP (Multi-Layer Perceptron) based architectures for the segmentation of glomeruli in PAS (Periodic AcidSchiff) st...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10138195/ |
| _version_ | 1797810677447917568 |
|---|---|
| author | Franchis N. Saikia Yuji Iwahori Taisei Suzuki M. K. Bhuyan Aili Wang Boonserm Kijsirikul |
| author_facet | Franchis N. Saikia Yuji Iwahori Taisei Suzuki M. K. Bhuyan Aili Wang Boonserm Kijsirikul |
| author_sort | Franchis N. Saikia |
| collection | DOAJ |
| description | Glomerulus segmentation in kidney tissue segments is a crucial nephropathology process used to diagnose renal diseases effectively. This study proposes a novel and robust application of MLP (Multi-Layer Perceptron) based architectures for the segmentation of glomeruli in PAS (Periodic AcidSchiff) stained whole renal images for effective diagnosis of renal diseases. For the segmentation challenge, the proposed unique solution uses MLP-UNet (Multi-Layer Perceptron U-Net), a novel design that evades using conventional convolution and self-attention mechanisms. Additionally, the study compares various approaches, including U-Net, and for the first time, trains the TransUNet model on the kidney WSI (Whole Slide Image) dataset. Dice Score and Dice Loss were used for training these models as the metric and loss function. Results showed that MLP-based architectures provide comparable results (89.96%) to pre-trained architectures like TransUNet (90.58%) with effectively 20% lesser parameters and no pre-training, and also produce superior Dice scores across the 5-fold cross-validation training and learn more quickly than conventional U-Net architectures. |
| first_indexed | 2024-03-13T07:12:27Z |
| format | Article |
| id | doaj.art-f742e7c046c44d1196c93712ec212fa6 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-13T07:12:27Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-f742e7c046c44d1196c93712ec212fa62023-06-05T23:00:17ZengIEEEIEEE Access2169-35362023-01-0111530345304710.1109/ACCESS.2023.328083110138195MLP-UNet: Glomerulus SegmentationFranchis N. Saikia0https://orcid.org/0000-0002-9544-5971Yuji Iwahori1https://orcid.org/0000-0002-6421-8186Taisei Suzuki2M. K. Bhuyan3https://orcid.org/0000-0003-2152-5466Aili Wang4https://orcid.org/0000-0002-9118-230XBoonserm Kijsirikul5https://orcid.org/0000-0002-9046-7151Department of Electronics and Communication Engineering, Indian Institute of Technology Guwahati, Guwahati, IndiaDepartment of Computer Science, Chubu University, Kasugai, JapanDepartment of Nephrology, Nagoya City University East Medical Center, Nagoya, JapanDepartment of Electronics and Electrical Engineering, Indian Institute of Technology Guwahati, Guwahati, IndiaHigher Education Key Laboratory for Measuring and Control Technology and Instrumentations of Heilongjiang, Harbin University of Science and Technology, Harbin, ChinaDepartment of Computer Engineering, Faculty of Engineering, Chulalongkorn University, Bangkok, ThailandGlomerulus segmentation in kidney tissue segments is a crucial nephropathology process used to diagnose renal diseases effectively. This study proposes a novel and robust application of MLP (Multi-Layer Perceptron) based architectures for the segmentation of glomeruli in PAS (Periodic AcidSchiff) stained whole renal images for effective diagnosis of renal diseases. For the segmentation challenge, the proposed unique solution uses MLP-UNet (Multi-Layer Perceptron U-Net), a novel design that evades using conventional convolution and self-attention mechanisms. Additionally, the study compares various approaches, including U-Net, and for the first time, trains the TransUNet model on the kidney WSI (Whole Slide Image) dataset. Dice Score and Dice Loss were used for training these models as the metric and loss function. Results showed that MLP-based architectures provide comparable results (89.96%) to pre-trained architectures like TransUNet (90.58%) with effectively 20% lesser parameters and no pre-training, and also produce superior Dice scores across the 5-fold cross-validation training and learn more quickly than conventional U-Net architectures.https://ieeexplore.ieee.org/document/10138195/Computer visionimage segmentationsemantic segmentationdeep learningbiomedical image processingartificial intelligence |
| spellingShingle | Franchis N. Saikia Yuji Iwahori Taisei Suzuki M. K. Bhuyan Aili Wang Boonserm Kijsirikul MLP-UNet: Glomerulus Segmentation IEEE Access Computer vision image segmentation semantic segmentation deep learning biomedical image processing artificial intelligence |
| title | MLP-UNet: Glomerulus Segmentation |
| title_full | MLP-UNet: Glomerulus Segmentation |
| title_fullStr | MLP-UNet: Glomerulus Segmentation |
| title_full_unstemmed | MLP-UNet: Glomerulus Segmentation |
| title_short | MLP-UNet: Glomerulus Segmentation |
| title_sort | mlp unet glomerulus segmentation |
| topic | Computer vision image segmentation semantic segmentation deep learning biomedical image processing artificial intelligence |
| url | https://ieeexplore.ieee.org/document/10138195/ |
| work_keys_str_mv | AT franchisnsaikia mlpunetglomerulussegmentation AT yujiiwahori mlpunetglomerulussegmentation AT taiseisuzuki mlpunetglomerulussegmentation AT mkbhuyan mlpunetglomerulussegmentation AT ailiwang mlpunetglomerulussegmentation AT boonsermkijsirikul mlpunetglomerulussegmentation |