Differential evolution-based neural architecture search for brain vessel segmentation
Brain vasculature analysis is critical in developing novel treatment targets for neurodegenerative diseases. Such an accurate analysis cannot be performed manually but requires a semi-automated or fully-automated approach. Deep learning methods have recently proven indispensable for the automated se...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-10-01
|
| Series: | Engineering Science and Technology, an International Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2215098623001805 |
| _version_ | 1797683049827139584 |
|---|---|
| author | Zeki Kuş Berna Kiraz Tuğçe Koçak Göksu Musa Aydın Esra Özkan Atay Vural Alper Kiraz Burhanettin Can |
| author_facet | Zeki Kuş Berna Kiraz Tuğçe Koçak Göksu Musa Aydın Esra Özkan Atay Vural Alper Kiraz Burhanettin Can |
| author_sort | Zeki Kuş |
| collection | DOAJ |
| description | Brain vasculature analysis is critical in developing novel treatment targets for neurodegenerative diseases. Such an accurate analysis cannot be performed manually but requires a semi-automated or fully-automated approach. Deep learning methods have recently proven indispensable for the automated segmentation and analysis of medical images. However, optimizing a deep learning network architecture is another challenge. Manually selecting deep learning network architectures and tuning their hyper-parameters requires a lot of expertise and effort. To solve this problem, neural architecture search (NAS) approaches that explore more efficient network architectures with high segmentation performance have been proposed in the literature. This study introduces differential evolution-based NAS approaches in which a novel search space is proposed for brain vessel segmentation. We select two architectures that are frequently used for medical image segmentation, i.e. U-Net and Attention U-Net, as baselines for NAS optimizations. The conventional differential evolution and the opposition-based differential evolution with novel search space are employed as search methods in NAS. Furthermore, we perform ablation studies and evaluate the effects of specific loss functions, model pruning, threshold selection and generalization performance on the proposed models. The experiments are conducted on two datasets providing 335 single-channel 8-bit gray-scale images. These datasets are a public volumetric cerebrovascular system dataset (vesseINN) and our own dataset called KUVESG. The proposed NAS approaches, namely UNAS-Net and Attention UNAS-Net architectures, yield better segmentation performance in terms of different segmentation metrics. More specifically, UNAS-Net with differential evolution reveals high dice score/sensitivity values of 79.57/81.48, respectively. Moreover, they provide shorter inference times by a factor of 9.15 than the baseline methods. |
| first_indexed | 2024-03-12T00:08:51Z |
| format | Article |
| id | doaj.art-1a94dc2207174d9b83aaabfdcd876ae4 |
| institution | Directory Open Access Journal |
| issn | 2215-0986 |
| language | English |
| last_indexed | 2024-03-12T00:08:51Z |
| publishDate | 2023-10-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Engineering Science and Technology, an International Journal |
| spelling | doaj.art-1a94dc2207174d9b83aaabfdcd876ae42023-09-16T05:30:56ZengElsevierEngineering Science and Technology, an International Journal2215-09862023-10-0146101502Differential evolution-based neural architecture search for brain vessel segmentationZeki Kuş0Berna Kiraz1Tuğçe Koçak Göksu2Musa Aydın3Esra Özkan4Atay Vural5Alper Kiraz6Burhanettin Can7Fatih Sultan Mehmet Vakif University, Department of Computer Engineering, Beyoğlu, Istanbul, 34445, TurkeyFatih Sultan Mehmet Vakif University, Department of Computer Engineering, Beyoğlu, Istanbul, 34445, Turkey; Fatih Sultan Mehmet Vakif University, Data Science Research and Application Center (VEBIM), Beyoğlu, Istanbul, 34445, Turkey; Corresponding author at: Fatih Sultan Mehmet Vakif University, Department of Computer Engineering, Beyoğlu, Istanbul, 34445, Turkey.Fatih Sultan Mehmet Vakif University, Department of Computer Engineering, Beyoğlu, Istanbul, 34445, TurkeyFatih Sultan Mehmet Vakif University, Department of Computer Engineering, Beyoğlu, Istanbul, 34445, TurkeyKoç University, Research Center for Translational Medicine (KUTTAM), Sarıyer, Istanbul, 34450, TurkeyKoç University, Research Center for Translational Medicine (KUTTAM), Sarıyer, Istanbul, 34450, Turkey; Koç University, School of Medicine Department of Neurology, Sarıyer, Istanbul, 34450, TurkeyKoç University, Research Center for Translational Medicine (KUTTAM), Sarıyer, Istanbul, 34450, Turkey; Koç University, Department of Physics, Sarıyer, Istanbul, 34450, Turkey; Koç University, Department of Electrical and Electronics Engineering, Sarıyer, Istanbul, 34450, TurkeyFatih Sultan Mehmet Vakif University, Department of Computer Engineering, Beyoğlu, Istanbul, 34445, Turkey; Fatih Sultan Mehmet Vakif University, Data Science Research and Application Center (VEBIM), Beyoğlu, Istanbul, 34445, TurkeyBrain vasculature analysis is critical in developing novel treatment targets for neurodegenerative diseases. Such an accurate analysis cannot be performed manually but requires a semi-automated or fully-automated approach. Deep learning methods have recently proven indispensable for the automated segmentation and analysis of medical images. However, optimizing a deep learning network architecture is another challenge. Manually selecting deep learning network architectures and tuning their hyper-parameters requires a lot of expertise and effort. To solve this problem, neural architecture search (NAS) approaches that explore more efficient network architectures with high segmentation performance have been proposed in the literature. This study introduces differential evolution-based NAS approaches in which a novel search space is proposed for brain vessel segmentation. We select two architectures that are frequently used for medical image segmentation, i.e. U-Net and Attention U-Net, as baselines for NAS optimizations. The conventional differential evolution and the opposition-based differential evolution with novel search space are employed as search methods in NAS. Furthermore, we perform ablation studies and evaluate the effects of specific loss functions, model pruning, threshold selection and generalization performance on the proposed models. The experiments are conducted on two datasets providing 335 single-channel 8-bit gray-scale images. These datasets are a public volumetric cerebrovascular system dataset (vesseINN) and our own dataset called KUVESG. The proposed NAS approaches, namely UNAS-Net and Attention UNAS-Net architectures, yield better segmentation performance in terms of different segmentation metrics. More specifically, UNAS-Net with differential evolution reveals high dice score/sensitivity values of 79.57/81.48, respectively. Moreover, they provide shorter inference times by a factor of 9.15 than the baseline methods.http://www.sciencedirect.com/science/article/pii/S2215098623001805Attention U-NetBrain vessel segmentationDifferential evolutionNeural architecture searchU-Net |
| spellingShingle | Zeki Kuş Berna Kiraz Tuğçe Koçak Göksu Musa Aydın Esra Özkan Atay Vural Alper Kiraz Burhanettin Can Differential evolution-based neural architecture search for brain vessel segmentation Engineering Science and Technology, an International Journal Attention U-Net Brain vessel segmentation Differential evolution Neural architecture search U-Net |
| title | Differential evolution-based neural architecture search for brain vessel segmentation |
| title_full | Differential evolution-based neural architecture search for brain vessel segmentation |
| title_fullStr | Differential evolution-based neural architecture search for brain vessel segmentation |
| title_full_unstemmed | Differential evolution-based neural architecture search for brain vessel segmentation |
| title_short | Differential evolution-based neural architecture search for brain vessel segmentation |
| title_sort | differential evolution based neural architecture search for brain vessel segmentation |
| topic | Attention U-Net Brain vessel segmentation Differential evolution Neural architecture search U-Net |
| url | http://www.sciencedirect.com/science/article/pii/S2215098623001805 |
| work_keys_str_mv | AT zekikus differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation AT bernakiraz differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation AT tugcekocakgoksu differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation AT musaaydın differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation AT esraozkan differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation AT atayvural differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation AT alperkiraz differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation AT burhanettincan differentialevolutionbasedneuralarchitecturesearchforbrainvesselsegmentation |