Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor Images
Among researchers using traditional and new machine learning and deep learning techniques, 2D medical image segmentation models are popular. Additionally, 3D volumetric data recently became more accessible, as a result of the high number of studies conducted in recent years regarding the creation of...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-08-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/17/6501 |
| _version_ | 1797493195601346560 |
|---|---|
| author | Jakhongir Nodirov Akmalbek Bobomirzaevich Abdusalomov Taeg Keun Whangbo |
| author_facet | Jakhongir Nodirov Akmalbek Bobomirzaevich Abdusalomov Taeg Keun Whangbo |
| author_sort | Jakhongir Nodirov |
| collection | DOAJ |
| description | Among researchers using traditional and new machine learning and deep learning techniques, 2D medical image segmentation models are popular. Additionally, 3D volumetric data recently became more accessible, as a result of the high number of studies conducted in recent years regarding the creation of 3D volumes. Using these 3D data, researchers have begun conducting research on creating 3D segmentation models, such as brain tumor segmentation and classification. Since a higher number of crucial features can be extracted using 3D data than 2D data, 3D brain tumor detection models have increased in popularity among researchers. Until now, various significant research works have focused on the 3D version of the U-Net and other popular models, such as 3D U-Net and V-Net, while doing superior research works. In this study, we used 3D brain image data and created a new architecture based on a 3D U-Net model that uses multiple skip connections with cost-efficient pretrained 3D MobileNetV2 blocks and attention modules. These pretrained MobileNetV2 blocks assist our architecture by providing smaller parameters to maintain operable model size in terms of our computational capability and help the model to converge faster. We added additional skip connections between the encoder and decoder blocks to ease the exchange of extracted features between the two blocks, which resulted in the maximum use of the features. We also used attention modules to filter out irrelevant features coming through the skip connections and, thus, preserved more computational power while achieving improved accuracy. |
| first_indexed | 2024-03-10T01:16:32Z |
| format | Article |
| id | doaj.art-d733c9ecaae74d7aa59be3a5d945d895 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T01:16:32Z |
| publishDate | 2022-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-d733c9ecaae74d7aa59be3a5d945d8952023-11-23T14:09:31ZengMDPI AGSensors1424-82202022-08-012217650110.3390/s22176501Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor ImagesJakhongir Nodirov0Akmalbek Bobomirzaevich Abdusalomov1Taeg Keun Whangbo2Department of IT Convergence Engineering, Gachon University, Sujeong-Gu, Seongnam-Si 461-701, Gyeonggi-Do, KoreaDepartment of Computer Engineering, Gachon University, Sujeong-Gu, Seongnam-Si 461-701, Gyeonggi-Do, KoreaDepartment of Computer Engineering, Gachon University, Sujeong-Gu, Seongnam-Si 461-701, Gyeonggi-Do, KoreaAmong researchers using traditional and new machine learning and deep learning techniques, 2D medical image segmentation models are popular. Additionally, 3D volumetric data recently became more accessible, as a result of the high number of studies conducted in recent years regarding the creation of 3D volumes. Using these 3D data, researchers have begun conducting research on creating 3D segmentation models, such as brain tumor segmentation and classification. Since a higher number of crucial features can be extracted using 3D data than 2D data, 3D brain tumor detection models have increased in popularity among researchers. Until now, various significant research works have focused on the 3D version of the U-Net and other popular models, such as 3D U-Net and V-Net, while doing superior research works. In this study, we used 3D brain image data and created a new architecture based on a 3D U-Net model that uses multiple skip connections with cost-efficient pretrained 3D MobileNetV2 blocks and attention modules. These pretrained MobileNetV2 blocks assist our architecture by providing smaller parameters to maintain operable model size in terms of our computational capability and help the model to converge faster. We added additional skip connections between the encoder and decoder blocks to ease the exchange of extracted features between the two blocks, which resulted in the maximum use of the features. We also used attention modules to filter out irrelevant features coming through the skip connections and, thus, preserved more computational power while achieving improved accuracy.https://www.mdpi.com/1424-8220/22/17/6501brain imagedeep learningCNNU-Netpretrained MobileNetV2attention modules |
| spellingShingle | Jakhongir Nodirov Akmalbek Bobomirzaevich Abdusalomov Taeg Keun Whangbo Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor Images Sensors brain image deep learning CNN U-Net pretrained MobileNetV2 attention modules |
| title | Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor Images |
| title_full | Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor Images |
| title_fullStr | Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor Images |
| title_full_unstemmed | Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor Images |
| title_short | Attention 3D U-Net with Multiple Skip Connections for Segmentation of Brain Tumor Images |
| title_sort | attention 3d u net with multiple skip connections for segmentation of brain tumor images |
| topic | brain image deep learning CNN U-Net pretrained MobileNetV2 attention modules |
| url | https://www.mdpi.com/1424-8220/22/17/6501 |
| work_keys_str_mv | AT jakhongirnodirov attention3dunetwithmultipleskipconnectionsforsegmentationofbraintumorimages AT akmalbekbobomirzaevichabdusalomov attention3dunetwithmultipleskipconnectionsforsegmentationofbraintumorimages AT taegkeunwhangbo attention3dunetwithmultipleskipconnectionsforsegmentationofbraintumorimages |