DOI QR코드

DOI QR Code

Radiomics and Deep Learning from Research to Clinical Workflow: Neuro-Oncologic Imaging

  • Ji Eun Park (Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center) ;
  • Philipp Kickingereder (Department of Neuroradiology, University of Heidelberg) ;
  • Ho Sung Kim (Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center)
  • Received : 2019.11.10
  • Accepted : 2020.03.29
  • Published : 2020.10.01

Abstract

Imaging plays a key role in the management of brain tumors, including the diagnosis, prognosis, and treatment response assessment. Radiomics and deep learning approaches, along with various advanced physiologic imaging parameters, hold great potential for aiding radiological assessments in neuro-oncology. The ongoing development of new technology needs to be validated in clinical trials and incorporated into the clinical workflow. However, none of the potential neuro-oncological applications for radiomics and deep learning has yet been realized in clinical practice. In this review, we summarize the current applications of radiomics and deep learning in neuro-oncology and discuss challenges in relation to evidence-based medicine and reporting guidelines, as well as potential applications in clinical workflows and routine clinical practice.

Keywords

Acknowledgement

This research was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIP) (grant number: NRF-2020R1A2B5B01001707 and NRF-2020R1A2C4001748).

References

  1. Kumar V, Gu Y, Basu S, Berglund A, Eschrich SA, Schabath MB, et al. Radiomics: the process and the challenges. Magn Reson Imaging 2012;30:1234-1248 https://doi.org/10.1016/j.mri.2012.06.010
  2. Lambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer 2012;48:441-446 https://doi.org/10.1016/j.ejca.2011.11.036
  3. Lee JG, Jun S, Cho YW, Lee H, Kim GB, Seo JB, et al. Deep learning in medical imaging: general overview. Korean J Radiol 2017;18:570-584 https://doi.org/10.3348/kjr.2017.18.4.570
  4. Topol E. Deep medicine: how artificial intelligence can make healthcare human again, 1st ed. NewYork: Basic Books, 2019
  5. Gillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures, they are data. Radiology 2016;278:563-577 https://doi.org/10.1148/radiol.2015151169
  6. Kim JY, Park JE, Jo Y, Shim WH, Nam SJ, Kim JH, et al. Incorporating diffusion- and perfusion-weighted MRI into a radiomics model improves diagnostic performance for pseudoprogression in glioblastoma patients. Neuro Oncol 2019;21:404-414 https://doi.org/10.1093/neuonc/noy133
  7. Kickingereder P, Gotz M, Muschelli J, Wick A, Neuberger U, Shinohara RT, et al. Large-scale radiomic profiling of recurrent glioblastoma identifies an imaging predictor for stratifying anti-angiogenic treatment response. Clin Cancer Res 2016;22:5765-5771 https://doi.org/10.1158/1078-0432.CCR-16-0702
  8. Grossmann P, Narayan V, Chang K, Rahman R, Abrey L, Reardon DA, et al. Quantitative imaging biomarkers for risk stratification of patients with recurrent glioblastoma treated with bevacizumab. Neuro Oncol 2017;19:1688-1697 https://doi.org/10.1093/neuonc/nox092
  9. Bhatia A, Birger M, Veeraraghavan H, Um H, Tixier F, McKenney AS, et al. MRI radiomic features are associated with survival in melanoma brain metastases treated with immune checkpoint inhibitors. Neuro Oncol 2019;21:1578-1586 https://doi.org/10.1093/neuonc/noz141
  10. Artzi M, Bressler I, Ben Bashat D. Differentiation between glioblastoma, brain metastasis and subtypes using radiomics analysis. J Magn Reson Imaging 2019;50:519-528 https://doi.org/10.1002/jmri.26643
  11. Kang D, Park JE, Kim YH, Kim JH, Oh JY, Kim J, et al. Diffusion radiomics as a diagnostic model for atypical manifestation of primary central nervous system lymphoma: development and multicenter external validation. Neuro Oncol 2018;20:1251-1261 https://doi.org/10.1093/neuonc/noy021
  12. Kniep HC, Madesta F, Schneider T, Hanning U, Schonfeld MH, Schon G, et al. Radiomics of brain MRI: utility in prediction of metastatic tumor type. Radiology 2019;290:479-487 https://doi.org/10.1148/radiol.2018180946
  13. Thakkar JP, Dolecek TA, Horbinski C, Ostrom QT, Lightner DD, Barnholtz-Sloan JS, et al. Epidemiologic and molecular prognostic review of glioblastoma. Cancer Epidemio Biomarkers Prev 2014;23:1985-1996 https://doi.org/10.1158/1055-9965.EPI-14-0275
  14. Kickingereder P, Burth S, Wick A, Gotz M, Eidel O, Schlemmer HP, et al. Radiomic profiling of glioblastoma: identifying an imaging predictor of patient survival with improved performance over established clinical and radiologic risk models. Radiology 2016;280:880-889 https://doi.org/10.1148/radiol.2016160845
  15. Bae S, Choi YS, Ahn SS, Chang JH, Kang SG, Kim EH, et al. Radiomic MRI phenotyping of glioblastoma: improving survival prediction. Radiology 2018;289:797-806 https://doi.org/10.1148/radiol.2018180200
  16. Kickingereder P, Neuberger U, Bonekamp D, Piechotta PL, Gotz M, Wick A, et al. Radiomic subtyping improves disease stratification beyond key molecular, clinical, and standard imaging characteristics in patients with glioblastoma. Neuro Oncol 2018;20:848-857 https://doi.org/10.1093/neuonc/nox188
  17. Li Q, Bai H, Chen Y, Sun Q, Liu L, Zhou S, et al. A fully-automatic multiparametric radiomics model: towards reproducible and prognostic imaging signature for prediction of overall survival in glioblastoma multiforme. Sci Rep 2017;7:14331
  18. Pan SJ, Yang Q. A survey on transfer learning. IEEE Trans Knowl Data Eng 2010;22:1345-1359 https://doi.org/10.1109/TKDE.2009.191
  19. Lao J, Chen Y, Li ZC, Li Q, Zhang J, Liu J, et al. A deep learning-based radiomics model for prediction of survival in glioblastoma multiforme. Sci Rep 2017;7:10353
  20. Nie D, Zhang H, Adeli E, Liu L, Shen D. 3D deep learning for multi-modal imaging-guided survival time prediction of brain tumor patients. Med Image Comput Comput Assist Interv 2016;9901:212-220 https://doi.org/10.1007/978-3-319-46723-8_25
  21. Hu X, Wong KK, Young GS, Guo L, Wong ST. Support vector machine multiparametric MRI identification of pseudoprogression from tumor recurrence in patients with resected glioblastoma. J Magn Reson Imaging 2011;33:296-305 https://doi.org/10.1002/jmri.22432
  22. Park JE, Kim HS, Goh MJ, Kim SJ, Kim JH. Pseudoprogression in patients with glioblastoma: assessment by using volume-weighted voxel-based multiparametric clustering of MR imaging data in an independent test set. Radiology 2015;275:792-802 https://doi.org/10.1148/radiol.14141414
  23. Elshafeey N, Kotrotsou A, Hassan A, Elshafei N, Hassan I, Ahmed S, et al. Multicenter study demonstrates radiomic features derived from magnetic resonance perfusion images identify pseudoprogression in glioblastoma. Nat Commun 2019;10:3170
  24. Qian X, Tan H, Zhang J, Zhao W, Chan MD, Zhou X. Stratification of pseudoprogression and true progression of glioblastoma multiform based on longitudinal diffusion tensor imaging without segmentation. Med Phys 2016;43:5889
  25. Jang BS, Jeon SH, Kim IH, Kim IA. Prediction of pseudoprogression versus progression using machine learning algorithm in glioblastoma. Scientific Reports 2018;8:12516
  26. Tang A, Tam R, Cadrin-Chenevert A, Guest W, Chong J, Barfett J, et al. Canadian association of radiologists white paper on artificial intelligence in radiology. Can Assoc Radiol J 2018;69:120-135 https://doi.org/10.1016/j.carj.2018.02.002
  27. Park SH, Han K. Methodologic guide for evaluating clinical performance and effect of artificial intelligence technology for medical diagnosis and prediction. Radiology 2018;286:800-809 https://doi.org/10.1148/radiol.2017171920
  28. Guyatt GH, Tugwell PX, Feeny DH, Drummond MF, Haynes RB. The role of before-after studies of therapeutic impact in the evaluation of diagnostic technologies. J Chronic Dis 1986;39:295-304 https://doi.org/10.1016/0021-9681(86)90051-2
  29. Knottnerus JA, Buntinx F. The evidence base of clinical diagnosis: theory and methods of diagnostic research, 2nd ed. London: BMJ Books, 2011
  30. Park JE, Kim D, Kim HS, Park SY, Kim JY, Cho SJ, et al. Quality of science and reporting of radiomics in oncologic studies: room for improvement according to radiomics quality score and TRIPOD statement. Eur Radiol 2020;30:523-536 https://doi.org/10.1007/s00330-019-06360-z
  31. Kim DW, Jang HY, Kim KW, Shin Y, Park SH. Design characteristics of studies reporting the performance of artificial intelligence algorithms for diagnostic analysis of medical images: results from recently published papers. Korean J Radiol 2019;20:405-410 https://doi.org/10.3348/kjr.2019.0025
  32. Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162:W1-W73 https://doi.org/10.7326/M14-0698
  33. Collins GS, Moons KGM. Reporting of artificial intelligence prediction models. Lancet 2019;393:1577-1579 https://doi.org/10.1016/S0140-6736(19)30037-6
  34. Bossuyt PM, Irwig L, Craig J, Glasziou P. Comparative accuracy: assessing new tests against existing diagnostic pathways. BMJ 2006;332:1089-1092 https://doi.org/10.1136/bmj.332.7549.1089
  35. Weston AD, Korfiatis P, Kline TL, Philbrick KA, Kostandy P, Sakinis T, et al. Automated abdominal segmentation of CT scans for body composition analysis using deep learning. Radiology 2019;290:669-679 https://doi.org/10.1148/radiol.2018181432
  36. Lin L, Dou Q, Jin YM, Zhou GQ, Tang YQ, Chen WL, et al. Deep learning for automated contouring of primary tumor volumes by MRI for nasopharyngeal carcinoma. Radiology 2019;291:677-686 https://doi.org/10.1148/radiol.2019182012
  37. Norman B, Pedoia V, Majumdar S. Use of 2D U-net convolutional neural networks for automated cartilage and meniscus segmentation of knee MR imaging data to determine relaxometry and morphometry. Radiology 2018;288:177-185 https://doi.org/10.1148/radiol.2018172322
  38. Parmar C, Rios Velazquez E, Leijenaar R, Jermoumi M, Carvalho S, Mak RH, et al. Robust radiomics feature quantification using semiautomatic volumetric segmentation. Plos One 2014 Jul 15 [Epub]. https://doi.org/10.1371/journal.pone.0102107
  39. Pavic M, Bogowicz M, Wurms X, Glatz S, Finazzi T, Riesterer O, et al. Influence of inter-observer delineation variability on radiomics stability in different tumor sites. Acta Oncol 2018;57:1070-1074 https://doi.org/10.1080/0284186X.2018.1445283
  40. Lambin P, Leijenaar RTH, Deist TM, Peerlings J, de Jong EEC, van Timmeren J, et al. Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol 2017;14:749-762 https://doi.org/10.1038/nrclinonc.2017.141
  41. Park JE, Kim HS. Radiomics as a quantitative imaging biomarker: practical considerations and the current standpoint in neuro-oncologic Studies. Nucl Med Mol Imaging 2018;52:99-108 https://doi.org/10.1007/s13139-017-0512-7
  42. Bakas S, Reyes M, Jakab A, Bauer S, Rempfler M, Crimi A, et al. Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the BRATS challenge [updated April 2019]. Cornell University, 2018. Available at: https://arxiv.org/abs/1811.02629. Accessed March 3, 2020
  43. Yang D. Standardized MRI assessment of high-grade glioma response: a review of the essential elements and pitfalls of the RANO criteria. Neurooncol Pract 2016;3:59-67 https://doi.org/10.1093/nop/npv023
  44. van den Bent MJ, Wefel JS, Schiff D, Taphoorn MJ, Jaeckle K, Junck L, et al. Response assessment in neuro-oncology (a report of the RANO group): assessment of outcome in trials of diffuse low-grade gliomas. Lancet Oncol 2011;12:583-593 https://doi.org/10.1016/S1470-2045(11)70057-2
  45. Wen PY, Macdonald DR, Reardon DA, Cloughesy TF, Sorensen AG, Galanis E, et al. Updated response assessment criteria for high-grade gliomas: response assessment in neuro-oncology working group. J Clin Oncol 2010;28:1963-1972 https://doi.org/10.1200/JCO.2009.26.3541
  46. Chow DS, Qi J, Guo X, Miloushev VZ, Iwamoto FM, Bruce JN, et al. Semiautomated volumetric measurement on postcontrast MR imaging for analysis of recurrent and residual disease in glioblastoma multiforme. AJNR Am J Neuroradiol 2014;35:498-503 https://doi.org/10.3174/ajnr.A3724
  47. Sorensen AG, Patel S, Harmath C, Bridges S, Synnott J, Sievers A, et al. Comparison of diameter and perimeter methods for tumor volume calculation. J Clin Oncol 2001;19:551-557 https://doi.org/10.1200/JCO.2001.19.2.551
  48. Korn RL, Crowley JJ. Overview: progression-free survival as an endpoint in clinical trials with solid tumors. Clin Cancer Res 2013;19:2607-2612 https://doi.org/10.1158/1078-0432.CCR-12-2934
  49. Kickingereder P, Isensee F, Tursunova I, Petersen J, Neuberger U, Bonekamp D, et al. Automated quantitative tumour response assessment of MRI in neuro-oncology with artificial neural networks: a multicentre, retrospective study. Lancet Oncol 2019;20:728-740 https://doi.org/10.1016/S1470-2045(19)30098-1
  50. Doi K. Computer-aided diagnosis in medical imaging: historical review, current status and future potential. Comput Med Imaging Graph 2007;31:198-211 https://doi.org/10.1016/j.compmedimag.2007.02.002
  51. Lehman CD, Wellman RD, Buist DS, Kerlikowske K, Tosteson AN, Miglioretti DL. Diagnostic accuracy of digital screening mammography with and without computer-aided detection. JAMA Intern Med 2015;175:1828-1837 https://doi.org/10.1001/jamainternmed.2015.5231
  52. Titano JJ, Badgeley M, Schefflein J, Pain M, Su A, Cai M, et al. Automated deep-neural-network surveillance of cranial images for acute neurologic events. Nat Med 2018;24:1337-1341 https://doi.org/10.1038/s41591-018-0147-y
  53. Ambrosini RD, Wang P, O'Dell WG. Computer-aided detection of metastatic brain tumors using automated three-dimensional template matching. J Magn Reson Imaging 2010;31:85-93 https://doi.org/10.1002/jmri.22009
  54. Perez-Ramirez U, Arana E, Moratal D. Brain metastases detection on MR by means of three-dimensional tumor-appearance template matching. J Magn Reson Imaging 2016;44:642-652 https://doi.org/10.1002/jmri.25207
  55. Jun Y, Eo T, Kim T, Shin H, Hwang D, Bae SH, et al. Deep-learned 3D black-blood imaging using automatic labelling technique and 3D convolutional neural networks for detecting metastatic brain tumors. Scientific Reports 2018;8:9450
  56. Louis DN, Perry A, Reifenberger G, von Deimling A, Figarella-Branger D, Cavenee WK, et al. The 2016 World Health Organization classification of tumors of the central nervous system: a summary. Acta Neuropathol 2016;131:803-820 https://doi.org/10.1007/s00401-016-1545-1
  57. Brandes AA, Franceschi E, Tosoni A, Blatt V, Pession A, Tallini G, et al. MGMT promoter methylation status can predict the incidence and outcome of pseudoprogression after concomitant radiochemotherapy in newly diagnosed glioblastoma patients. J Clin Oncol 2008;26:2192-2197 https://doi.org/10.1200/JCO.2007.14.8163
  58. Gan HK, Cvrljevic AN, Johns TG. The epidermal growth factor receptor variant III (EGFRvIII): where wild things are altered. FEBS J 2013;280:5350-5370 https://doi.org/10.1111/febs.12393
  59. Yu J, Shi Z, Lian Y, Li Z, Liu T, Gao Y, et al. Noninvasive IDH1 mutation estimation based on a quantitative radiomics approach for grade II glioma. Eur Radiol 2017;27:3509-3522 https://doi.org/10.1007/s00330-016-4653-3
  60. Zhang X, Tian Q, Wang L, Liu Y, Li B, Liang Z, et al. Radiomics strategy for molecular subtype stratification of lower-grade glioma: detecting IDH and TP53 mutations based on multimodal MRI. J Magn Reson Imaging 2018;48:916-926 https://doi.org/10.1002/jmri.25960
  61. Kim M, Jung SY, Park JE, Jo Y, Park SY, Nam SJ, et al. Diffusion- and perfusion-weighted MRI radiomics model may predict isocitrate dehydrogenase (IDH) mutation and tumor aggressiveness in diffuse lower grade glioma. Eur Radiol 2020;30:2142-2151 https://doi.org/10.1007/s00330-019-06548-3
  62. Arita H, Kinoshita M, Kawaguchi A, Takahashi M, Narita Y, Terakawa Y, et al. Lesion location implemented magnetic resonance imaging radiomics for predicting IDH and TERT promoter mutations in grade II/III gliomas. Sci Rep 2018;8:11773
  63. Gevaert O, Mitchell LA, Achrol AS, Xu J, Echegaray S, Steinberg GK, et al. Glioblastoma multiforme: exploratory radiogenomic analysis by using quantitative image features. Radiology 2014;273:168-174 https://doi.org/10.1148/radiol.14131731
  64. Eichinger P, Alberts E, Delbridge C, Trebeschi S, Valentinitsch A, Bette S, et al. Diffusion tensor image features predict IDH genotype in newly diagnosed WHO grade II/III gliomas. Sci Rep 2017;7:13396
  65. Zhou H, Vallieres M, Bai HX, Su C, Tang H, Oldridge D, et al. MRI features predict survival and molecular markers in diffuse lower-grade gliomas. Neuro Oncol 2017;19:862-870 https://doi.org/10.1093/neuonc/now256
  66. Chang K, Bai HX, Zhou H, Su C, Bi WL, Agbodza E, et al. Residual convolutional neural network for the determination of IDH status in low- and high-grade gliomas from MR imaging. Clin Cancer Res 2018;24:1073-1081 https://doi.org/10.1158/1078-0432.CCR-17-2236
  67. Chang P, Grinband J, Weinberg BD, Bardis M, Khy M, Cadena G, et al. Deep-learning convolutional neural networks accurately classify genetic mutations in gliomas. AJNR Am J Neuroradiol 2018;39:1201-1207 https://doi.org/10.3174/ajnr.A5667
  68. Han L, Kamdar MR. MRI to MGMT: predicting methylation status in glioblastoma patients using convolutional recurrent neural networks. Pac Symp Biocomput 2018;23:331-342 https://doi.org/10.1142/9789813235533_0031
  69. Liang S, Zhang R, Liang D, Song T, Ai T, Xia C, et al. Multimodal 3D DenseNet for IDH genotype prediction in gliomas. Genes (Basel) 2018;9:382
  70. Kickingereder P, Bonekamp D, Nowosielski M, Kratz A, Sill M, Burth S, et al. Radiogenomics of glioblastoma: machine learning-based classification of molecular characteristics by using multiparametric and multiregional MR imaging features. Radiology 2016;281:907-918 https://doi.org/10.1148/radiol.2016161382
  71. Hu LS, Ning S, Eschbacher JM, Baxter LC, Gaw N, Ranjbar S, et al. Radiogenomics to characterize regional genetic heterogeneity in glioblastoma. Neuro Oncol 2017;19:128-137 https://doi.org/10.1093/neuonc/now135
  72. Park JE, Kim HS, Park SY, Nam SJ, Chun SM, Jo Y, et al. Prediction of core signaling pathway by using diffusion-and perfusion-based MRI radiomics and next-generation sequencing in isocitrate dehydrogenase wild-type glioblastoma. Radiology 2020;294:388-397 https://doi.org/10.1148/radiol.2019190913
  73. Rathore S, Akbari H, Rozycki M, Abdullah KG, Nasrallah MP, Binder ZA, et al. Radiomic MRI signature reveals three distinct subtypes of glioblastoma with different clinical and molecular characteristics, offering prognostic value beyond IDH1. Sci Rep 2018;8:5087
  74. Marcus DS, Olsen TR, Ramaratnam M, Buckner RL. The extensible neuroimaging archive toolkit: an informatics platform for managing, exploring, and sharing neuroimaging data. Neuroinformatics 2007;5:11-34 https://doi.org/10.1385/NI:5:1:11
  75. Ravi S, Larochelle H. Optimization as a model for few-shot learning. OpenReview.net Web site. https://openreview.net/forum?id=rJY0-Kcll¬eId=ryq49XyLg. Published November 5, 2016. Accessed May 27, 2019