Learnable PM diffusion coefficients and reformative coordinate attention network for low dose CT denoising
Various deep learning methods have recently been used for low dose CT (LDCT) denoising. Aggressive denoising may destroy the edge and fine anatomical structures of CT images. Therefore a key issue in LDCT denoising tasks is the difficulty of balancing noise/artifact suppression and edge/structure pr...
Uloženo v:
| Vydáno v: | Physics in medicine & biology Ročník 68; číslo 24 |
|---|---|
| Hlavní autoři: | , , , , , , , , |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
England
21.12.2023
|
| Témata: | |
| ISSN: | 1361-6560, 1361-6560 |
| On-line přístup: | Zjistit podrobnosti o přístupu |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Abstract | Various deep learning methods have recently been used for low dose CT (LDCT) denoising. Aggressive denoising may destroy the edge and fine anatomical structures of CT images. Therefore a key issue in LDCT denoising tasks is the difficulty of balancing noise/artifact suppression and edge/structure preservation.
We proposed an LDCT denoising network based on the encoder-decoder structure, namely the Learnable PM diffusion coefficient and efficient attention network (PMA-Net). First, using the powerful feature modeling capability of partial differential equations, we constructed a multiple learnable edge module to generate precise edge information, incorporating the anisotropic image processing idea of Perona-Malik (PM) model into the neural network. Second, a multiscale reformative coordinate attention module was designed to extract multiscale information. Non-overlapping dilated convolution capturing abundant contextual content was combined with coordinate attention which could embed the spatial location information of important features into the channel attention map. Finally, we imposed additional constraints on the edge information using edge-enhanced multiscale perceptual loss to avoid structure loss and over-smoothing.
Experiments are conducted on simulated and real datasets. The quantitative and qualitative results show that the proposed method has better performance in suppressing noise/artifacts and preserving edges/structures.
This work proposes a novel edge feature extraction method that unfolds partial differential equation into neural networks, which contributes to the interpretability and clinical application value of neural network. |
|---|---|
| AbstractList | Objective.Various deep learning methods have recently been used for low dose CT (LDCT) denoising. Aggressive denoising may destroy the edge and fine anatomical structures of CT images. Therefore a key issue in LDCT denoising tasks is the difficulty of balancing noise/artifact suppression and edge/structure preservation.Approach.We proposed an LDCT denoising network based on the encoder-decoder structure, namely the Learnable PM diffusion coefficient and efficient attention network (PMA-Net). First, using the powerful feature modeling capability of partial differential equations, we constructed a multiple learnable edge module to generate precise edge information, incorporating the anisotropic image processing idea of Perona-Malik (PM) model into the neural network. Second, a multiscale reformative coordinate attention module was designed to extract multiscale information. Non-overlapping dilated convolution capturing abundant contextual content was combined with coordinate attention which could embed the spatial location information of important features into the channel attention map. Finally, we imposed additional constraints on the edge information using edge-enhanced multiscale perceptual loss to avoid structure loss and over-smoothing.Main results.Experiments are conducted on simulated and real datasets. The quantitative and qualitative results show that the proposed method has better performance in suppressing noise/artifacts and preserving edges/structures.Significance.This work proposes a novel edge feature extraction method that unfolds partial differential equation into neural networks, which contributes to the interpretability and clinical application value of neural network.Objective.Various deep learning methods have recently been used for low dose CT (LDCT) denoising. Aggressive denoising may destroy the edge and fine anatomical structures of CT images. Therefore a key issue in LDCT denoising tasks is the difficulty of balancing noise/artifact suppression and edge/structure preservation.Approach.We proposed an LDCT denoising network based on the encoder-decoder structure, namely the Learnable PM diffusion coefficient and efficient attention network (PMA-Net). First, using the powerful feature modeling capability of partial differential equations, we constructed a multiple learnable edge module to generate precise edge information, incorporating the anisotropic image processing idea of Perona-Malik (PM) model into the neural network. Second, a multiscale reformative coordinate attention module was designed to extract multiscale information. Non-overlapping dilated convolution capturing abundant contextual content was combined with coordinate attention which could embed the spatial location information of important features into the channel attention map. Finally, we imposed additional constraints on the edge information using edge-enhanced multiscale perceptual loss to avoid structure loss and over-smoothing.Main results.Experiments are conducted on simulated and real datasets. The quantitative and qualitative results show that the proposed method has better performance in suppressing noise/artifacts and preserving edges/structures.Significance.This work proposes a novel edge feature extraction method that unfolds partial differential equation into neural networks, which contributes to the interpretability and clinical application value of neural network. Various deep learning methods have recently been used for low dose CT (LDCT) denoising. Aggressive denoising may destroy the edge and fine anatomical structures of CT images. Therefore a key issue in LDCT denoising tasks is the difficulty of balancing noise/artifact suppression and edge/structure preservation. We proposed an LDCT denoising network based on the encoder-decoder structure, namely the Learnable PM diffusion coefficient and efficient attention network (PMA-Net). First, using the powerful feature modeling capability of partial differential equations, we constructed a multiple learnable edge module to generate precise edge information, incorporating the anisotropic image processing idea of Perona-Malik (PM) model into the neural network. Second, a multiscale reformative coordinate attention module was designed to extract multiscale information. Non-overlapping dilated convolution capturing abundant contextual content was combined with coordinate attention which could embed the spatial location information of important features into the channel attention map. Finally, we imposed additional constraints on the edge information using edge-enhanced multiscale perceptual loss to avoid structure loss and over-smoothing. Experiments are conducted on simulated and real datasets. The quantitative and qualitative results show that the proposed method has better performance in suppressing noise/artifacts and preserving edges/structures. This work proposes a novel edge feature extraction method that unfolds partial differential equation into neural networks, which contributes to the interpretability and clinical application value of neural network. |
| Author | Cheng, Weiting Yan, Rongbiao Zhang, Haowen Liu, Yi Chen, Yang Hou, Ruifeng Gui, Zhiguo Li, Shu Zhang, Pengcheng |
| Author_xml | – sequence: 1 givenname: Haowen surname: Zhang fullname: Zhang, Haowen organization: State Key Laboratory of Dynamic Testing Technology, North University of China, Taiyuan 030051, People's Republic of China – sequence: 2 givenname: Pengcheng orcidid: 0000-0003-3560-1167 surname: Zhang fullname: Zhang, Pengcheng organization: State Key Laboratory of Dynamic Testing Technology, North University of China, Taiyuan 030051, People's Republic of China – sequence: 3 givenname: Weiting surname: Cheng fullname: Cheng, Weiting organization: State Key Laboratory of Dynamic Testing Technology, North University of China, Taiyuan 030051, People's Republic of China – sequence: 4 givenname: Shu surname: Li fullname: Li, Shu organization: State Key Laboratory of Dynamic Testing Technology, North University of China, Taiyuan 030051, People's Republic of China – sequence: 5 givenname: Rongbiao surname: Yan fullname: Yan, Rongbiao organization: State Key Laboratory of Dynamic Testing Technology, North University of China, Taiyuan 030051, People's Republic of China – sequence: 6 givenname: Ruifeng surname: Hou fullname: Hou, Ruifeng organization: State Key Laboratory of Dynamic Testing Technology, North University of China, Taiyuan 030051, People's Republic of China – sequence: 7 givenname: Zhiguo orcidid: 0000-0002-8991-9907 surname: Gui fullname: Gui, Zhiguo organization: State Key Laboratory of Dynamic Testing Technology, North University of China, Taiyuan 030051, People's Republic of China – sequence: 8 givenname: Yi surname: Liu fullname: Liu, Yi organization: Key Laboratory of Computer Network and Information Integration (Southeast University), Ministry of Education, Nanjing 210096, People's Republic of China – sequence: 9 givenname: Yang orcidid: 0000-0002-5660-6349 surname: Chen fullname: Chen, Yang organization: Jiangsu Provincial Joint International Research Laboratory of Medical Information Processing, Southeast University, Nanjing, People's Republic of China |
| BackLink | https://www.ncbi.nlm.nih.gov/pubmed/37536336$$D View this record in MEDLINE/PubMed |
| BookMark | eNpNkDtPwzAUhS1URB-wMyGPLKF-xUlGVPGSimAoc-TE18glsYvtUPHvCaIgpnN0z3fucOZo4rwDhM4puaKkLJeUS5rJXJKlakFzfoRmf6fJPz9F8xi3hFBaMnGCprzIueRcztB2DSo41XSAnx-xtsYM0XqHWw_G2NaCSxErp3EA40Ovkv2AMfRBW6cSYJXSiHw3HKS9D294xHDn91j7CHi1wRqct9G611N0bFQX4eygC_Rye7NZ3Wfrp7uH1fU6a4XIU9ZwXpmCMSYKKSujtQIBrCwoAKWakQryRmhmOBSqZQKUoHkrNakUbbjQhC3Q5c_fXfDvA8RU9za20HXKgR9izUohKyaJzEf04oAOTQ-63gXbq_BZ_-7DvgBhI2tC |
| CitedBy_id | crossref_primary_10_1088_1361_6560_ad45a4 crossref_primary_10_1007_s13534_024_00419_7 crossref_primary_10_1016_j_neucom_2025_129667 |
| ContentType | Journal Article |
| Copyright | 2023 Institute of Physics and Engineering in Medicine. |
| Copyright_xml | – notice: 2023 Institute of Physics and Engineering in Medicine. |
| DBID | CGR CUY CVF ECM EIF NPM 7X8 |
| DOI | 10.1088/1361-6560/aced33 |
| DatabaseName | Medline MEDLINE MEDLINE (Ovid) MEDLINE MEDLINE PubMed MEDLINE - Academic |
| DatabaseTitle | MEDLINE Medline Complete MEDLINE with Full Text PubMed MEDLINE (Ovid) MEDLINE - Academic |
| DatabaseTitleList | MEDLINE - Academic MEDLINE |
| Database_xml | – sequence: 1 dbid: NPM name: PubMed url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database – sequence: 2 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database |
| DeliveryMethod | no_fulltext_linktorsrc |
| Discipline | Medicine Biology Physics |
| EISSN | 1361-6560 |
| ExternalDocumentID | 37536336 |
| Genre | Journal Article |
| GroupedDBID | --- -DZ -~X 123 1JI 4.4 5B3 5RE 5VS 5ZH 7.M 7.Q AAGCD AAJIO AAJKP AATNI ABCXL ABHWH ABJNI ABLJU ABQJV ABVAM ACAFW ACGFS ACHIP ADEQX AEFHF AEINN AENEX AFYNE AKPSB ALMA_UNASSIGNED_HOLDINGS AOAED ASPBG ATQHT AVWKF AZFZN CBCFC CEBXE CGR CJUJL CRLBU CS3 CUY CVF DU5 EBS ECM EDWGO EIF EJD EMSAF EPQRW EQZZN F5P IHE IJHAN IOP IZVLO KOT LAP M45 N5L N9A NPM P2P PJBAE R4D RIN RNS RO9 ROL RPA SY9 TN5 W28 XPP 7X8 ABUFD |
| ID | FETCH-LOGICAL-c445t-b339f722247669fddae4e2871ee11d209e5b4d2f3e7ac24ea415c6d09a1b34d02 |
| IEDL.DBID | 7X8 |
| ISICitedReferencesCount | 5 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001121394600001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 1361-6560 |
| IngestDate | Sat Nov 01 14:53:33 EDT 2025 Tue Sep 09 02:34:16 EDT 2025 |
| IsDoiOpenAccess | false |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 24 |
| Keywords | encoder–decoder structure LDCT attention mechanism perceptual loss Perona–Malik model |
| Language | English |
| License | 2023 Institute of Physics and Engineering in Medicine. |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c445t-b339f722247669fddae4e2871ee11d209e5b4d2f3e7ac24ea415c6d09a1b34d02 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
| ORCID | 0000-0002-8991-9907 0000-0003-3560-1167 0000-0002-5660-6349 |
| OpenAccessLink | https://doi.org/10.1088/1361-6560/aced33 |
| PMID | 37536336 |
| PQID | 2846926065 |
| PQPubID | 23479 |
| ParticipantIDs | proquest_miscellaneous_2846926065 pubmed_primary_37536336 |
| PublicationCentury | 2000 |
| PublicationDate | 2023-12-21 |
| PublicationDateYYYYMMDD | 2023-12-21 |
| PublicationDate_xml | – month: 12 year: 2023 text: 2023-12-21 day: 21 |
| PublicationDecade | 2020 |
| PublicationPlace | England |
| PublicationPlace_xml | – name: England |
| PublicationTitle | Physics in medicine & biology |
| PublicationTitleAlternate | Phys Med Biol |
| PublicationYear | 2023 |
| SSID | ssj0011824 |
| Score | 2.4585054 |
| Snippet | Various deep learning methods have recently been used for low dose CT (LDCT) denoising. Aggressive denoising may destroy the edge and fine anatomical... Objective.Various deep learning methods have recently been used for low dose CT (LDCT) denoising. Aggressive denoising may destroy the edge and fine anatomical... |
| SourceID | proquest pubmed |
| SourceType | Aggregation Database Index Database |
| SubjectTerms | Anisotropy Image Processing, Computer-Assisted Neural Networks, Computer Signal-To-Noise Ratio Tomography, X-Ray Computed |
| Title | Learnable PM diffusion coefficients and reformative coordinate attention network for low dose CT denoising |
| URI | https://www.ncbi.nlm.nih.gov/pubmed/37536336 https://www.proquest.com/docview/2846926065 |
| Volume | 68 |
| WOSCitedRecordID | wos001121394600001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1JS8NAFB7UqnhxqVvdGMFraJKZTDInkWLxYEsPFXoLswUqklTTVvz3vpek9iQIXnKYJBBm3rz53pLvI-SOR8pJprknuMIyozCe4gy1XliWJRwweU2Z_xwPh8lkIkdNwq1s2ipXPrFy1LYwmCPvghsVEsC3iO5n7x6qRmF1tZHQ2CQtBlAGrTqerKsIgJ1rUVsReEgy05QpYWN1f8a6yjiLsrm_AczqoOkf_PcTD8l-AzHpQ20TR2TD5W2yU4tOfrXJ7qApp8Ng1f9pymPyWvGs4m9UdDSgqJqywDQaNYWrOCaw3YKq3FI4UGuUu3RwEyLXaQ5olSJLZ9U3SfO6r5zCY_St-KS2KB3tjSn4t2KKiYkT8tJ_HPeevEaGwTOcR3NPMyazGHAEj4WQmbXKcYeBlnNBYENfukhzG2bMxcqE3CnABEZYX6pAM2798JRs5UXuzglVibEslL62vuaR1ipMAh4YaaOAaSZFh9yuZjYFM8fahcpdsSjT9dx2yFm9POms5uNIGYRcgjFx8Ye3L8keCsZjQ0oYXJFWBpvcXZNts5xPy4-byn7gOhwNvgEGH8-8 |
| linkProvider | ProQuest |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Learnable+PM+diffusion+coefficients+and+reformative+coordinate+attention+network+for+low+dose+CT+denoising&rft.jtitle=Physics+in+medicine+%26+biology&rft.au=Zhang%2C+Haowen&rft.au=Zhang%2C+Pengcheng&rft.au=Cheng%2C+Weiting&rft.au=Li%2C+Shu&rft.date=2023-12-21&rft.issn=1361-6560&rft.eissn=1361-6560&rft.volume=68&rft.issue=24&rft_id=info:doi/10.1088%2F1361-6560%2Faced33&rft.externalDBID=NO_FULL_TEXT |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1361-6560&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1361-6560&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1361-6560&client=summon |