File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: The impact of radiomic feature selection on clinical modelling

TitleThe impact of radiomic feature selection on clinical modelling
Authors
Issue Date2020
PublisherSpringerOpen. The Journal's web site is located at http://www.springer.com/medicine/radiology/journal/13244
Citation
European Congress of Radiology (ECR) 2020, Virtual Congress, Vienna, Austria, 15-17 July 2020. Book of Abstracts in Insights into Imaging, 2020, v. 11 n. Suppl. 1, article no. 34: p. 624, abstract no. RPS 1705-11 How to Cite?
AbstractPurpose: Software choice often dictates what radiomic features can be extracted from images. This study analyses how different feature sets affect clinical models in two cancer cohorts: PET oesophagus (in-house dataset) and CT lung (open dataset). Methods and materials: 95 patients (65 training and 30 validation cohorts) who had undergone pre-treatment 18F-FDG-PET studies were included and classified as those who achieved complete pathological response (pCR) and those who did not. The primary tumours were segmented using a fixed threshold approach, radiomic features were extracted, features were annotated according to the international biomarker standardisation initiative. Feature reduction with the minimum redundancy maximum relevance method was performed. Logistical regression and random forest models for clinical outcomes were constructed and compared using ROC curves, decision curves, and McNemar's test. Results: Of the 5 feature sets tested, AUC values for the two models ranged between 0.6 and 0.87. Reduced feature sets did not have much overlap (10- 25%) and showed substantial differences in their ability to predict outcomes with our models and datasets. A combined superset showed the best performance (AUC 0.87) but did not show much reduction in feature correlation. Conclusion: High-dimensional data requires extensive feature reduction and increased scrutiny for overfitting, but if a proper methodology is applied, the results of combining multiple feature sets may be beneficial for modelling with radiomics in cancer. Limitations: The PET dataset is not very large (low incidence and expensive modality) and the event frequency is slightly below 40%. The fixed threshold segmentation may be less accurate for lesion size, but it was chosen to maximise radiomics stability. Ethics committee approval: The study was approved by the local ethics committee in accordance with the Helsinki Declaration and all patient information was anonymised prior to analysis.
Persistent Identifierhttp://hdl.handle.net/10722/293774
ISSN
2023 Impact Factor: 4.1
2023 SCImago Journal Rankings: 1.240

 

DC FieldValueLanguage
dc.contributor.authorvan Lunenburg, JTJ-
dc.contributor.authorChiu, WHK-
dc.date.accessioned2020-11-23T08:21:37Z-
dc.date.available2020-11-23T08:21:37Z-
dc.date.issued2020-
dc.identifier.citationEuropean Congress of Radiology (ECR) 2020, Virtual Congress, Vienna, Austria, 15-17 July 2020. Book of Abstracts in Insights into Imaging, 2020, v. 11 n. Suppl. 1, article no. 34: p. 624, abstract no. RPS 1705-11-
dc.identifier.issn1869-4101-
dc.identifier.urihttp://hdl.handle.net/10722/293774-
dc.description.abstractPurpose: Software choice often dictates what radiomic features can be extracted from images. This study analyses how different feature sets affect clinical models in two cancer cohorts: PET oesophagus (in-house dataset) and CT lung (open dataset). Methods and materials: 95 patients (65 training and 30 validation cohorts) who had undergone pre-treatment 18F-FDG-PET studies were included and classified as those who achieved complete pathological response (pCR) and those who did not. The primary tumours were segmented using a fixed threshold approach, radiomic features were extracted, features were annotated according to the international biomarker standardisation initiative. Feature reduction with the minimum redundancy maximum relevance method was performed. Logistical regression and random forest models for clinical outcomes were constructed and compared using ROC curves, decision curves, and McNemar's test. Results: Of the 5 feature sets tested, AUC values for the two models ranged between 0.6 and 0.87. Reduced feature sets did not have much overlap (10- 25%) and showed substantial differences in their ability to predict outcomes with our models and datasets. A combined superset showed the best performance (AUC 0.87) but did not show much reduction in feature correlation. Conclusion: High-dimensional data requires extensive feature reduction and increased scrutiny for overfitting, but if a proper methodology is applied, the results of combining multiple feature sets may be beneficial for modelling with radiomics in cancer. Limitations: The PET dataset is not very large (low incidence and expensive modality) and the event frequency is slightly below 40%. The fixed threshold segmentation may be less accurate for lesion size, but it was chosen to maximise radiomics stability. Ethics committee approval: The study was approved by the local ethics committee in accordance with the Helsinki Declaration and all patient information was anonymised prior to analysis.-
dc.languageeng-
dc.publisherSpringerOpen. The Journal's web site is located at http://www.springer.com/medicine/radiology/journal/13244-
dc.relation.ispartofInsights into Imaging-
dc.relation.ispartofEuropean Congress of Radiology (ERC) 2020-
dc.titleThe impact of radiomic feature selection on clinical modelling-
dc.typeConference_Paper-
dc.identifier.emailChiu, WHK: kwhchiu@hku.hk-
dc.identifier.authorityChiu, WHK=rp02074-
dc.description.natureabstract-
dc.identifier.hkuros320293-
dc.identifier.volume11-
dc.identifier.issueSuppl. 1-
dc.identifier.spagearticle no. 34: p. 624-
dc.identifier.epagearticle no. 34: p. 624-
dc.publisher.placeGermany-
dc.identifier.partofdoi10.1186/s13244-020-00851-0-
dc.identifier.issnl1869-4101-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats