No Cover Image

E-Thesis 20168 views 122 downloads

3D Medical Image Lossless Compressor Using Deep Learning Approaches / Omniah Nagoor

Swansea University Author: Omniah Nagoor

  • Nagoor_Omniah_PhD_Thesis_Final_Redacted_Signature.pdf

    PDF | E-Thesis – open access

    Copyright: The author, Omniah H. J. Nagoor, 2022.

    Download (78.75MB)

DOI (Published version): 10.23889/SUthesis.61753

Abstract

The ever-increasing importance of accelerated information processing, communica-tion, and storing are major requirements within the big-data era revolution. With the extensive rise in data availability, handy information acquisition, and growing data rate, a critical challenge emerges in efficient han...

Full description

Published: Swansea 2022
Institution: Swansea University
Degree level: Doctoral
Degree name: Ph.D
Supervisor: Jones, Mark W.
URI: https://cronfa.swan.ac.uk/Record/cronfa61753
Tags: Add Tag
my 1
phd 1
thesis 1
first_indexed 2022-11-01T15:01:35Z
last_indexed 2023-01-13T19:22:43Z
id cronfa61753
recordtype RisThesis
fullrecord <?xml version="1.0"?><rfc1807><datestamp>2022-11-01T15:58:11.1958474</datestamp><bib-version>v2</bib-version><id>61753</id><entry>2022-11-01</entry><title>3D Medical Image Lossless Compressor Using Deep Learning Approaches</title><swanseaauthors><author><sid>ad8a0ed9b747350e0d626fe4398a9fe0</sid><firstname>Omniah</firstname><surname>Nagoor</surname><name>Omniah Nagoor</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2022-11-01</date><deptcode>SCS</deptcode><abstract>The ever-increasing importance of accelerated information processing, communica-tion, and storing are major requirements within the big-data era revolution. With the extensive rise in data availability, handy information acquisition, and growing data rate, a critical challenge emerges in e&#xFB03;cient handling. Even with advanced technical hardware developments and multiple Graphics Processing Units (GPUs) availability, this demand is still highly promoted to utilise these technologies e&#xFB00;ectively. Health-care systems are one of the domains yielding explosive data growth. Especially when considering their modern scanners abilities, which annually produce higher-resolution and more densely sampled medical images, with increasing requirements for massive storage capacity. The bottleneck in data transmission and storage would essentially be handled with an e&#xFB00;ective compression method. Since medical information is critical and imposes an in&#xFB02;uential role in diagnosis accuracy, it is strongly encouraged to guarantee exact reconstruction with no loss in quality, which is the main objective of any lossless compression algorithm. Given the revolutionary impact of Deep Learning (DL) methods in solving many tasks while achieving the state of the art results, includ-ing data compression, this opens tremendous opportunities for contributions. While considerable e&#xFB00;orts have been made to address lossy performance using learning-based approaches, less attention was paid to address lossless compression. This PhD thesis investigates and proposes novel learning-based approaches for compressing 3D medical images losslessly.Firstly, we formulate the lossless compression task as a supervised sequential prediction problem, whereby a model learns a projection function to predict a target voxel given sequence of samples from its spatially surrounding voxels. Using such 3D local sampling information e&#xFB03;ciently exploits spatial similarities and redundancies in a volumetric medical context by utilising such a prediction paradigm. The proposed NN-based data predictor is trained to minimise the di&#xFB00;erences with the original data values while the residual errors are encoded using arithmetic coding to allow lossless reconstruction.Following this, we explore the e&#xFB00;ectiveness of Recurrent Neural Networks (RNNs) as a 3D predictor for learning the mapping function from the spatial medical domain (16 bit-depths). We analyse Long Short-Term Memory (LSTM) models&#x2019; generalisabil-ity and robustness in capturing the 3D spatial dependencies of a voxel&#x2019;s neighbourhood while utilising samples taken from various scanning settings. We evaluate our proposed MedZip models in compressing unseen Computerized Tomography (CT) and Magnetic Resonance Imaging (MRI) modalities losslessly, compared to other state-of-the-art lossless compression standards.This work investigates input con&#xFB01;gurations and sampling schemes for a many-to-one sequence prediction model, speci&#xFB01;cally for compressing 3D medical images (16 bit-depths) losslessly. The main objective is to determine the optimal practice for enabling the proposed LSTM model to achieve a high compression ratio and fast encoding-decoding performance. A solution for a non-deterministic environments problem was also proposed, allowing models to run in parallel form without much compression performance drop. Compared to well-known lossless codecs, experimental evaluations were carried out on datasets acquired by di&#xFB00;erent hospitals, representing di&#xFB00;erent body segments, and have distinct scanning modalities (i.e. CT and MRI).To conclude, we present a novel data-driven sampling scheme utilising weighted gradient scores for training LSTM prediction-based models. The objective is to determine whether some training samples are signi&#xFB01;cantly more informative than others, speci&#xFB01;cally in medical domains where samples are available on a scale of billions. The e&#xFB00;ectiveness of models trained on the presented importance sampling scheme was evaluated compared to alternative strategies such as uniform, Gaussian, and sliced-based sampling.</abstract><type>E-Thesis</type><journal/><volume/><journalNumber/><paginationStart/><paginationEnd/><publisher/><placeOfPublication>Swansea</placeOfPublication><isbnPrint/><isbnElectronic/><issnPrint/><issnElectronic/><keywords>Lossless Compression, Deep learning, 3D Predictor, Medical Image Compression, Sequence prediction model, LSTM, Volumetric Data Compression, Neural Network, CT, MRI</keywords><publishedDay>28</publishedDay><publishedMonth>10</publishedMonth><publishedYear>2022</publishedYear><publishedDate>2022-10-28</publishedDate><doi>10.23889/SUthesis.61753</doi><url/><notes>ORCiD identifier: https://orcid.org/0000-0002-6329-1293</notes><college>COLLEGE NANME</college><department>Computer Science</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>SCS</DepartmentCode><institution>Swansea University</institution><supervisor>Jones, Mark W.</supervisor><degreelevel>Doctoral</degreelevel><degreename>Ph.D</degreename><apcterm/><funders/><projectreference/><lastEdited>2022-11-01T15:58:11.1958474</lastEdited><Created>2022-11-01T14:57:06.4669608</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></authors><documents><document><filename>61753__25633__075e79f24f014a93a92c6561a5a051b7.pdf</filename><originalFilename>Nagoor_Omniah_PhD_Thesis_Final_Redacted_Signature.pdf</originalFilename><uploaded>2022-11-01T15:20:33.7049863</uploaded><type>Output</type><contentLength>82578163</contentLength><contentType>application/pdf</contentType><version>E-Thesis &#x2013; open access</version><cronfaStatus>true</cronfaStatus><documentNotes>Copyright: The author, Omniah H. J. Nagoor, 2022.</documentNotes><copyrightCorrect>true</copyrightCorrect><language>eng</language></document></documents><OutputDurs/></rfc1807>
spelling 2022-11-01T15:58:11.1958474 v2 61753 2022-11-01 3D Medical Image Lossless Compressor Using Deep Learning Approaches ad8a0ed9b747350e0d626fe4398a9fe0 Omniah Nagoor Omniah Nagoor true false 2022-11-01 SCS The ever-increasing importance of accelerated information processing, communica-tion, and storing are major requirements within the big-data era revolution. With the extensive rise in data availability, handy information acquisition, and growing data rate, a critical challenge emerges in efficient handling. Even with advanced technical hardware developments and multiple Graphics Processing Units (GPUs) availability, this demand is still highly promoted to utilise these technologies effectively. Health-care systems are one of the domains yielding explosive data growth. Especially when considering their modern scanners abilities, which annually produce higher-resolution and more densely sampled medical images, with increasing requirements for massive storage capacity. The bottleneck in data transmission and storage would essentially be handled with an effective compression method. Since medical information is critical and imposes an influential role in diagnosis accuracy, it is strongly encouraged to guarantee exact reconstruction with no loss in quality, which is the main objective of any lossless compression algorithm. Given the revolutionary impact of Deep Learning (DL) methods in solving many tasks while achieving the state of the art results, includ-ing data compression, this opens tremendous opportunities for contributions. While considerable efforts have been made to address lossy performance using learning-based approaches, less attention was paid to address lossless compression. This PhD thesis investigates and proposes novel learning-based approaches for compressing 3D medical images losslessly.Firstly, we formulate the lossless compression task as a supervised sequential prediction problem, whereby a model learns a projection function to predict a target voxel given sequence of samples from its spatially surrounding voxels. Using such 3D local sampling information efficiently exploits spatial similarities and redundancies in a volumetric medical context by utilising such a prediction paradigm. The proposed NN-based data predictor is trained to minimise the differences with the original data values while the residual errors are encoded using arithmetic coding to allow lossless reconstruction.Following this, we explore the effectiveness of Recurrent Neural Networks (RNNs) as a 3D predictor for learning the mapping function from the spatial medical domain (16 bit-depths). We analyse Long Short-Term Memory (LSTM) models’ generalisabil-ity and robustness in capturing the 3D spatial dependencies of a voxel’s neighbourhood while utilising samples taken from various scanning settings. We evaluate our proposed MedZip models in compressing unseen Computerized Tomography (CT) and Magnetic Resonance Imaging (MRI) modalities losslessly, compared to other state-of-the-art lossless compression standards.This work investigates input configurations and sampling schemes for a many-to-one sequence prediction model, specifically for compressing 3D medical images (16 bit-depths) losslessly. The main objective is to determine the optimal practice for enabling the proposed LSTM model to achieve a high compression ratio and fast encoding-decoding performance. A solution for a non-deterministic environments problem was also proposed, allowing models to run in parallel form without much compression performance drop. Compared to well-known lossless codecs, experimental evaluations were carried out on datasets acquired by different hospitals, representing different body segments, and have distinct scanning modalities (i.e. CT and MRI).To conclude, we present a novel data-driven sampling scheme utilising weighted gradient scores for training LSTM prediction-based models. The objective is to determine whether some training samples are significantly more informative than others, specifically in medical domains where samples are available on a scale of billions. The effectiveness of models trained on the presented importance sampling scheme was evaluated compared to alternative strategies such as uniform, Gaussian, and sliced-based sampling. E-Thesis Swansea Lossless Compression, Deep learning, 3D Predictor, Medical Image Compression, Sequence prediction model, LSTM, Volumetric Data Compression, Neural Network, CT, MRI 28 10 2022 2022-10-28 10.23889/SUthesis.61753 ORCiD identifier: https://orcid.org/0000-0002-6329-1293 COLLEGE NANME Computer Science COLLEGE CODE SCS Swansea University Jones, Mark W. Doctoral Ph.D 2022-11-01T15:58:11.1958474 2022-11-01T14:57:06.4669608 Faculty of Science and Engineering School of Mathematics and Computer Science - Computer Science Omniah Nagoor 1 61753__25633__075e79f24f014a93a92c6561a5a051b7.pdf Nagoor_Omniah_PhD_Thesis_Final_Redacted_Signature.pdf 2022-11-01T15:20:33.7049863 Output 82578163 application/pdf E-Thesis – open access true Copyright: The author, Omniah H. J. Nagoor, 2022. true eng
title 3D Medical Image Lossless Compressor Using Deep Learning Approaches
spellingShingle 3D Medical Image Lossless Compressor Using Deep Learning Approaches
Omniah Nagoor
title_short 3D Medical Image Lossless Compressor Using Deep Learning Approaches
title_full 3D Medical Image Lossless Compressor Using Deep Learning Approaches
title_fullStr 3D Medical Image Lossless Compressor Using Deep Learning Approaches
title_full_unstemmed 3D Medical Image Lossless Compressor Using Deep Learning Approaches
title_sort 3D Medical Image Lossless Compressor Using Deep Learning Approaches
author_id_str_mv ad8a0ed9b747350e0d626fe4398a9fe0
author_id_fullname_str_mv ad8a0ed9b747350e0d626fe4398a9fe0_***_Omniah Nagoor
author Omniah Nagoor
author2 Omniah Nagoor
format E-Thesis
publishDate 2022
institution Swansea University
doi_str_mv 10.23889/SUthesis.61753
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 The ever-increasing importance of accelerated information processing, communica-tion, and storing are major requirements within the big-data era revolution. With the extensive rise in data availability, handy information acquisition, and growing data rate, a critical challenge emerges in efficient handling. Even with advanced technical hardware developments and multiple Graphics Processing Units (GPUs) availability, this demand is still highly promoted to utilise these technologies effectively. Health-care systems are one of the domains yielding explosive data growth. Especially when considering their modern scanners abilities, which annually produce higher-resolution and more densely sampled medical images, with increasing requirements for massive storage capacity. The bottleneck in data transmission and storage would essentially be handled with an effective compression method. Since medical information is critical and imposes an influential role in diagnosis accuracy, it is strongly encouraged to guarantee exact reconstruction with no loss in quality, which is the main objective of any lossless compression algorithm. Given the revolutionary impact of Deep Learning (DL) methods in solving many tasks while achieving the state of the art results, includ-ing data compression, this opens tremendous opportunities for contributions. While considerable efforts have been made to address lossy performance using learning-based approaches, less attention was paid to address lossless compression. This PhD thesis investigates and proposes novel learning-based approaches for compressing 3D medical images losslessly.Firstly, we formulate the lossless compression task as a supervised sequential prediction problem, whereby a model learns a projection function to predict a target voxel given sequence of samples from its spatially surrounding voxels. Using such 3D local sampling information efficiently exploits spatial similarities and redundancies in a volumetric medical context by utilising such a prediction paradigm. The proposed NN-based data predictor is trained to minimise the differences with the original data values while the residual errors are encoded using arithmetic coding to allow lossless reconstruction.Following this, we explore the effectiveness of Recurrent Neural Networks (RNNs) as a 3D predictor for learning the mapping function from the spatial medical domain (16 bit-depths). We analyse Long Short-Term Memory (LSTM) models’ generalisabil-ity and robustness in capturing the 3D spatial dependencies of a voxel’s neighbourhood while utilising samples taken from various scanning settings. We evaluate our proposed MedZip models in compressing unseen Computerized Tomography (CT) and Magnetic Resonance Imaging (MRI) modalities losslessly, compared to other state-of-the-art lossless compression standards.This work investigates input configurations and sampling schemes for a many-to-one sequence prediction model, specifically for compressing 3D medical images (16 bit-depths) losslessly. The main objective is to determine the optimal practice for enabling the proposed LSTM model to achieve a high compression ratio and fast encoding-decoding performance. A solution for a non-deterministic environments problem was also proposed, allowing models to run in parallel form without much compression performance drop. Compared to well-known lossless codecs, experimental evaluations were carried out on datasets acquired by different hospitals, representing different body segments, and have distinct scanning modalities (i.e. CT and MRI).To conclude, we present a novel data-driven sampling scheme utilising weighted gradient scores for training LSTM prediction-based models. The objective is to determine whether some training samples are significantly more informative than others, specifically in medical domains where samples are available on a scale of billions. The effectiveness of models trained on the presented importance sampling scheme was evaluated compared to alternative strategies such as uniform, Gaussian, and sliced-based sampling.
published_date 2022-10-28T04:20:49Z
_version_ 1763754376321564672
score 10.9983425