A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRI
Abstract Parkinson’s disease ranks as the second most prevalent neurological disorder after Alzheimer’s disease. Convolutional neural networks (CNNs) have been extensively employed in Parkinson’s disease (PD) detection using MR images. However, CNN models generally focus on local features while pron...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-04-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-93671-5 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849314703083307008 |
|---|---|
| author | Sayyed Shahid Hussain Pir Masoom Shah Hussain Dawood Xu Degang Ahmad Alshamayleh Muhammad Adnan Khan Taher M. Ghazal |
| author_facet | Sayyed Shahid Hussain Pir Masoom Shah Hussain Dawood Xu Degang Ahmad Alshamayleh Muhammad Adnan Khan Taher M. Ghazal |
| author_sort | Sayyed Shahid Hussain |
| collection | DOAJ |
| description | Abstract Parkinson’s disease ranks as the second most prevalent neurological disorder after Alzheimer’s disease. Convolutional neural networks (CNNs) have been extensively employed in Parkinson’s disease (PD) detection using MR images. However, CNN models generally focus on local features while prone to capture global representations. On the other hand, the vision transformer (ViT) excels at capturing global features through its self-attention mechanism, but it compromises local feature representations. Additionally, the varying magnitude of MR data poses a challenge for ViT, potentially leading to the gradient vanishing problem. To address these limitations, this paper proposed a novel framework that combines the Swin-Transformer and CNN to capture both local and global features effectively. To mitigate the gradient vanishing issue in ViT, we used skipped connections and cosine attention mechanism in VIT that preserves the output distribution regardless of input magnitude variations. The proposed model comprises three primary blocks: Transformer-block, convolutional block, and dense-block. The input image is processed concurrently by the cosine transformer and convolutional block. Subsequently, the extracted features from both blocks are concatenated and fed to the dense block for decision-making. The proposed model achieved promesing results of 96%, 97%, 95%, and 95% in terms of accuracy, sensitivity, specificity, and area under the curve, respectively. |
| format | Article |
| id | doaj-art-3785fcc59e6c447cb8311520f8e1e1e4 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-3785fcc59e6c447cb8311520f8e1e1e42025-08-20T03:52:23ZengNature PortfolioScientific Reports2045-23222025-04-0115111410.1038/s41598-025-93671-5A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRISayyed Shahid Hussain0Pir Masoom Shah1Hussain Dawood2Xu Degang3Ahmad Alshamayleh4Muhammad Adnan Khan5Taher M. Ghazal6School of Automation, Central South UniversitySchool of Computer Science and Engineering, Central South UniversitySchool of Computing, Skyline University CollegeSchool of Automation, Central South UniversityDepartment of Data Science and Artificial Intelligence, Hourani Center for Applied Scientific Research, Al-Ahliyya Amman UniversitySchool of Computing, Skyline University CollegeNetworks and Cybersecurity Department, Hourani Center for Applied Scientific Research, Al-Ahliyya Amman UniversityAbstract Parkinson’s disease ranks as the second most prevalent neurological disorder after Alzheimer’s disease. Convolutional neural networks (CNNs) have been extensively employed in Parkinson’s disease (PD) detection using MR images. However, CNN models generally focus on local features while prone to capture global representations. On the other hand, the vision transformer (ViT) excels at capturing global features through its self-attention mechanism, but it compromises local feature representations. Additionally, the varying magnitude of MR data poses a challenge for ViT, potentially leading to the gradient vanishing problem. To address these limitations, this paper proposed a novel framework that combines the Swin-Transformer and CNN to capture both local and global features effectively. To mitigate the gradient vanishing issue in ViT, we used skipped connections and cosine attention mechanism in VIT that preserves the output distribution regardless of input magnitude variations. The proposed model comprises three primary blocks: Transformer-block, convolutional block, and dense-block. The input image is processed concurrently by the cosine transformer and convolutional block. Subsequently, the extracted features from both blocks are concatenated and fed to the dense block for decision-making. The proposed model achieved promesing results of 96%, 97%, 95%, and 95% in terms of accuracy, sensitivity, specificity, and area under the curve, respectively.https://doi.org/10.1038/s41598-025-93671-5Parkinson’s diseaseSwin-transformerCosine similarityCNNMRI |
| spellingShingle | Sayyed Shahid Hussain Pir Masoom Shah Hussain Dawood Xu Degang Ahmad Alshamayleh Muhammad Adnan Khan Taher M. Ghazal A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRI Scientific Reports Parkinson’s disease Swin-transformer Cosine similarity CNN MRI |
| title | A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRI |
| title_full | A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRI |
| title_fullStr | A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRI |
| title_full_unstemmed | A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRI |
| title_short | A swin transformer and CNN fusion framework for accurate Parkinson disease classification in MRI |
| title_sort | swin transformer and cnn fusion framework for accurate parkinson disease classification in mri |
| topic | Parkinson’s disease Swin-transformer Cosine similarity CNN MRI |
| url | https://doi.org/10.1038/s41598-025-93671-5 |
| work_keys_str_mv | AT sayyedshahidhussain aswintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT pirmasoomshah aswintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT hussaindawood aswintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT xudegang aswintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT ahmadalshamayleh aswintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT muhammadadnankhan aswintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT tahermghazal aswintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT sayyedshahidhussain swintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT pirmasoomshah swintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT hussaindawood swintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT xudegang swintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT ahmadalshamayleh swintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT muhammadadnankhan swintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri AT tahermghazal swintransformerandcnnfusionframeworkforaccurateparkinsondiseaseclassificationinmri |