No Cover Image

Journal article 802 views 138 downloads

Sampling strategies for learning-based 3D medical image compression

Omniah Nagoor, Joss Whittle Orcid Logo, Jingjing Deng, Benjamin Mora Orcid Logo, Mark Jones Orcid Logo

Machine Learning with Applications, Volume: 8, Start page: 100273

Swansea University Authors: Omniah Nagoor, Jingjing Deng, Benjamin Mora Orcid Logo, Mark Jones Orcid Logo

  • 1-s2.0-S2666827022000135-main.pdf

    PDF | Version of Record

    © 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license

    Download (4.31MB)

Check full text

DOI (Published version): 10.1016/j.mlwa.2022.100273

Abstract

Recent achievements of sequence prediction models in numerous domains, including compression, provide great potential for novel learning-based codecs. In such models, the input sequence’s shape and size play a crucial role in learning the mapping function of the data distribution to the target outpu...

Full description

Published in: Machine Learning with Applications
ISSN: 2666-8270
Published: Elsevier BV 2022
Online Access: Check full text

URI: https://cronfa.swan.ac.uk/Record/cronfa59383
Tags: Add Tag
No Tags, Be the first to tag this record!
first_indexed 2022-02-12T09:59:01Z
last_indexed 2023-01-12T12:55:28Z
id cronfa59383
recordtype SURis
fullrecord <?xml version="1.0"?><rfc1807><datestamp>2023-01-11T14:59:56.1051337</datestamp><bib-version>v2</bib-version><id>59383</id><entry>2022-02-12</entry><title>Sampling strategies for learning-based 3D medical image compression</title><swanseaauthors><author><sid>ad8a0ed9b747350e0d626fe4398a9fe0</sid><firstname>Omniah</firstname><surname>Nagoor</surname><name>Omniah Nagoor</name><active>true</active><ethesisStudent>false</ethesisStudent></author><author><sid>6f6d01d585363d6dc1622640bb4fcb3f</sid><firstname>Jingjing</firstname><surname>Deng</surname><name>Jingjing Deng</name><active>true</active><ethesisStudent>false</ethesisStudent></author><author><sid>557f93dfae240600e5bd4398bf203821</sid><ORCID>0000-0002-2945-3519</ORCID><firstname>Benjamin</firstname><surname>Mora</surname><name>Benjamin Mora</name><active>true</active><ethesisStudent>false</ethesisStudent></author><author><sid>2e1030b6e14fc9debd5d5ae7cc335562</sid><ORCID>0000-0001-8991-1190</ORCID><firstname>Mark</firstname><surname>Jones</surname><name>Mark Jones</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2022-02-12</date><deptcode>SCS</deptcode><abstract>Recent achievements of sequence prediction models in numerous domains, including compression, provide great potential for novel learning-based codecs. In such models, the input sequence&#x2019;s shape and size play a crucial role in learning the mapping function of the data distribution to the target output. This work examines numerous input configurations and sampling schemes for a many-to-one sequence prediction model, specifically for compressing 3D medical images (16-bit depth) losslessly. The main objective is to determine the optimal practice for enabling the proposed Long Short-Term Memory (LSTM) model to achieve high compression ratio and fast encoding-decoding performance.Our LSTM models are trained with 4-fold cross-validation on 12 high-resolution CT dataset while measuring model&#x2019;s compression ratios and execution time. Several configurations of sequences have been evaluated, and our results demonstrate that pyramid-shaped sampling represents the best trade-off between performance and compression ratio (up to&#x2009;3x). We solve a problem of non-deterministic environments that allow our models to run in parallel without much compression performance drop.Experimental evaluation was carried out on datasets acquired by different hospitals, representing different body segments, and distinct scanning modalities (CT and MRI). Our new methodology allows straightforward parallelisation that speeds-up the decoder by up to&#x2009;37x compared to previous methods. Overall, the trained models demonstrate efficiency and generalisability for compressing 3D medical images losslessly while still outperforming well-known lossless methods by approximately 17% and 12%. To the best of our knowledge, this is the first study that focuses on voxel-wise predictions of volumetric medical imaging for lossless compression.</abstract><type>Journal Article</type><journal>Machine Learning with Applications</journal><volume>8</volume><journalNumber/><paginationStart>100273</paginationStart><paginationEnd/><publisher>Elsevier BV</publisher><placeOfPublication/><isbnPrint/><isbnElectronic/><issnPrint>2666-8270</issnPrint><issnElectronic/><keywords>3D predictors; Deep learning; Lossless compression; Medical image compression; Sequence prediction model; LSTM</keywords><publishedDay>15</publishedDay><publishedMonth>6</publishedMonth><publishedYear>2022</publishedYear><publishedDate>2022-06-15</publishedDate><doi>10.1016/j.mlwa.2022.100273</doi><url/><notes/><college>COLLEGE NANME</college><department>Computer Science</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>SCS</DepartmentCode><institution>Swansea University</institution><apcterm>SU College/Department paid the OA fee</apcterm><funders/><projectreference/><lastEdited>2023-01-11T14:59:56.1051337</lastEdited><Created>2022-02-12T09:55:36.4264124</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Mathematics and Computer Science - Computer Science</level></path><authors><author><firstname>Omniah</firstname><surname>Nagoor</surname><order>1</order></author><author><firstname>Joss</firstname><surname>Whittle</surname><orcid>0000-0002-4147-7185</orcid><order>2</order></author><author><firstname>Jingjing</firstname><surname>Deng</surname><order>3</order></author><author><firstname>Benjamin</firstname><surname>Mora</surname><orcid>0000-0002-2945-3519</orcid><order>4</order></author><author><firstname>Mark</firstname><surname>Jones</surname><orcid>0000-0001-8991-1190</orcid><order>5</order></author></authors><documents><document><filename>59383__22463__985603d06e234bacaeff67910dcf561a.pdf</filename><originalFilename>1-s2.0-S2666827022000135-main.pdf</originalFilename><uploaded>2022-02-26T12:48:21.8501875</uploaded><type>Output</type><contentLength>4515581</contentLength><contentType>application/pdf</contentType><version>Version of Record</version><cronfaStatus>true</cronfaStatus><documentNotes>&#xA9; 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license</documentNotes><copyrightCorrect>true</copyrightCorrect><language>eng</language><licence>https://creativecommons.org/licenses/by/4.0/</licence></document></documents><OutputDurs/></rfc1807>
spelling 2023-01-11T14:59:56.1051337 v2 59383 2022-02-12 Sampling strategies for learning-based 3D medical image compression ad8a0ed9b747350e0d626fe4398a9fe0 Omniah Nagoor Omniah Nagoor true false 6f6d01d585363d6dc1622640bb4fcb3f Jingjing Deng Jingjing Deng true false 557f93dfae240600e5bd4398bf203821 0000-0002-2945-3519 Benjamin Mora Benjamin Mora true false 2e1030b6e14fc9debd5d5ae7cc335562 0000-0001-8991-1190 Mark Jones Mark Jones true false 2022-02-12 SCS Recent achievements of sequence prediction models in numerous domains, including compression, provide great potential for novel learning-based codecs. In such models, the input sequence’s shape and size play a crucial role in learning the mapping function of the data distribution to the target output. This work examines numerous input configurations and sampling schemes for a many-to-one sequence prediction model, specifically for compressing 3D medical images (16-bit depth) losslessly. The main objective is to determine the optimal practice for enabling the proposed Long Short-Term Memory (LSTM) model to achieve high compression ratio and fast encoding-decoding performance.Our LSTM models are trained with 4-fold cross-validation on 12 high-resolution CT dataset while measuring model’s compression ratios and execution time. Several configurations of sequences have been evaluated, and our results demonstrate that pyramid-shaped sampling represents the best trade-off between performance and compression ratio (up to 3x). We solve a problem of non-deterministic environments that allow our models to run in parallel without much compression performance drop.Experimental evaluation was carried out on datasets acquired by different hospitals, representing different body segments, and distinct scanning modalities (CT and MRI). Our new methodology allows straightforward parallelisation that speeds-up the decoder by up to 37x compared to previous methods. Overall, the trained models demonstrate efficiency and generalisability for compressing 3D medical images losslessly while still outperforming well-known lossless methods by approximately 17% and 12%. To the best of our knowledge, this is the first study that focuses on voxel-wise predictions of volumetric medical imaging for lossless compression. Journal Article Machine Learning with Applications 8 100273 Elsevier BV 2666-8270 3D predictors; Deep learning; Lossless compression; Medical image compression; Sequence prediction model; LSTM 15 6 2022 2022-06-15 10.1016/j.mlwa.2022.100273 COLLEGE NANME Computer Science COLLEGE CODE SCS Swansea University SU College/Department paid the OA fee 2023-01-11T14:59:56.1051337 2022-02-12T09:55:36.4264124 Faculty of Science and Engineering School of Mathematics and Computer Science - Computer Science Omniah Nagoor 1 Joss Whittle 0000-0002-4147-7185 2 Jingjing Deng 3 Benjamin Mora 0000-0002-2945-3519 4 Mark Jones 0000-0001-8991-1190 5 59383__22463__985603d06e234bacaeff67910dcf561a.pdf 1-s2.0-S2666827022000135-main.pdf 2022-02-26T12:48:21.8501875 Output 4515581 application/pdf Version of Record true © 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license true eng https://creativecommons.org/licenses/by/4.0/
title Sampling strategies for learning-based 3D medical image compression
spellingShingle Sampling strategies for learning-based 3D medical image compression
Omniah Nagoor
Jingjing Deng
Benjamin Mora
Mark Jones
title_short Sampling strategies for learning-based 3D medical image compression
title_full Sampling strategies for learning-based 3D medical image compression
title_fullStr Sampling strategies for learning-based 3D medical image compression
title_full_unstemmed Sampling strategies for learning-based 3D medical image compression
title_sort Sampling strategies for learning-based 3D medical image compression
author_id_str_mv ad8a0ed9b747350e0d626fe4398a9fe0
6f6d01d585363d6dc1622640bb4fcb3f
557f93dfae240600e5bd4398bf203821
2e1030b6e14fc9debd5d5ae7cc335562
author_id_fullname_str_mv ad8a0ed9b747350e0d626fe4398a9fe0_***_Omniah Nagoor
6f6d01d585363d6dc1622640bb4fcb3f_***_Jingjing Deng
557f93dfae240600e5bd4398bf203821_***_Benjamin Mora
2e1030b6e14fc9debd5d5ae7cc335562_***_Mark Jones
author Omniah Nagoor
Jingjing Deng
Benjamin Mora
Mark Jones
author2 Omniah Nagoor
Joss Whittle
Jingjing Deng
Benjamin Mora
Mark Jones
format Journal article
container_title Machine Learning with Applications
container_volume 8
container_start_page 100273
publishDate 2022
institution Swansea University
issn 2666-8270
doi_str_mv 10.1016/j.mlwa.2022.100273
publisher Elsevier BV
college_str Faculty of Science and Engineering
hierarchytype
hierarchy_top_id facultyofscienceandengineering
hierarchy_top_title Faculty of Science and Engineering
hierarchy_parent_id facultyofscienceandengineering
hierarchy_parent_title Faculty of Science and Engineering
department_str School of Mathematics and Computer Science - Computer Science{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Mathematics and Computer Science - Computer Science
document_store_str 1
active_str 0
description Recent achievements of sequence prediction models in numerous domains, including compression, provide great potential for novel learning-based codecs. In such models, the input sequence’s shape and size play a crucial role in learning the mapping function of the data distribution to the target output. This work examines numerous input configurations and sampling schemes for a many-to-one sequence prediction model, specifically for compressing 3D medical images (16-bit depth) losslessly. The main objective is to determine the optimal practice for enabling the proposed Long Short-Term Memory (LSTM) model to achieve high compression ratio and fast encoding-decoding performance.Our LSTM models are trained with 4-fold cross-validation on 12 high-resolution CT dataset while measuring model’s compression ratios and execution time. Several configurations of sequences have been evaluated, and our results demonstrate that pyramid-shaped sampling represents the best trade-off between performance and compression ratio (up to 3x). We solve a problem of non-deterministic environments that allow our models to run in parallel without much compression performance drop.Experimental evaluation was carried out on datasets acquired by different hospitals, representing different body segments, and distinct scanning modalities (CT and MRI). Our new methodology allows straightforward parallelisation that speeds-up the decoder by up to 37x compared to previous methods. Overall, the trained models demonstrate efficiency and generalisability for compressing 3D medical images losslessly while still outperforming well-known lossless methods by approximately 17% and 12%. To the best of our knowledge, this is the first study that focuses on voxel-wise predictions of volumetric medical imaging for lossless compression.
published_date 2022-06-15T04:16:39Z
_version_ 1763754114198536192
score 10.999161

Similar Items