MiMICRI: Towards Domain-centered Counterfactual Explanations of Cardiovascular Image Classification Models

The recent prevalence of publicly accessible, large medical imaging datasets has led to a proliferation of artificial intelligence (AI) models for cardiovascular image classification and analysis. At the same time, the potentially significant impacts of these models have motivated the development of...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:FAccT '24 : proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency (ACM FAccT '24) : June 3rd-6th 2024, Rio de Janeiro, Brazil. ACM Conference on Fairness, Accountability, and Transparency (2024 : Rio de Ja Jg. 2024; S. 1861
Hauptverfasser: Guo, Grace, Deng, Lifu, Tandon, Animesh, Endert, Alex, Kwon, Bum Chul
Format: Journal Article
Sprache:Englisch
Veröffentlicht: United States 01.06.2024
Schlagworte:
Online-Zugang:Weitere Angaben
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract The recent prevalence of publicly accessible, large medical imaging datasets has led to a proliferation of artificial intelligence (AI) models for cardiovascular image classification and analysis. At the same time, the potentially significant impacts of these models have motivated the development of a range of explainable AI (XAI) methods that aim to explain model predictions given certain image inputs. However, many of these methods are not developed or evaluated with domain experts, and explanations are not contextualized in terms of medical expertise or domain knowledge. In this paper, we propose a novel framework and python library, MiMICRI, that provides domain-centered counterfactual explanations of cardiovascular image classification models. MiMICRI helps users interactively select and replace segments of medical images that correspond to morphological structures. From the counterfactuals generated, users can then assess the influence of each segment on model predictions, and validate the model against known medical facts. We evaluate this library with two medical experts. Our evaluation demonstrates that a domain-centered XAI approach can enhance the interpretability of model explanations, and help experts reason about models in terms of relevant domain knowledge. However, concerns were also surfaced about the clinical plausibility of the counterfactuals generated. We conclude with a discussion on the generalizability and trustworthiness of the MiMICRI framework, as well as the implications of our findings on the development of domain-centered XAI methods for model interpretability in healthcare contexts.
AbstractList The recent prevalence of publicly accessible, large medical imaging datasets has led to a proliferation of artificial intelligence (AI) models for cardiovascular image classification and analysis. At the same time, the potentially significant impacts of these models have motivated the development of a range of explainable AI (XAI) methods that aim to explain model predictions given certain image inputs. However, many of these methods are not developed or evaluated with domain experts, and explanations are not contextualized in terms of medical expertise or domain knowledge. In this paper, we propose a novel framework and python library, MiMICRI, that provides domain-centered counterfactual explanations of cardiovascular image classification models. MiMICRI helps users interactively select and replace segments of medical images that correspond to morphological structures. From the counterfactuals generated, users can then assess the influence of each segment on model predictions, and validate the model against known medical facts. We evaluate this library with two medical experts. Our evaluation demonstrates that a domain-centered XAI approach can enhance the interpretability of model explanations, and help experts reason about models in terms of relevant domain knowledge. However, concerns were also surfaced about the clinical plausibility of the counterfactuals generated. We conclude with a discussion on the generalizability and trustworthiness of the MiMICRI framework, as well as the implications of our findings on the development of domain-centered XAI methods for model interpretability in healthcare contexts.
The recent prevalence of publicly accessible, large medical imaging datasets has led to a proliferation of artificial intelligence (AI) models for cardiovascular image classification and analysis. At the same time, the potentially significant impacts of these models have motivated the development of a range of explainable AI (XAI) methods that aim to explain model predictions given certain image inputs. However, many of these methods are not developed or evaluated with domain experts, and explanations are not contextualized in terms of medical expertise or domain knowledge. In this paper, we propose a novel framework and python library, MiMICRI, that provides domain-centered counterfactual explanations of cardiovascular image classification models. MiMICRI helps users interactively select and replace segments of medical images that correspond to morphological structures. From the counterfactuals generated, users can then assess the influence of each segment on model predictions, and validate the model against known medical facts. We evaluate this library with two medical experts. Our evaluation demonstrates that a domain-centered XAI approach can enhance the interpretability of model explanations, and help experts reason about models in terms of relevant domain knowledge. However, concerns were also surfaced about the clinical plausibility of the counterfactuals generated. We conclude with a discussion on the generalizability and trustworthiness of the MiMICRI framework, as well as the implications of our findings on the development of domain-centered XAI methods for model interpretability in healthcare contexts.The recent prevalence of publicly accessible, large medical imaging datasets has led to a proliferation of artificial intelligence (AI) models for cardiovascular image classification and analysis. At the same time, the potentially significant impacts of these models have motivated the development of a range of explainable AI (XAI) methods that aim to explain model predictions given certain image inputs. However, many of these methods are not developed or evaluated with domain experts, and explanations are not contextualized in terms of medical expertise or domain knowledge. In this paper, we propose a novel framework and python library, MiMICRI, that provides domain-centered counterfactual explanations of cardiovascular image classification models. MiMICRI helps users interactively select and replace segments of medical images that correspond to morphological structures. From the counterfactuals generated, users can then assess the influence of each segment on model predictions, and validate the model against known medical facts. We evaluate this library with two medical experts. Our evaluation demonstrates that a domain-centered XAI approach can enhance the interpretability of model explanations, and help experts reason about models in terms of relevant domain knowledge. However, concerns were also surfaced about the clinical plausibility of the counterfactuals generated. We conclude with a discussion on the generalizability and trustworthiness of the MiMICRI framework, as well as the implications of our findings on the development of domain-centered XAI methods for model interpretability in healthcare contexts.
Author Endert, Alex
Guo, Grace
Deng, Lifu
Tandon, Animesh
Kwon, Bum Chul
Author_xml – sequence: 1
  givenname: Grace
  surname: Guo
  fullname: Guo, Grace
  organization: Georgia Institute of Technology Atlanta, Georgia, USA
– sequence: 2
  givenname: Lifu
  surname: Deng
  fullname: Deng, Lifu
  organization: Cleveland Clinic Cleveland, Ohio, USA
– sequence: 3
  givenname: Animesh
  surname: Tandon
  fullname: Tandon, Animesh
  organization: Cleveland Clinic Cleveland, Ohio, USA
– sequence: 4
  givenname: Alex
  surname: Endert
  fullname: Endert, Alex
  organization: Georgia Institute of Technology Atlanta, Georgia, USA
– sequence: 5
  givenname: Bum Chul
  surname: Kwon
  fullname: Kwon, Bum Chul
  organization: IBM Research Cambridge, Massachusetts, USA
BackLink https://www.ncbi.nlm.nih.gov/pubmed/39877054$$D View this record in MEDLINE/PubMed
BookMark eNo1kDlPxDAUhF2AOBZqOuSSJuAjcWI6FBaItCsktNTRi_2CjJx4iROOf0-ApZopvhmN5pjs9aFHQs44u-Q8za6kkowzdSlVphnnB-RQ6iLPWZYekde1W1flU3VNN-EDBhvpbejA9YnBfsQBLS3D9ONaMOMEni4_tx56GF3oIw0tLeeQC-8QzeRhoFUHL0hLDzG61plfjq6DRR9PyH4LPuLpThfk-W65KR-S1eN9Vd6sEpAFGxOltc0lL5gAgxwUN6mWLG84GGtEw61Ak0MKOsuRKWm5ZUI2ugCRQisNiAW5-OvdDuFtwjjWnYsG_TwbwxRryRXT6XwEm9HzHTo1Hdp6O7gOhq_6_x_xDZI6Y18
ContentType Journal Article
DBID NPM
7X8
DOI 10.1145/3630106.3659011
DatabaseName PubMed
MEDLINE - Academic
DatabaseTitle PubMed
MEDLINE - Academic
DatabaseTitleList PubMed
MEDLINE - Academic
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
ExternalDocumentID 39877054
Genre Journal Article
GrantInformation_xml – fundername: NHLBI NIH HHS
  grantid: K23 HL150279
GroupedDBID NPM
7X8
ID FETCH-LOGICAL-a380t-699d731802ace1a61c49307b1acdc2b1d2ec7a4a957e063d1d023b98a24af3ca2
IEDL.DBID 7X8
ISICitedReferencesCount 0
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001253359300125&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
IngestDate Fri Jul 11 16:42:42 EDT 2025
Thu Feb 06 04:52:33 EST 2025
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed false
IsScholarly false
Keywords explainable AI
human-centered AI
interactive visualizations
counterfactual explanation
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-a380t-699d731802ace1a61c49307b1acdc2b1d2ec7a4a957e063d1d023b98a24af3ca2
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
OpenAccessLink https://pubmed.ncbi.nlm.nih.gov/PMC11774553
PMID 39877054
PQID 3160940110
PQPubID 23479
ParticipantIDs proquest_miscellaneous_3160940110
pubmed_primary_39877054
PublicationCentury 2000
PublicationDate 20240601
PublicationDateYYYYMMDD 2024-06-01
PublicationDate_xml – month: 6
  year: 2024
  text: 20240601
  day: 1
PublicationDecade 2020
PublicationPlace United States
PublicationPlace_xml – name: United States
PublicationTitle FAccT '24 : proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency (ACM FAccT '24) : June 3rd-6th 2024, Rio de Janeiro, Brazil. ACM Conference on Fairness, Accountability, and Transparency (2024 : Rio de Ja
PublicationTitleAlternate FACCT 24 (2024)
PublicationYear 2024
Score 1.8865079
Snippet The recent prevalence of publicly accessible, large medical imaging datasets has led to a proliferation of artificial intelligence (AI) models for...
SourceID proquest
pubmed
SourceType Aggregation Database
Index Database
StartPage 1861
Title MiMICRI: Towards Domain-centered Counterfactual Explanations of Cardiovascular Image Classification Models
URI https://www.ncbi.nlm.nih.gov/pubmed/39877054
https://www.proquest.com/docview/3160940110
Volume 2024
WOSCitedRecordID wos001253359300125&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/eLvHCXMwpV3NS8MwFA_qPHjxA7_mFxG8xrVJmzReRKbDgRtDpuxWXpMUJq6ddvr3m6Qd4kEQvPTUQnlJ3vu9l997P4QuINeCR1qQmIMgUcYkAcWAJDKnCQfKtVdreH4Qw2EymchRU3CrGlrl0id6R61L5WrkHRZyN-rNRqvr-RtxqlHudrWR0FhFLWahjKN0iUnSTPAJo7jDOHNJzyXjvsfydxjpw0lv678_so02GyCJb-qV30ErpthFL4OpNetj_wqPPR22wrflzOb-xHEwnSondj3oTpkafOMIdhQ8qAuCFS5z3P1BT8X9mXU32AtnOkqRfw87-bTXag899e7G3XvSqCkQYEmwIFxKLZgb-AbKhMBDFUl7wLMQlFY0CzU1SkAEMhbG4hYdahvOM5kAjSBnCug-WivKwhwi7O8vAykCoSBiNJdc6dyIBCBQNqWO2-h8abrU7lZ3BQGFKT-q9Nt4bXRQ2z-d12M1Urt-QlgEefSHr4_RBrXoouZsnaBWbs-qOUXr6nMxrd7P_Dawz-Fo8AW_EcCK
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=MiMICRI%3A+Towards+Domain-centered+Counterfactual+Explanations+of+Cardiovascular+Image+Classification+Models&rft.jtitle=FAccT+%2724+%3A+proceedings+of+the+2024+ACM+Conference+on+Fairness%2C+Accountability%2C+and+Transparency+%28ACM+FAccT+%2724%29+%3A+June+3rd-6th+2024%2C+Rio+de+Janeiro%2C+Brazil.+ACM+Conference+on+Fairness%2C+Accountability%2C+and+Transparency+%282024+%3A+Rio+de+Ja&rft.au=Guo%2C+Grace&rft.au=Deng%2C+Lifu&rft.au=Tandon%2C+Animesh&rft.au=Endert%2C+Alex&rft.date=2024-06-01&rft.volume=2024&rft.spage=1861&rft_id=info:doi/10.1145%2F3630106.3659011&rft.externalDBID=NO_FULL_TEXT