Chelsea A. H. Sargeant, Angela Davey, Alan McWilliam
{"title":"从4DCT运动补偿重建中提取放射组学的适用性。","authors":"Chelsea A. H. Sargeant, Angela Davey, Alan McWilliam","doi":"10.1002/acm2.70254","DOIUrl":null,"url":null,"abstract":"<div>\n \n \n <section>\n \n <h3> Purpose</h3>\n \n <p>The most appropriate 4DCT image for radiomics analysis remains uncertain, with options including individual respiratory phases and motion-compensated (MC) reconstructions. This study compares radiomic features derived from MC reconstructions and patient-specific, 4D phases, evaluating their influence on feature selection, robustness to registration algorithms and implications for radiomics analysis.</p>\n </section>\n \n <section>\n \n <h3> Methods</h3>\n \n <p>This study included 223 NSCLC patients. MC reconstructions (MC<sub>mean</sub> and MC<sub>median</sub>) were generated using three different registration algorithms and compared to a patient-specific optimal 4D phase (TOp), defined as the phase where radiomic feature values exhibited the smallest variability. Ninety-three features, extracted from the tumor region, underwent unsupervised selection to assess how image type influenced feature selection and redundancy. The impact of image type on distant failure prediction was evaluated using bootstrapped univariable Cox regression (<i>p</i> < 0.05) and multivariable modeling. Model performance was assessed across 500 bootstrap resamples, with feature selection frequency, concordance index (CI), and Akaike Information Criterion (AIC) recorded.</p>\n </section>\n \n <section>\n \n <h3> Results</h3>\n \n <p>Over 60% of selected features differed between 4D phases and MC reconstructions, indicating image type influences feature selection. The proportion of remaining features varied: 11.8% (TOp), 15.1% (MC<sub>mean</sub>), and 12.9% (MC<sub>median</sub>). The single-phase model, TOp, achieved a CI of 0.72 [0.64–0.77] and an AIC of 267.53, but did not demonstrate clear superiority over MC-based models. Both MC<sub>median</sub> and TOp showed modest improvements over the clinical model, suggesting both image types offer comparable prognostic potential. MC reconstructions were largely robust across registration algorithms, but MC<sub>mean</sub> and MC<sub>median</sub> reconstructions should not be used interchangeably.</p>\n </section>\n \n <section>\n \n <h3> Conclusion</h3>\n \n <p>This study highlights differences in radiomic feature selection and predictive performance between MC reconstructions and individual 4DCT phases. MC reconstructions were a viable alternative, demonstrating robustness across registration algorithms. Both approaches can be integrated into radiomics pipelines, but image type selection should be carefully considered.</p>\n </section>\n </div>","PeriodicalId":14989,"journal":{"name":"Journal of Applied Clinical Medical Physics","volume":"26 10","pages":""},"PeriodicalIF":2.2000,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12479209/pdf/","citationCount":"0","resultStr":"{\"title\":\"The suitability of radiomics extracted from 4DCT motion-compensated reconstruction\",\"authors\":\"Chelsea A. H. Sargeant, Angela Davey, Alan McWilliam\",\"doi\":\"10.1002/acm2.70254\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div>\\n \\n \\n <section>\\n \\n <h3> Purpose</h3>\\n \\n <p>The most appropriate 4DCT image for radiomics analysis remains uncertain, with options including individual respiratory phases and motion-compensated (MC) reconstructions. This study compares radiomic features derived from MC reconstructions and patient-specific, 4D phases, evaluating their influence on feature selection, robustness to registration algorithms and implications for radiomics analysis.</p>\\n </section>\\n \\n <section>\\n \\n <h3> Methods</h3>\\n \\n <p>This study included 223 NSCLC patients. MC reconstructions (MC<sub>mean</sub> and MC<sub>median</sub>) were generated using three different registration algorithms and compared to a patient-specific optimal 4D phase (TOp), defined as the phase where radiomic feature values exhibited the smallest variability. Ninety-three features, extracted from the tumor region, underwent unsupervised selection to assess how image type influenced feature selection and redundancy. The impact of image type on distant failure prediction was evaluated using bootstrapped univariable Cox regression (<i>p</i> < 0.05) and multivariable modeling. Model performance was assessed across 500 bootstrap resamples, with feature selection frequency, concordance index (CI), and Akaike Information Criterion (AIC) recorded.</p>\\n </section>\\n \\n <section>\\n \\n <h3> Results</h3>\\n \\n <p>Over 60% of selected features differed between 4D phases and MC reconstructions, indicating image type influences feature selection. The proportion of remaining features varied: 11.8% (TOp), 15.1% (MC<sub>mean</sub>), and 12.9% (MC<sub>median</sub>). The single-phase model, TOp, achieved a CI of 0.72 [0.64–0.77] and an AIC of 267.53, but did not demonstrate clear superiority over MC-based models. Both MC<sub>median</sub> and TOp showed modest improvements over the clinical model, suggesting both image types offer comparable prognostic potential. MC reconstructions were largely robust across registration algorithms, but MC<sub>mean</sub> and MC<sub>median</sub> reconstructions should not be used interchangeably.</p>\\n </section>\\n \\n <section>\\n \\n <h3> Conclusion</h3>\\n \\n <p>This study highlights differences in radiomic feature selection and predictive performance between MC reconstructions and individual 4DCT phases. MC reconstructions were a viable alternative, demonstrating robustness across registration algorithms. Both approaches can be integrated into radiomics pipelines, but image type selection should be carefully considered.</p>\\n </section>\\n </div>\",\"PeriodicalId\":14989,\"journal\":{\"name\":\"Journal of Applied Clinical Medical Physics\",\"volume\":\"26 10\",\"pages\":\"\"},\"PeriodicalIF\":2.2000,\"publicationDate\":\"2025-09-29\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12479209/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Applied Clinical Medical Physics\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://aapm.onlinelibrary.wiley.com/doi/10.1002/acm2.70254\",\"RegionNum\":4,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Applied Clinical Medical Physics","FirstCategoryId":"3","ListUrlMain":"https://aapm.onlinelibrary.wiley.com/doi/10.1002/acm2.70254","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
The suitability of radiomics extracted from 4DCT motion-compensated reconstruction
Purpose
The most appropriate 4DCT image for radiomics analysis remains uncertain, with options including individual respiratory phases and motion-compensated (MC) reconstructions. This study compares radiomic features derived from MC reconstructions and patient-specific, 4D phases, evaluating their influence on feature selection, robustness to registration algorithms and implications for radiomics analysis.
Methods
This study included 223 NSCLC patients. MC reconstructions (MCmean and MCmedian) were generated using three different registration algorithms and compared to a patient-specific optimal 4D phase (TOp), defined as the phase where radiomic feature values exhibited the smallest variability. Ninety-three features, extracted from the tumor region, underwent unsupervised selection to assess how image type influenced feature selection and redundancy. The impact of image type on distant failure prediction was evaluated using bootstrapped univariable Cox regression (p < 0.05) and multivariable modeling. Model performance was assessed across 500 bootstrap resamples, with feature selection frequency, concordance index (CI), and Akaike Information Criterion (AIC) recorded.
Results
Over 60% of selected features differed between 4D phases and MC reconstructions, indicating image type influences feature selection. The proportion of remaining features varied: 11.8% (TOp), 15.1% (MCmean), and 12.9% (MCmedian). The single-phase model, TOp, achieved a CI of 0.72 [0.64–0.77] and an AIC of 267.53, but did not demonstrate clear superiority over MC-based models. Both MCmedian and TOp showed modest improvements over the clinical model, suggesting both image types offer comparable prognostic potential. MC reconstructions were largely robust across registration algorithms, but MCmean and MCmedian reconstructions should not be used interchangeably.
Conclusion
This study highlights differences in radiomic feature selection and predictive performance between MC reconstructions and individual 4DCT phases. MC reconstructions were a viable alternative, demonstrating robustness across registration algorithms. Both approaches can be integrated into radiomics pipelines, but image type selection should be carefully considered.
期刊介绍:
Journal of Applied Clinical Medical Physics is an international Open Access publication dedicated to clinical medical physics. JACMP welcomes original contributions dealing with all aspects of medical physics from scientists working in the clinical medical physics around the world. JACMP accepts only online submission.
JACMP will publish:
-Original Contributions: Peer-reviewed, investigations that represent new and significant contributions to the field. Recommended word count: up to 7500.
-Review Articles: Reviews of major areas or sub-areas in the field of clinical medical physics. These articles may be of any length and are peer reviewed.
-Technical Notes: These should be no longer than 3000 words, including key references.
-Letters to the Editor: Comments on papers published in JACMP or on any other matters of interest to clinical medical physics. These should not be more than 1250 (including the literature) and their publication is only based on the decision of the editor, who occasionally asks experts on the merit of the contents.
-Book Reviews: The editorial office solicits Book Reviews.
-Announcements of Forthcoming Meetings: The Editor may provide notice of forthcoming meetings, course offerings, and other events relevant to clinical medical physics.
-Parallel Opposed Editorial: We welcome topics relevant to clinical practice and medical physics profession. The contents can be controversial debate or opposed aspects of an issue. One author argues for the position and the other against. Each side of the debate contains an opening statement up to 800 words, followed by a rebuttal up to 500 words. Readers interested in participating in this series should contact the moderator with a proposed title and a short description of the topic