Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network
Accurate segmentation of 3D micro CT scans is a key step in the process of analysis of the microstructure of porous materials. In polar ice core studies, the environmental effects on the firn column could be detected if the microstructure is digitized accurately. The most challenging task is to obta...
Published in: | Applied Computing and Geosciences |
---|---|
Main Authors: | , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Elsevier
2024
|
Subjects: | |
Online Access: | https://doi.org/10.1016/j.acags.2024.100193 https://doaj.org/article/e03c0cd5691f4696b9da9391e248f510 |
id |
ftdoajarticles:oai:doaj.org/article:e03c0cd5691f4696b9da9391e248f510 |
---|---|
record_format |
openpolar |
spelling |
ftdoajarticles:oai:doaj.org/article:e03c0cd5691f4696b9da9391e248f510 2024-09-30T14:36:30+00:00 Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network Faramarz Bagherzadeh Johannes Freitag Udo Frese Frank Wilhelms 2024-09-01T00:00:00Z https://doi.org/10.1016/j.acags.2024.100193 https://doaj.org/article/e03c0cd5691f4696b9da9391e248f510 EN eng Elsevier http://www.sciencedirect.com/science/article/pii/S2590197424000405 https://doaj.org/toc/2590-1974 2590-1974 doi:10.1016/j.acags.2024.100193 https://doaj.org/article/e03c0cd5691f4696b9da9391e248f510 Applied Computing and Geosciences, Vol 23, Iss , Pp 100193- (2024) Micro CT 3D image segmentation Deep learning 3D unet FCN Ice core Geography. Anthropology. Recreation G Geology QE1-996.5 Electronic computers. Computer science QA75.5-76.95 article 2024 ftdoajarticles https://doi.org/10.1016/j.acags.2024.100193 2024-09-17T16:00:44Z Accurate segmentation of 3D micro CT scans is a key step in the process of analysis of the microstructure of porous materials. In polar ice core studies, the environmental effects on the firn column could be detected if the microstructure is digitized accurately. The most challenging task is to obtain the microstructure parameters of the bubbly ice section. To identify the minimum, necessary resolution, the bubbly ice micro CT scans with different resolutions (120, 60, 30, 12 μm) were compared object-wise via a region pairing algorithm. When the minimum resolution was found to be 60 μm, for generating the training/validation dataset, 4 ice core samples from a depth range of 96 to 108 meters (bubbly ice) were scanned with 120 μm (input images) and another time with 4 times higher resolution (30 μm) to build ground truth. A specific pipeline was designed with non-rigid image registration to create an accurate ground truth from 4 times higher resolution scans. Then, two SOTA deep learning models (3D-Unet and FCN) were trained and later validated to perform super-resolution segmentation by taking input of 120μm resolution data and giving the output of binary segmented with two times higher resolution (60μm). Finally, the outputs of CNN models were compared with traditional rule-based and unsupervised methods on blind test data. It is observed the 3D-Unet can segment low-resolution scans with an accuracy of 96% and an f1-score of 80.8% while preserving microstructure having less than 2% error in porosity and SSA. Article in Journal/Newspaper ice core Directory of Open Access Journals: DOAJ Articles Applied Computing and Geosciences 23 100193 |
institution |
Open Polar |
collection |
Directory of Open Access Journals: DOAJ Articles |
op_collection_id |
ftdoajarticles |
language |
English |
topic |
Micro CT 3D image segmentation Deep learning 3D unet FCN Ice core Geography. Anthropology. Recreation G Geology QE1-996.5 Electronic computers. Computer science QA75.5-76.95 |
spellingShingle |
Micro CT 3D image segmentation Deep learning 3D unet FCN Ice core Geography. Anthropology. Recreation G Geology QE1-996.5 Electronic computers. Computer science QA75.5-76.95 Faramarz Bagherzadeh Johannes Freitag Udo Frese Frank Wilhelms Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network |
topic_facet |
Micro CT 3D image segmentation Deep learning 3D unet FCN Ice core Geography. Anthropology. Recreation G Geology QE1-996.5 Electronic computers. Computer science QA75.5-76.95 |
description |
Accurate segmentation of 3D micro CT scans is a key step in the process of analysis of the microstructure of porous materials. In polar ice core studies, the environmental effects on the firn column could be detected if the microstructure is digitized accurately. The most challenging task is to obtain the microstructure parameters of the bubbly ice section. To identify the minimum, necessary resolution, the bubbly ice micro CT scans with different resolutions (120, 60, 30, 12 μm) were compared object-wise via a region pairing algorithm. When the minimum resolution was found to be 60 μm, for generating the training/validation dataset, 4 ice core samples from a depth range of 96 to 108 meters (bubbly ice) were scanned with 120 μm (input images) and another time with 4 times higher resolution (30 μm) to build ground truth. A specific pipeline was designed with non-rigid image registration to create an accurate ground truth from 4 times higher resolution scans. Then, two SOTA deep learning models (3D-Unet and FCN) were trained and later validated to perform super-resolution segmentation by taking input of 120μm resolution data and giving the output of binary segmented with two times higher resolution (60μm). Finally, the outputs of CNN models were compared with traditional rule-based and unsupervised methods on blind test data. It is observed the 3D-Unet can segment low-resolution scans with an accuracy of 96% and an f1-score of 80.8% while preserving microstructure having less than 2% error in porosity and SSA. |
format |
Article in Journal/Newspaper |
author |
Faramarz Bagherzadeh Johannes Freitag Udo Frese Frank Wilhelms |
author_facet |
Faramarz Bagherzadeh Johannes Freitag Udo Frese Frank Wilhelms |
author_sort |
Faramarz Bagherzadeh |
title |
Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network |
title_short |
Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network |
title_full |
Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network |
title_fullStr |
Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network |
title_full_unstemmed |
Resolution enhancement and segmentation of polar bubbly ice micro CT scans via 3D convolutional neural network |
title_sort |
resolution enhancement and segmentation of polar bubbly ice micro ct scans via 3d convolutional neural network |
publisher |
Elsevier |
publishDate |
2024 |
url |
https://doi.org/10.1016/j.acags.2024.100193 https://doaj.org/article/e03c0cd5691f4696b9da9391e248f510 |
genre |
ice core |
genre_facet |
ice core |
op_source |
Applied Computing and Geosciences, Vol 23, Iss , Pp 100193- (2024) |
op_relation |
http://www.sciencedirect.com/science/article/pii/S2590197424000405 https://doaj.org/toc/2590-1974 2590-1974 doi:10.1016/j.acags.2024.100193 https://doaj.org/article/e03c0cd5691f4696b9da9391e248f510 |
op_doi |
https://doi.org/10.1016/j.acags.2024.100193 |
container_title |
Applied Computing and Geosciences |
container_volume |
23 |
container_start_page |
100193 |
_version_ |
1811639545804881920 |