FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2913053403> ?p ?o ?g. }

• W2913053403 endingPage "22" @default.
• W2913053403 startingPage "21" @default.
• W2913053403 abstract "HomeRadiologyVol. 291, No. 1 PreviousNext Reviews and CommentaryFree AccessEditorialMammographic Parenchymal Analysis: Can We Do Better with Digital Assistance?Kitt Shaffer Kitt Shaffer Author AffiliationsFrom the Department of Radiology, Boston Medical Center and Boston University School of Medicine, 820 Harrison Ave, FGH 3001, Boston, MA 02118.Address correspondence to the author (e-mail: [email protected]).Kitt Shaffer Published Online:Feb 12 2019https://doi.org/10.1148/radiol.2019190085MoreSectionsPDF ToolsImage ViewerAdd to favoritesCiteTrack CitationsPermissionsReprints ShareShare onFacebookTwitterLinked In See also the article by Li et al in this issue.IntroductionAs radiologists, we always strive to improve both our sensitivity (ability to detect subtle abnormalities) and our specificity (ability to classify the things we find). Nowhere is this more apparent than in breast imaging. Excluding skin cancers, breast cancer is the most common cancer in women, and breast imaging was the first successful widespread image-based cancer screening (1), followed by CT for lung cancer (2), US for thyroid cancer (3), and CT and/or MRI for hepatoma (4). Although there have been many controversies regarding the optimal use of mammography in screening (5), no one denies its positive effects in terms of earlier breast cancer diagnosis. In this issue of Radiology, Dr Li and associates describe the results of adding a sophisticated analysis of parenchymal patterns to the analysis of known breast mass features to improve our ability to detect cancer with accuracy (6).The attempt at classification of breast parenchymal patterns has a long and not very successful history (7). It is clear to anyone who looks at mammograms that women vary in how their breast parenchyma looks on images. It is also clear that denser stroma is associated in some way with a higher risk of cancer, due to obscuration of clinically significant findings, physiologic effects in the tissues that predispose them to the development of cancer, or, possibly, a combination of both (8).What has not been clear is how best to quantify and classify the wide range of appearances of breast tissue on mammograms. The article by Dr Li and associates takes a different approach than most. Current practice is to determine the average density of the tissue overall, or what proportion of the breast is “dense” compared with fatty, despite research indicating that these visual assessments are not consistent (9). Instead, Li et al performed a texture analysis on a region of interest in the breast opposite from a known breast mass. From a list of 45 texture features, they used a neural network to select the particular features with the best ability to predict tumor risk. The features with the highest predictive value (skewness and power law beta) are not things that a viewer of images would be able to identify on their own, even if they understood what these mathematical concepts represent. In fact, the list of features analyzed includes fractal analysis, edge analysis, Fourier analysis, and neighborhood analysis, all of which are far from everyday activities for radiologists.After the texture analysis was completed, the additive role of these features was assessed by comparing the predictive value of tumor imaging features (also by using complex computational methods) to that of tumor features plus parenchymal texture analysis. In other words, how much did inclusion of analysis of uninvolved parenchyma change the overall predictive value of the study? The data set included a mix of malignant and benign tumors, with Breast Imaging Reporting and Data System category 4 or 5 assigned by the original interpreting physician. The addition of texture analysis of uninvolved breast parenchyma improved the area under the receiver operating characteristic curve from 0.79 to 0.84, which was statistically significant. Although many studies have used neural networks to analyze tumor features or parenchymal patterns, to my knowledge this is the first study to combine such an analysis of tumor with an analysis of the uninvolved parenchyma in the same patient.The application of such methodology to actual mammographic interpretation is a big step up from the current state of computer-assisted detection (CAD) used in most clinical breast imaging settings. The majority of CAD markings on mammograms are ignored by radiologists, as they are on normal structures, known benign findings, or normal tissue. The sophistication level of commercially available CAD programs is not comparable to that of the systems used by Dr Li and colleagues. However, in the future, it is conceivable that automated texture analysis may take the place of our current crude “density scale.” This future analysis could provide a better determination of an individual’s risk for breast cancer based on a more accurate assessment of stromal physiology, as shown with sophisticated analysis of parenchymal patterns. This is particularly important given the clinical concern about how to manage dense breasts (10).There are some important limitations of this study, mainly related to the small number of breast lesions included. Because of this, it was not possible to have independent training and testing sets for the neural network. This can be a source of bias. In addition, all images were from a single institution and obtained with one type of mammographic unit, so generalization to a wider set of mammograms might yield different results. However, the study does raise the bar for our analysis of breast density, which could lead to more accurate and reproducible ways to provide our patients with the best possible care and the most accurate readings of their screening mammograms.Disclosures of Conflicts of Interest: disclosed no relevant relationships.References1. Gershon-Cohen J, Forman M. Mammography of cancer. Bull N Y Acad Med 1964;40:674–689. Medline, Google Scholar2. Wender R, Fontham ET, Barrera E Jr, et al. American Cancer Society lung cancer screening guidelines. CA Cancer J Clin 2013;63(2):107–117. Crossref, Medline, Google Scholar3. Lin JS, Bowles EJA, Williams SB, Morrison CC. Screening for thyroid cancer: updated evidence report and systematic review for the US Preventive Services Task Force. JAMA 2017;317(18):1888–1903. Crossref, Medline, Google Scholar4. Kudo M, Matsui O, Izumi N, et al. Surveillance and diagnostic algorithm for hepatocellular carcinoma proposed by the Liver Cancer Study Group of Japan: 2014 update. Oncology 2014;87(Suppl 1):7–21. Crossref, Medline, Google Scholar5. Ray KM, Joe BN, Freimanis RI, Sickles EA, Hendrick RE. Screening mammography in women 40–49 years old: current evidence. AJR Am J Roentgenol 2018;210(2):264–270. Crossref, Medline, Google Scholar6. Li H, Mendel KR, Lan L, Sheth D, Giger ML. Digital mammography in breast cancer: additive value of radiomics of breast parenchyma. Radiology 2019;291:15–20. Link, Google Scholar7. Wolfe JN. Breast patterns as an index of risk for developing breast cancer. AJR Am J Roentgenol 1976;126(6):1130–1137. Crossref, Medline, Google Scholar8. Hooley RJ. Breast density legislation and clinical evidence. Radiol Clin North Am 2017;55(3):513–526. Crossref, Medline, Google Scholar9. Melnikow J, Fenton JJ, Whitlock EP, et al. Supplemental screening for breast cancer in women with dense breasts: a systematic review for the U.S. Preventive Service Task Force. Ann Intern Med 2016;164(4):268–278. Crossref, Medline, Google Scholar10. Ng KH, Lau S. Vision 20/20: mammographic breast density and its clinical applications. Med Phys 2015;42(12):7059–7077. Crossref, Medline, Google ScholarArticle HistoryReceived: Jan 11 2019Revision requested: Jan 15 2019Revision received: Jan 15 2019Accepted: Jan 16 2019Published online: Feb 12 2019Published in print: Apr 2019 FiguresReferencesRelatedDetailsAccompanying This ArticleDigital Mammography in Breast Cancer: Additive Value of Radiomics of Breast ParenchymaFeb 12 2019RadiologyRecommended Articles Is There a Place for Lymphatic Contrast-enhanced US in Thyroid Cancer?Radiology2023Volume: 307Issue: 4Breast Cancer Risk Prediction Using Deep LearningRadiology2021Volume: 301Issue: 3pp. 559-560Digital Mammography in Breast Cancer: Additive Value of Radiomics of Breast ParenchymaRadiology2019Volume: 291Issue: 1pp. 15-20Cancer Genomics and Important Oncologic Mutations: A Contemporary Guide for Body ImagersRadiology2017Volume: 283Issue: 2pp. 314-340Lymphatic Contrast-enhanced US to Improve the Diagnosis of Cervical Lymph Node Metastasis from Thyroid CancerRadiology2023Volume: 307Issue: 4See More RSNA Education Exhibits Li Fraumeni Syndrome: Genomics, Oncogenesis and Where to Look on Whole-Body Surveillance ImagingDigital Posters2019Hereditary Gynecological Neoplasia Syndromes: 2020 UpdateDigital Posters2020Li-fraumeni Syndrome And Breast Cancer: What, Why, How And Who.Digital Posters2021 RSNA Case Collection Dedifferentiated Thyroid CancerRSNA Case Collection2020Xanthogranulomatous mastitisRSNA Case Collection2022Fibroadenoma of the breastRSNA Case Collection2020 Vol. 291, No. 1 Metrics Altmetric Score PDF download" @default.
• W2913053403 created "2019-02-21" @default.
• W2913053403 creator A5067314445 @default.
• W2913053403 date "2019-04-01" @default.
• W2913053403 modified "2023-09-23" @default.
• W2913053403 title "Mammographic Parenchymal Analysis: Can We Do Better with Digital Assistance?" @default.
• W2913053403 doi "https://doi.org/10.1148/radiol.2019190085" @default.
• W2913053403 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/30747593" @default.
• W2913053403 hasPublicationYear "2019" @default.
• W2913053403 type Work @default.
• W2913053403 sameAs 2913053403 @default.
• W2913053403 citedByCount "0" @default.
• W2913053403 crossrefType "journal-article" @default.
• W2913053403 hasAuthorship W2913053403A5067314445 @default.
• W2913053403 hasConcept C142724271 @default.
• W2913053403 hasConcept C196822366 @default.
• W2913053403 hasConcept C2989005 @default.
• W2913053403 hasConcept C71924100 @default.
• W2913053403 hasConceptScore W2913053403C142724271 @default.
• W2913053403 hasConceptScore W2913053403C196822366 @default.
• W2913053403 hasConceptScore W2913053403C2989005 @default.
• W2913053403 hasConceptScore W2913053403C71924100 @default.
• W2913053403 hasIssue "1" @default.
• W2913053403 hasLocation W29130534031 @default.
• W2913053403 hasLocation W29130534032 @default.
• W2913053403 hasOpenAccess W2913053403 @default.
• W2913053403 hasPrimaryLocation W29130534031 @default.
• W2913053403 hasRelatedWork W1778826760 @default.
• W2913053403 hasRelatedWork W1995515455 @default.
• W2913053403 hasRelatedWork W2080531066 @default.
• W2913053403 hasRelatedWork W2095073570 @default.
• W2913053403 hasRelatedWork W2244323977 @default.
• W2913053403 hasRelatedWork W2368873087 @default.
• W2913053403 hasRelatedWork W2748952813 @default.
• W2913053403 hasRelatedWork W2899084033 @default.
• W2913053403 hasRelatedWork W3031052312 @default.
• W2913053403 hasRelatedWork W3032375762 @default.
• W2913053403 hasVolume "291" @default.
• W2913053403 isParatext "false" @default.
• W2913053403 isRetracted "false" @default.
• W2913053403 magId "2913053403" @default.
• W2913053403 workType "article" @default.