{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,10,24]],"date-time":"2025-10-24T08:21:13Z","timestamp":1761294073005,"version":"3.37.3"},"reference-count":54,"publisher":"Oxford University Press (OUP)","issue":"11","license":[{"start":{"date-parts":[[2020,2,26]],"date-time":"2020-02-26T00:00:00Z","timestamp":1582675200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/academic.oup.com\/journals\/pages\/open_access\/funder_policies\/chorus\/standard_publication_model"}],"funder":[{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["F31CA221061","T32EB016640","R01CA157553"],"award-info":[{"award-number":["F31CA221061","T32EB016640","R01CA157553"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000001","name":"National Science Foundation","doi-asserted-by":"publisher","award":["1722516"],"award-info":[{"award-number":["1722516"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2020,6,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:sec>\n                  <jats:title>Motivation<\/jats:title>\n                  <jats:p>Cancer heterogeneity is observed at multiple biological levels. To improve our understanding of these differences and their relevance in medicine, approaches to link organ- and tissue-level information from diagnostic images and cellular-level information from genomics are needed. However, these \u2018radiogenomic\u2019 studies often use linear or shallow models, depend on feature selection, or consider one gene at a time to map images to genes. Moreover, no study has systematically attempted to understand the molecular basis of imaging traits based on the interpretation of what the neural network has learned. These studies are thus limited in their ability to understand the transcriptomic drivers of imaging traits, which could provide additional context for determining clinical outcomes.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Results<\/jats:title>\n                  <jats:p>We present a neural network-based approach that takes high-dimensional gene expression data as input and performs non-linear mapping to an imaging trait. To interpret the models, we propose gene masking and gene saliency to extract learned relationships from radiogenomic neural networks. In glioblastoma patients, our models outperformed comparable classifiers (&amp;gt;0.10 AUC) and our interpretation methods were validated using a similar model to identify known relationships between genes and molecular subtypes. We found that tumor imaging traits had specific transcription patterns, e.g. edema and genes related to cellular invasion, and 10 radiogenomic traits were significantly predictive of survival. We demonstrate that neural networks can model transcriptomic heterogeneity to reflect differences in imaging and can be used to derive radiogenomic traits with clinical value.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Availability and implementation<\/jats:title>\n                  <jats:p>https:\/\/github.com\/novasmedley\/deepRadiogenomics.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Contact<\/jats:title>\n                  <jats:p>whsu@mednet.ucla.edu<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Supplementary information<\/jats:title>\n                  <jats:p>Supplementary data are available at Bioinformatics online.<\/jats:p>\n               <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btaa126","type":"journal-article","created":{"date-parts":[[2020,2,20]],"date-time":"2020-02-20T12:53:15Z","timestamp":1582203195000},"page":"3537-3548","source":"Crossref","is-referenced-by-count":9,"title":["Discovering and interpreting transcriptomic drivers of imaging traits using neural networks"],"prefix":"10.1093","volume":"36","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0581-7477","authenticated-orcid":false,"given":"Nova F","family":"Smedley","sequence":"first","affiliation":[{"name":"Medical & Imaging Informatics"},{"name":"Department of Radiological Sciences"},{"name":"Department of Bioengineering"}]},{"given":"Suzie","family":"El-Saden","sequence":"additional","affiliation":[{"name":"Medical & Imaging Informatics"},{"name":"Department of Radiological Sciences"}]},{"given":"William","family":"Hsu","sequence":"additional","affiliation":[{"name":"Medical & Imaging Informatics"},{"name":"Department of Radiological Sciences"},{"name":"Department of Bioengineering"},{"name":"Bioinformatics IDP , University of California Los Angeles, Los Angeles, CA 90024, USA"}]}],"member":"286","published-online":{"date-parts":[[2020,2,26]]},"reference":[{"key":"2023062312015626600_btaa126-B1","doi-asserted-by":"crossref","first-page":"4006","DOI":"10.1038\/ncomms5006","article-title":"Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach","volume":"5","author":"Aerts","year":"2014","journal-title":"Nat. Commun"},{"key":"2023062312015626600_btaa126-B2","doi-asserted-by":"crossref","first-page":"D8","DOI":"10.1093\/nar\/gkx1095","article-title":"Database resources of the National Center for Biotechnology Information","volume":"46","author":"Agarwala","year":"2018","journal-title":"Nucleic Acids Res"},{"key":"2023062312015626600_btaa126-B3","doi-asserted-by":"crossref","DOI":"10.1561\/9781601982957","volume-title":"Learning Deep Architectures for AI","author":"Bengio","year":"2009"},{"key":"2023062312015626600_btaa126-B4","doi-asserted-by":"crossref","first-page":"8690","DOI":"10.18632\/oncotarget.2366","article-title":"Intracellular and extracellular domains of protein tyrosine phosphatase PTPRZ-B differentially regulate glioma cell growth and motility","volume":"5","author":"Bourgonje","year":"2014","journal-title":"Oncotarget"},{"key":"2023062312015626600_btaa126-B5","doi-asserted-by":"crossref","first-page":"1073","DOI":"10.1158\/1078-0432.CCR-17-2236","article-title":"Residual convolutional neural network for the determination of IDH status in low- and high-grade gliomas from MR imaging","volume":"24","author":"Chang","year":"2018","journal-title":"Clin. Cancer Res"},{"year":"2016","author":"Chen","key":"2023062312015626600_btaa126-B6"},{"key":"2023062312015626600_btaa126-B7","doi-asserted-by":"crossref","first-page":"1832","DOI":"10.1093\/bioinformatics\/btw074","article-title":"Gene expression inference with deep learning","volume":"32","author":"Chen","year":"2016","journal-title":"Bioinformatics"},{"year":"2015","author":"Chollet","key":"2023062312015626600_btaa126-B8"},{"key":"2023062312015626600_btaa126-B9","doi-asserted-by":"crossref","first-page":"30","DOI":"10.1186\/1755-8794-7-30","article-title":"Imaging genomic mapping of an invasive MRI phenotype predicts patient outcome and metabolic dysfunction: a TCGA Glioma Phenotype Research Group project","volume":"7","author":"Colen","year":"2014","journal-title":"BMC Med. Genomics"},{"key":"2023062312015626600_btaa126-B10","doi-asserted-by":"crossref","first-page":"7285","DOI":"10.1073\/pnas.1507125112","article-title":"A survey of human brain transcriptome diversity at the single cell level","volume":"112","author":"Darmanis","year":"2015","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"2023062312015626600_btaa126-B11","doi-asserted-by":"crossref","first-page":"5213","DOI":"10.1073\/pnas.0801279105","article-title":"Identification of noninvasive imaging surrogates for brain tumor gene-expression modules","volume":"105","author":"Diehn","year":"2008","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"2023062312015626600_btaa126-B12","doi-asserted-by":"crossref","first-page":"387","DOI":"10.1148\/radiol.12111607","article-title":"Non-small cell lung cancer: identifying prognostic imaging biomarkers by leveraging public gene expression microarray data-methods and preliminary results","volume":"264","author":"Gevaert","year":"2012","journal-title":"Radiology"},{"key":"2023062312015626600_btaa126-B13","doi-asserted-by":"crossref","first-page":"168","DOI":"10.1148\/radiol.14131731","article-title":"Glioblastoma multiforme: exploratory radiogenomic analysis by using quantitative image features","volume":"273","author":"Gevaert","year":"2014","journal-title":"Radiology"},{"key":"2023062312015626600_btaa126-B14","doi-asserted-by":"crossref","first-page":"41674","DOI":"10.1038\/srep41674","article-title":"Predictive radiogenomics modeling of EGFR mutation status in lung cancer","volume":"7","author":"Gevaert","year":"2017","journal-title":"Sci. Rep"},{"volume-title":"Deep Learning","year":"2016","author":"Goodfellow","key":"2023062312015626600_btaa126-B15"},{"key":"2023062312015626600_btaa126-B16","doi-asserted-by":"crossref","DOI":"10.7554\/eLife.23421","article-title":"Defining the biological basis of radiomic phenotypes in lung cancer","volume":"6","author":"Grossmann","year":"2017","journal-title":"Elife"},{"key":"2023062312015626600_btaa126-B17","doi-asserted-by":"crossref","first-page":"041007","DOI":"10.1117\/1.JMI.2.4.041007","article-title":"Prediction of clinical phenotypes in invasive breast carcinomas from the integration of radiomics and genomics data","volume":"2","author":"Guo","year":"2015","journal-title":"J. Med. Imaging"},{"key":"2023062312015626600_btaa126-B18","doi-asserted-by":"crossref","first-page":"560","DOI":"10.1148\/radiol.13120118","article-title":"MR imaging predictors of molecular profile and survival: multi-institutional study of the TCGA glioblastoma data set","volume":"267","author":"Gutman","year":"2013","journal-title":"Radiology"},{"key":"2023062312015626600_btaa126-B19","doi-asserted-by":"crossref","first-page":"276","DOI":"10.1007\/s10278-019-00179-2","article-title":"Predicting breast cancer molecular subtype with MRI dataset utilizing convolutional neural network algorithm","volume":"32","author":"Ha","year":"2019","journal-title":"J. Digit. Imaging"},{"key":"2023062312015626600_btaa126-B20","doi-asserted-by":"crossref","first-page":"128","DOI":"10.1093\/neuonc\/now135","article-title":"Radiogenomics to characterize regional genetic heterogeneity in glioblastoma","volume":"19","author":"Hu","year":"2017","journal-title":"Neuro. Oncol"},{"key":"2023062312015626600_btaa126-B21","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1148\/radiol.13130078","article-title":"Illuminating radiogenomic characteristics of glioblastoma multiforme through Integration of MR imaging, messenger RNA expression, and DNA copy number variation","volume":"270","author":"Jamshidi","year":"2014","journal-title":"Radiology"},{"key":"2023062312015626600_btaa126-B22","doi-asserted-by":"crossref","first-page":"907","DOI":"10.1148\/radiol.2016161382","article-title":"Radiogenomics of glioblastoma: machine learning-based classification of molecular characteristics by using multiparametric and multiregional MR imaging features","volume":"281","author":"Kickingereder","year":"2016","journal-title":"Radiology"},{"key":"2023062312015626600_btaa126-B23","doi-asserted-by":"crossref","first-page":"622","DOI":"10.1007\/s10278-017-0009-z","article-title":"Residual deep convolutional neural network predicts MGMT methylation status","volume":"30","author":"Korfiatis","year":"2017","journal-title":"J. Digit. Imaging"},{"year":"2017","author":"Kotikalapudi","key":"2023062312015626600_btaa126-B24"},{"key":"2023062312015626600_btaa126-B25","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"Lecun","year":"2015","journal-title":"Nature"},{"year":"2018","author":"Li","key":"2023062312015626600_btaa126-B26"},{"key":"2023062312015626600_btaa126-B27","doi-asserted-by":"crossref","first-page":"1739","DOI":"10.1093\/bioinformatics\/btr260","article-title":"Molecular signatures database (MSigDB) 3.0","volume":"27","author":"Liberzon","year":"2011","journal-title":"Bioinformatics"},{"key":"2023062312015626600_btaa126-B28","doi-asserted-by":"crossref","first-page":"417","DOI":"10.1016\/j.cels.2015.12.004","article-title":"The molecular signatures database hallmark gene set collection","volume":"1","author":"Liberzon","year":"2015","journal-title":"Cell Syst"},{"key":"2023062312015626600_btaa126-B29","doi-asserted-by":"crossref","first-page":"1061","DOI":"10.1038\/nature07385","article-title":"Comprehensive genomic characterization defines human glioblastoma genes and core pathways","volume":"455","author":"McLendon","year":"2008","journal-title":"Nature"},{"key":"2023062312015626600_btaa126-B30","doi-asserted-by":"crossref","first-page":"33257","DOI":"10.18632\/oncotarget.8921","article-title":"c-MYC inhibition impairs hypoxia response in glioblastoma multiforme","volume":"7","author":"Mongiardi","year":"2016","journal-title":"Oncotarget"},{"key":"2023062312015626600_btaa126-B31","doi-asserted-by":"crossref","DOI":"10.1074\/mcp.M111.014647","article-title":"The matrisome: in silico definition and in vivo characterization by proteomics of normal and tumor extracellular matrices","volume":"11","author":"Naba","year":"2012","journal-title":"Mol. Cell. Proteomics"},{"key":"2023062312015626600_btaa126-B32","doi-asserted-by":"crossref","first-page":"626","DOI":"10.1093\/neuonc\/not008","article-title":"Identifying the mesenchymal molecular subtype of glioblastoma using quantitative volumetric analysis of anatomic magnetic resonance images","volume":"15","author":"Naeini","year":"2013","journal-title":"Neuro. Oncol"},{"key":"2023062312015626600_btaa126-B33","doi-asserted-by":"crossref","first-page":"1445","DOI":"10.2353\/ajpath.2007.070011","article-title":"Genetic pathways to primary and secondary glioblastoma","volume":"170","author":"Ohgaki","year":"2007","journal-title":"Am. J. Pathol"},{"key":"2023062312015626600_btaa126-B34","doi-asserted-by":"crossref","first-page":"1842","DOI":"10.1001\/jama.2013.280319","article-title":"Glioblastoma and other malignant gliomas","volume":"310","author":"Omuro","year":"2013","journal-title":"JAMA"},{"key":"2023062312015626600_btaa126-B35","doi-asserted-by":"crossref","first-page":"1807","DOI":"10.1126\/science.1164382","article-title":"An integrated genomic analysis of human glioblastoma multiforme","volume":"321","author":"Parsons","year":"2008","journal-title":"Science"},{"key":"2023062312015626600_btaa126-B36","doi-asserted-by":"crossref","first-page":"1396","DOI":"10.1126\/science.1254257","article-title":"Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma","volume":"344","author":"Patel","year":"2014","journal-title":"Science"},{"key":"2023062312015626600_btaa126-B37","first-page":"2825","article-title":"Scikit-learn: machine learning in Python","volume":"12","author":"Pedregosa","year":"2011","journal-title":"J. Mach. Learn. Res"},{"key":"2023062312015626600_btaa126-B38","doi-asserted-by":"crossref","first-page":"1458","DOI":"10.1093\/brain\/aww046","article-title":"Corticosteroids compromise survival in glioblastoma","volume":"139","author":"Pitter","year":"2016","journal-title":"Brain"},{"key":"2023062312015626600_btaa126-B39","first-page":"2466","article-title":"MR imaging correlates of survival in patients with high-grade gliomas","volume":"26","author":"Pope","year":"2005","journal-title":"Am. J. Neuroradiol"},{"key":"2023062312015626600_btaa126-B40","doi-asserted-by":"crossref","first-page":"268","DOI":"10.1148\/radiol.2491072000","article-title":"Relationship between gene expression and enhancement in glioblastoma multiforme: exploratory DNA microarray analysis","volume":"249","author":"Pope","year":"2008","journal-title":"Radiology"},{"key":"2023062312015626600_btaa126-B41","doi-asserted-by":"crossref","first-page":"660","DOI":"10.1126\/science.aaf2666","article-title":"An anatomic transcriptional atlas of human glioblastoma","volume":"360","author":"Puchalski","year":"2018","journal-title":"Science"},{"year":"2016","author":"Scarpace","key":"2023062312015626600_btaa126-B42"},{"key":"2023062312015626600_btaa126-B43","doi-asserted-by":"crossref","first-page":"675","DOI":"10.1038\/nbt1306","article-title":"Decoding global gene expression programs in liver cancer by noninvasive imaging","volume":"25","author":"Segal","year":"2007","journal-title":"Nat. Biotechnol"},{"year":"2016","author":"Sergushichev","key":"2023062312015626600_btaa126-B44"},{"year":"2014","author":"Simonyan","key":"2023062312015626600_btaa126-B45"},{"key":"2023062312015626600_btaa126-B46","doi-asserted-by":"crossref","first-page":"15545","DOI":"10.1073\/pnas.0506580102","article-title":"Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles","volume":"102","author":"Subramanian","year":"2005","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"2023062312015626600_btaa126-B48","doi-asserted-by":"crossref","first-page":"98","DOI":"10.1016\/j.ccr.2009.12.020","article-title":"Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1","volume":"17","author":"Verhaak","year":"2010","journal-title":"Cancer Cell"},{"key":"2023062312015626600_btaa126-B49","doi-asserted-by":"crossref","first-page":"58","DOI":"10.3174\/ajnr.A4491","article-title":"MR imaging\u2013based analysis of glioblastoma multiforme: estimation of IDH1 mutation status","volume":"37","author":"Yamashita","year":"2016","journal-title":"Am. J. Neuroradiol"},{"first-page":"818","year":"2014","author":"Zeiler","key":"2023062312015626600_btaa126-B50"},{"key":"2023062312015626600_btaa126-B51","doi-asserted-by":"crossref","first-page":"109","DOI":"10.1093\/neuonc\/now121","article-title":"Multimodal MRI features predict isocitrate dehydrogenase genotype in high-grade gliomas","volume":"19","author":"Zhang","year":"2017","journal-title":"Neuro. Oncol"},{"key":"2023062312015626600_btaa126-B52","doi-asserted-by":"crossref","first-page":"37","DOI":"10.1016\/j.neuron.2015.11.013","article-title":"Purification and characterization of progenitor and mature human astrocytes reveals transcriptional and functional differences with mouse","volume":"89","author":"Zhang","year":"2016","journal-title":"Neuron"},{"key":"2023062312015626600_btaa126-B53","doi-asserted-by":"crossref","first-page":"17787","DOI":"10.1038\/srep17787","article-title":"Deciphering genomic underpinnings of quantitative MRI-based radiomic phenotypes of invasive breast carcinoma","volume":"5","author":"Zhu","year":"2015","journal-title":"Sci. Rep"},{"key":"2023062312015626600_btaa126-B54","doi-asserted-by":"crossref","first-page":"e25451","DOI":"10.1371\/journal.pone.0025451","article-title":"Radiogenomic mapping of edema\/cellular invasion MRI-phenotypes in glioblastoma multiforme","volume":"6","author":"Zinn","year":"2011","journal-title":"PLoS One"},{"key":"2023062312015626600_btaa126-B55","doi-asserted-by":"crossref","first-page":"6288","DOI":"10.1158\/1078-0432.CCR-17-3420","article-title":"A coclinical radiogenomic validation study: conserved magnetic resonance radiomic appearance of periostin-expressing glioblastoma in patients and xenograft models","volume":"24","author":"Zinn","year":"2018","journal-title":"Clin. Cancer Res"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/academic.oup.com\/bioinformatics\/advance-article-pdf\/doi\/10.1093\/bioinformatics\/btaa126\/33125574\/btaa126.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/36\/11\/3537\/50670871\/bioinformatics_36_11_3537.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/36\/11\/3537\/50670871\/bioinformatics_36_11_3537.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,6,24]],"date-time":"2023-06-24T21:17:35Z","timestamp":1687641455000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/36\/11\/3537\/5758261"}},"subtitle":[],"editor":[{"given":"Alfonso","family":"Valencia","sequence":"additional","affiliation":[]}],"short-title":[],"issued":{"date-parts":[[2020,2,26]]},"references-count":54,"journal-issue":{"issue":"11","published-print":{"date-parts":[[2020,6,1]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btaa126","relation":{},"ISSN":["1367-4803","1367-4811"],"issn-type":[{"type":"print","value":"1367-4803"},{"type":"electronic","value":"1367-4811"}],"subject":[],"published-other":{"date-parts":[[2020,6]]},"published":{"date-parts":[[2020,2,26]]}}}