UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias
Biomedical image segmentation (BIS) task is challenging due to the variations in organ types, position, shape, size, scale, orientation, and image contrast. Conventional methods lack accurate and automated designs. Artificial intelligence (AI)-based UNet has recently dominated BIS. This is the first...
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9999432/ |
| _version_ | 1828070749528653824 |
|---|---|
| author | Jasjit S. Suri Mrinalini Bhagawati Sushant Agarwal Sudip Paul Amit Pandey Suneet K. Gupta Luca Saba Kosmas I. Paraskevas Narendra N. Khanna John R. Laird Amer M. Johri Manudeep K. Kalra Mostafa M. Fouda Mostafa Fatemi Subbaram Naidu |
| author_facet | Jasjit S. Suri Mrinalini Bhagawati Sushant Agarwal Sudip Paul Amit Pandey Suneet K. Gupta Luca Saba Kosmas I. Paraskevas Narendra N. Khanna John R. Laird Amer M. Johri Manudeep K. Kalra Mostafa M. Fouda Mostafa Fatemi Subbaram Naidu |
| author_sort | Jasjit S. Suri |
| collection | DOAJ |
| description | Biomedical image segmentation (BIS) task is challenging due to the variations in organ types, position, shape, size, scale, orientation, and image contrast. Conventional methods lack accurate and automated designs. Artificial intelligence (AI)-based UNet has recently dominated BIS. This is the first review of its kind that microscopically addressed UNet types by complexity, stratification of UNet by its components, addressing UNet in vascular vs. non-vascular framework, the key to segmentation challenge vs. UNet-based architecture, and finally interfacing the three facets of AI, the pruning, the explainable AI (XAI), and the AI-bias. PRISMA was used to select 267 UNet-based studies. Five classes were identified and labeled as conventional UNet, superior UNet, attention-channel UNet, hybrid UNet, and ensemble UNet. We discovered 81 variations of UNet by considering six kinds of components, namely encoder, decoder, skip connection, bridge network, loss function, and their combination. Vascular vs. non-vascular UNet architecture was compared. AP(ai)Bias 2.0-UNet was identified in these UNet classes based on (i) attributes of UNet architecture and its performance, (ii) explainable AI (XAI), and, (iii) pruning (compression). Five bias methods such as (i) ranking, (ii) radial, (iii) regional area, (iv) PROBAST, and (v) ROBINS-I were applied and compared using a Venn diagram. Vascular and non-vascular UNet systems dominated with sUNet classes with attention. Most of the studies suffered from a low interest in XAI and pruning strategies. None of the UNet models qualified to be bias-free. There is a need to move from paper-to-practice paradigms for clinical evaluation and settings. |
| first_indexed | 2024-04-11T00:43:50Z |
| format | Article |
| id | doaj.art-b520592ad96040899b648c951e77ddfa |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-11T00:43:50Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-b520592ad96040899b648c951e77ddfa2023-01-06T00:00:26ZengIEEEIEEE Access2169-35362023-01-011159564510.1109/ACCESS.2022.32325619999432UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and BiasJasjit S. Suri0https://orcid.org/0000-0001-6499-396XMrinalini Bhagawati1Sushant Agarwal2https://orcid.org/0000-0003-2360-3886Sudip Paul3https://orcid.org/0000-0001-9856-539XAmit Pandey4Suneet K. Gupta5https://orcid.org/0000-0002-5086-8401Luca Saba6https://orcid.org/0000-0003-3610-8526Kosmas I. Paraskevas7https://orcid.org/0000-0001-6865-2919Narendra N. Khanna8John R. Laird9https://orcid.org/0000-0003-2095-2191Amer M. Johri10https://orcid.org/0000-0001-7044-8212Manudeep K. Kalra11Mostafa M. Fouda12https://orcid.org/0000-0003-1790-8640Mostafa Fatemi13https://orcid.org/0000-0002-6603-9077Subbaram Naidu14https://orcid.org/0000-0002-6284-399XStroke Diagnostic and Monitoring Division, AtheroPoint, Roseville, CA, USADepartment of Biomedical Engineering, North-Eastern Hill University, Shillong, IndiaTechNet Cyber Solutions Pvt. Ltd., Andhra Pradesh, IndiaDepartment of Biomedical Engineering, North-Eastern Hill University, Shillong, IndiaDepartment of Computer Science, Bennett University, Greater Noida, IndiaDepartment of Computer Science, Bennett University, Greater Noida, IndiaDepartment of Radiology, University of Cagliari, Cagliari, ItalyDepartment of Vascular Surgery, Central Clinic of Athens, Athens, GreeceDepartment of Cardiology, Indraprastha Apollo Hospitals, New Delhi, IndiaCardiology Department, St. Helena Hospital, St. Helena, CA, USADepartment of Medicine, Division of Cardiology, Queen’s University, Kingston, CanadaDepartment of Radiology, Massachusetts General Hospital, Boston, MA, USADepartment of Electrical and Computer Engineering, Idaho State University, Pocatello, ID, USADepartment of Physiology and Biomedical Engineering, Mayo Clinic College of Medicine and Science, Rochester, MN, USAElectrical Engineering Department, University of Minnesota, Duluth, MN, USABiomedical image segmentation (BIS) task is challenging due to the variations in organ types, position, shape, size, scale, orientation, and image contrast. Conventional methods lack accurate and automated designs. Artificial intelligence (AI)-based UNet has recently dominated BIS. This is the first review of its kind that microscopically addressed UNet types by complexity, stratification of UNet by its components, addressing UNet in vascular vs. non-vascular framework, the key to segmentation challenge vs. UNet-based architecture, and finally interfacing the three facets of AI, the pruning, the explainable AI (XAI), and the AI-bias. PRISMA was used to select 267 UNet-based studies. Five classes were identified and labeled as conventional UNet, superior UNet, attention-channel UNet, hybrid UNet, and ensemble UNet. We discovered 81 variations of UNet by considering six kinds of components, namely encoder, decoder, skip connection, bridge network, loss function, and their combination. Vascular vs. non-vascular UNet architecture was compared. AP(ai)Bias 2.0-UNet was identified in these UNet classes based on (i) attributes of UNet architecture and its performance, (ii) explainable AI (XAI), and, (iii) pruning (compression). Five bias methods such as (i) ranking, (ii) radial, (iii) regional area, (iv) PROBAST, and (v) ROBINS-I were applied and compared using a Venn diagram. Vascular and non-vascular UNet systems dominated with sUNet classes with attention. Most of the studies suffered from a low interest in XAI and pruning strategies. None of the UNet models qualified to be bias-free. There is a need to move from paper-to-practice paradigms for clinical evaluation and settings.https://ieeexplore.ieee.org/document/9999432/Image segmentationvascularnon-vascularUNet classesUNet variationsUNet-components |
| spellingShingle | Jasjit S. Suri Mrinalini Bhagawati Sushant Agarwal Sudip Paul Amit Pandey Suneet K. Gupta Luca Saba Kosmas I. Paraskevas Narendra N. Khanna John R. Laird Amer M. Johri Manudeep K. Kalra Mostafa M. Fouda Mostafa Fatemi Subbaram Naidu UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias IEEE Access Image segmentation vascular non-vascular UNet classes UNet variations UNet-components |
| title | UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias |
| title_full | UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias |
| title_fullStr | UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias |
| title_full_unstemmed | UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias |
| title_short | UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias |
| title_sort | unet deep learning architecture for segmentation of vascular and non vascular images a microscopic look at unet components buffered with pruning explainable artificial intelligence and bias |
| topic | Image segmentation vascular non-vascular UNet classes UNet variations UNet-components |
| url | https://ieeexplore.ieee.org/document/9999432/ |
| work_keys_str_mv | AT jasjitssuri unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT mrinalinibhagawati unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT sushantagarwal unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT sudippaul unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT amitpandey unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT suneetkgupta unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT lucasaba unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT kosmasiparaskevas unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT narendrankhanna unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT johnrlaird unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT amermjohri unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT manudeepkkalra unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT mostafamfouda unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT mostafafatemi unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias AT subbaramnaidu unetdeeplearningarchitectureforsegmentationofvascularandnonvascularimagesamicroscopiclookatunetcomponentsbufferedwithpruningexplainableartificialintelligenceandbias |