MR imaging-based radiomic signatures of distinct molecular subgroups of medulloblastoma

M. Iv, M. Zhou, K. Shpanskaya, S. Perreault, Z. Wang, E. Tranvinh, B. Lanzman, S. Vajapeyam, N. A. Vitanza, P. G. Fisher, Yoon-Jae Cho, S. Laughlin, V. Ramaswamy, M. D. Taylor, S. H. Cheshier, G. A. Grant, T. Young Poussaint, O. Gevaert, K. W. Yeom

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

BACKGROUND AND PURPOSE: Distinct molecular subgroups of pediatric medulloblastoma confer important differences in prognosis and therapy. Currently, tissue sampling is the only method to obtain information for classification. Our goal was to develop and validate radiomic and machine learning approaches for predicting molecular subgroups of pediatric medulloblastoma. MATERIALS AND METHODS: In this multi-institutional retrospective study, we evaluated MR imaging datasets of 109 pediatric patients with medulloblastoma from 3 children's hospitals from January 2001 to January 2014. A computational framework was developed to extract MR imaging- based radiomic features from tumor segmentations, and we tested 2 predictive models: a double 10-fold cross-validation using a combined dataset consisting of all 3 patient cohorts and a 3-dataset cross-validation, in which training was performed on 2 cohorts and testing was performed on the third independent cohort. We used the Wilcoxon rank sum test for feature selection with assessment of area under the receiver operating characteristic curve to evaluate model performance. RESULTS: Of 590 MR imaging- derived radiomic features, including intensity-based histograms, tumor edge-sharpness, Gabor features, and local area integral invariant features, extracted from imaging-derived tumor segmentations, tumor edge-sharpness was most useful for predicting sonic hedgehog and group 4 tumors. Receiver operating characteristic analysis revealed superior performance of the double 10-fold cross-validation model for predicting sonic hedgehog, group 3, and group 4 tumors when using combined T1- and T2-weighted images (area under the curve 0.79, 0.70, and 0.83, respectively). With the independent 3-dataset cross-validation strategy, select radiomic features were predictive of sonic hedgehog (area under the curve 0.70 - 0.73) and group 4 (area under the curve 0.76 - 0.80) medulloblastoma. CONCLUSIONS: This study provides proof-of-concept results for the application of radiomic and machine learning approaches to a multi-institutional dataset for the prediction of medulloblastoma subgroups.

Original languageEnglish (US)
Pages (from-to)154-161
Number of pages8
JournalAmerican Journal of Neuroradiology
Volume40
Issue number1
DOIs
StatePublished - Jan 1 2019

Fingerprint

Medulloblastoma
Area Under Curve
Neoplasms
Pediatrics
Nonparametric Statistics
ROC Curve
Retrospective Studies
Datasets

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Clinical Neurology

Cite this

Iv, M., Zhou, M., Shpanskaya, K., Perreault, S., Wang, Z., Tranvinh, E., ... Yeom, K. W. (2019). MR imaging-based radiomic signatures of distinct molecular subgroups of medulloblastoma. American Journal of Neuroradiology, 40(1), 154-161. https://doi.org/10.3174/ajnr.A5899

MR imaging-based radiomic signatures of distinct molecular subgroups of medulloblastoma. / Iv, M.; Zhou, M.; Shpanskaya, K.; Perreault, S.; Wang, Z.; Tranvinh, E.; Lanzman, B.; Vajapeyam, S.; Vitanza, N. A.; Fisher, P. G.; Cho, Yoon-Jae; Laughlin, S.; Ramaswamy, V.; Taylor, M. D.; Cheshier, S. H.; Grant, G. A.; Young Poussaint, T.; Gevaert, O.; Yeom, K. W.

In: American Journal of Neuroradiology, Vol. 40, No. 1, 01.01.2019, p. 154-161.

Research output: Contribution to journalArticle

Iv, M, Zhou, M, Shpanskaya, K, Perreault, S, Wang, Z, Tranvinh, E, Lanzman, B, Vajapeyam, S, Vitanza, NA, Fisher, PG, Cho, Y-J, Laughlin, S, Ramaswamy, V, Taylor, MD, Cheshier, SH, Grant, GA, Young Poussaint, T, Gevaert, O & Yeom, KW 2019, 'MR imaging-based radiomic signatures of distinct molecular subgroups of medulloblastoma', American Journal of Neuroradiology, vol. 40, no. 1, pp. 154-161. https://doi.org/10.3174/ajnr.A5899
Iv, M. ; Zhou, M. ; Shpanskaya, K. ; Perreault, S. ; Wang, Z. ; Tranvinh, E. ; Lanzman, B. ; Vajapeyam, S. ; Vitanza, N. A. ; Fisher, P. G. ; Cho, Yoon-Jae ; Laughlin, S. ; Ramaswamy, V. ; Taylor, M. D. ; Cheshier, S. H. ; Grant, G. A. ; Young Poussaint, T. ; Gevaert, O. ; Yeom, K. W. / MR imaging-based radiomic signatures of distinct molecular subgroups of medulloblastoma. In: American Journal of Neuroradiology. 2019 ; Vol. 40, No. 1. pp. 154-161.
@article{902824cf2012460498f566e30c20b34f,
title = "MR imaging-based radiomic signatures of distinct molecular subgroups of medulloblastoma",
abstract = "BACKGROUND AND PURPOSE: Distinct molecular subgroups of pediatric medulloblastoma confer important differences in prognosis and therapy. Currently, tissue sampling is the only method to obtain information for classification. Our goal was to develop and validate radiomic and machine learning approaches for predicting molecular subgroups of pediatric medulloblastoma. MATERIALS AND METHODS: In this multi-institutional retrospective study, we evaluated MR imaging datasets of 109 pediatric patients with medulloblastoma from 3 children's hospitals from January 2001 to January 2014. A computational framework was developed to extract MR imaging- based radiomic features from tumor segmentations, and we tested 2 predictive models: a double 10-fold cross-validation using a combined dataset consisting of all 3 patient cohorts and a 3-dataset cross-validation, in which training was performed on 2 cohorts and testing was performed on the third independent cohort. We used the Wilcoxon rank sum test for feature selection with assessment of area under the receiver operating characteristic curve to evaluate model performance. RESULTS: Of 590 MR imaging- derived radiomic features, including intensity-based histograms, tumor edge-sharpness, Gabor features, and local area integral invariant features, extracted from imaging-derived tumor segmentations, tumor edge-sharpness was most useful for predicting sonic hedgehog and group 4 tumors. Receiver operating characteristic analysis revealed superior performance of the double 10-fold cross-validation model for predicting sonic hedgehog, group 3, and group 4 tumors when using combined T1- and T2-weighted images (area under the curve 0.79, 0.70, and 0.83, respectively). With the independent 3-dataset cross-validation strategy, select radiomic features were predictive of sonic hedgehog (area under the curve 0.70 - 0.73) and group 4 (area under the curve 0.76 - 0.80) medulloblastoma. CONCLUSIONS: This study provides proof-of-concept results for the application of radiomic and machine learning approaches to a multi-institutional dataset for the prediction of medulloblastoma subgroups.",
author = "M. Iv and M. Zhou and K. Shpanskaya and S. Perreault and Z. Wang and E. Tranvinh and B. Lanzman and S. Vajapeyam and Vitanza, {N. A.} and Fisher, {P. G.} and Yoon-Jae Cho and S. Laughlin and V. Ramaswamy and Taylor, {M. D.} and Cheshier, {S. H.} and Grant, {G. A.} and {Young Poussaint}, T. and O. Gevaert and Yeom, {K. W.}",
year = "2019",
month = "1",
day = "1",
doi = "10.3174/ajnr.A5899",
language = "English (US)",
volume = "40",
pages = "154--161",
journal = "American Journal of Neuroradiology",
issn = "0195-6108",
publisher = "American Society of Neuroradiology",
number = "1",

}

TY - JOUR

T1 - MR imaging-based radiomic signatures of distinct molecular subgroups of medulloblastoma

AU - Iv, M.

AU - Zhou, M.

AU - Shpanskaya, K.

AU - Perreault, S.

AU - Wang, Z.

AU - Tranvinh, E.

AU - Lanzman, B.

AU - Vajapeyam, S.

AU - Vitanza, N. A.

AU - Fisher, P. G.

AU - Cho, Yoon-Jae

AU - Laughlin, S.

AU - Ramaswamy, V.

AU - Taylor, M. D.

AU - Cheshier, S. H.

AU - Grant, G. A.

AU - Young Poussaint, T.

AU - Gevaert, O.

AU - Yeom, K. W.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - BACKGROUND AND PURPOSE: Distinct molecular subgroups of pediatric medulloblastoma confer important differences in prognosis and therapy. Currently, tissue sampling is the only method to obtain information for classification. Our goal was to develop and validate radiomic and machine learning approaches for predicting molecular subgroups of pediatric medulloblastoma. MATERIALS AND METHODS: In this multi-institutional retrospective study, we evaluated MR imaging datasets of 109 pediatric patients with medulloblastoma from 3 children's hospitals from January 2001 to January 2014. A computational framework was developed to extract MR imaging- based radiomic features from tumor segmentations, and we tested 2 predictive models: a double 10-fold cross-validation using a combined dataset consisting of all 3 patient cohorts and a 3-dataset cross-validation, in which training was performed on 2 cohorts and testing was performed on the third independent cohort. We used the Wilcoxon rank sum test for feature selection with assessment of area under the receiver operating characteristic curve to evaluate model performance. RESULTS: Of 590 MR imaging- derived radiomic features, including intensity-based histograms, tumor edge-sharpness, Gabor features, and local area integral invariant features, extracted from imaging-derived tumor segmentations, tumor edge-sharpness was most useful for predicting sonic hedgehog and group 4 tumors. Receiver operating characteristic analysis revealed superior performance of the double 10-fold cross-validation model for predicting sonic hedgehog, group 3, and group 4 tumors when using combined T1- and T2-weighted images (area under the curve 0.79, 0.70, and 0.83, respectively). With the independent 3-dataset cross-validation strategy, select radiomic features were predictive of sonic hedgehog (area under the curve 0.70 - 0.73) and group 4 (area under the curve 0.76 - 0.80) medulloblastoma. CONCLUSIONS: This study provides proof-of-concept results for the application of radiomic and machine learning approaches to a multi-institutional dataset for the prediction of medulloblastoma subgroups.

AB - BACKGROUND AND PURPOSE: Distinct molecular subgroups of pediatric medulloblastoma confer important differences in prognosis and therapy. Currently, tissue sampling is the only method to obtain information for classification. Our goal was to develop and validate radiomic and machine learning approaches for predicting molecular subgroups of pediatric medulloblastoma. MATERIALS AND METHODS: In this multi-institutional retrospective study, we evaluated MR imaging datasets of 109 pediatric patients with medulloblastoma from 3 children's hospitals from January 2001 to January 2014. A computational framework was developed to extract MR imaging- based radiomic features from tumor segmentations, and we tested 2 predictive models: a double 10-fold cross-validation using a combined dataset consisting of all 3 patient cohorts and a 3-dataset cross-validation, in which training was performed on 2 cohorts and testing was performed on the third independent cohort. We used the Wilcoxon rank sum test for feature selection with assessment of area under the receiver operating characteristic curve to evaluate model performance. RESULTS: Of 590 MR imaging- derived radiomic features, including intensity-based histograms, tumor edge-sharpness, Gabor features, and local area integral invariant features, extracted from imaging-derived tumor segmentations, tumor edge-sharpness was most useful for predicting sonic hedgehog and group 4 tumors. Receiver operating characteristic analysis revealed superior performance of the double 10-fold cross-validation model for predicting sonic hedgehog, group 3, and group 4 tumors when using combined T1- and T2-weighted images (area under the curve 0.79, 0.70, and 0.83, respectively). With the independent 3-dataset cross-validation strategy, select radiomic features were predictive of sonic hedgehog (area under the curve 0.70 - 0.73) and group 4 (area under the curve 0.76 - 0.80) medulloblastoma. CONCLUSIONS: This study provides proof-of-concept results for the application of radiomic and machine learning approaches to a multi-institutional dataset for the prediction of medulloblastoma subgroups.

UR - http://www.scopus.com/inward/record.url?scp=85059897735&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85059897735&partnerID=8YFLogxK

U2 - 10.3174/ajnr.A5899

DO - 10.3174/ajnr.A5899

M3 - Article

C2 - 30523141

AN - SCOPUS:85059897735

VL - 40

SP - 154

EP - 161

JO - American Journal of Neuroradiology

JF - American Journal of Neuroradiology

SN - 0195-6108

IS - 1

ER -