D Rodriguez Gutierrez1, A Awwad2, L Meijer3, M Manita4, T Jaspan5, R A Dineen2, R G Grundy1, D P Auer6. 1. From the Division of Radiological and Imaging Sciences (D.R.G., A.A., M.M., R.A.D., R.G.G., D.P.A.)Children's Brain Tumor Research Centre (D.R.G., L.M., R.G.G., D.P.A.), University of Nottingham, Nottingham, UK. 2. From the Division of Radiological and Imaging Sciences (D.R.G., A.A., M.M., R.A.D., R.G.G., D.P.A.)Nottingham University Hospital Trust (A.A., L.M., T.J., R.A.D.), Nottingham, UK. 3. Children's Brain Tumor Research Centre (D.R.G., L.M., R.G.G., D.P.A.), University of Nottingham, Nottingham, UKNottingham University Hospital Trust (A.A., L.M., T.J., R.A.D.), Nottingham, UK. 4. From the Division of Radiological and Imaging Sciences (D.R.G., A.A., M.M., R.A.D., R.G.G., D.P.A.). 5. Nottingham University Hospital Trust (A.A., L.M., T.J., R.A.D.), Nottingham, UK. 6. From the Division of Radiological and Imaging Sciences (D.R.G., A.A., M.M., R.A.D., R.G.G., D.P.A.) dorothee.auer@nottingham.ac.uk.
Abstract
BACKGROUND AND PURPOSE: Qualitative radiologic MR imaging review affords limited differentiation among types of pediatric posterior fossa brain tumors and cannot detect histologic or molecular subtypes, which could help to stratify treatment. This study aimed to improve current posterior fossa discrimination of histologic tumor type by using support vector machine classifiers on quantitative MR imaging features. MATERIALS AND METHODS: This retrospective study included preoperative MRI in 40 children with posterior fossa tumors (17 medulloblastomas, 16 pilocytic astrocytomas, and 7 ependymomas). Shape, histogram, and textural features were computed from contrast-enhanced T2WI and T1WI and diffusivity (ADC) maps. Combinations of features were used to train tumor-type-specific classifiers for medulloblastoma, pilocytic astrocytoma, and ependymoma types in separation and as a joint posterior fossa classifier. A tumor-subtype classifier was also produced for classic medulloblastoma. The performance of different classifiers was assessed and compared by using randomly selected subsets of training and test data. RESULTS: ADC histogram features (25th and 75th percentiles and skewness) yielded the best classification of tumor type (on average >95.8% of medulloblastomas, >96.9% of pilocytic astrocytomas, and >94.3% of ependymomas by using 8 training samples). The resulting joint posterior fossa classifier correctly assigned >91.4% of the posterior fossa tumors. For subtype classification, 89.4% of classic medulloblastomas were correctly classified on the basis of ADC texture features extracted from the Gray-Level Co-Occurence Matrix. CONCLUSIONS: Support vector machine-based classifiers using ADC histogram features yielded very good discrimination among pediatric posterior fossa tumor types, and ADC textural features show promise for further subtype discrimination. These findings suggest an added diagnostic value of quantitative feature analysis of diffusion MR imaging in pediatric neuro-oncology.
BACKGROUND AND PURPOSE: Qualitative radiologic MR imaging review affords limited differentiation among types of pediatric posterior fossa brain tumors and cannot detect histologic or molecular subtypes, which could help to stratify treatment. This study aimed to improve current posterior fossa discrimination of histologic tumor type by using support vector machine classifiers on quantitative MR imaging features. MATERIALS AND METHODS: This retrospective study included preoperative MRI in 40 children with posterior fossa tumors (17 medulloblastomas, 16 pilocytic astrocytomas, and 7 ependymomas). Shape, histogram, and textural features were computed from contrast-enhanced T2WI and T1WI and diffusivity (ADC) maps. Combinations of features were used to train tumor-type-specific classifiers for medulloblastoma, pilocytic astrocytoma, and ependymoma types in separation and as a joint posterior fossa classifier. A tumor-subtype classifier was also produced for classic medulloblastoma. The performance of different classifiers was assessed and compared by using randomly selected subsets of training and test data. RESULTS: ADC histogram features (25th and 75th percentiles and skewness) yielded the best classification of tumor type (on average >95.8% of medulloblastomas, >96.9% of pilocytic astrocytomas, and >94.3% of ependymomas by using 8 training samples). The resulting joint posterior fossa classifier correctly assigned >91.4% of the posterior fossa tumors. For subtype classification, 89.4% of classic medulloblastomas were correctly classified on the basis of ADC texture features extracted from the Gray-Level Co-Occurence Matrix. CONCLUSIONS: Support vector machine-based classifiers using ADC histogram features yielded very good discrimination among pediatric posterior fossa tumor types, and ADC textural features show promise for further subtype discrimination. These findings suggest an added diagnostic value of quantitative feature analysis of diffusion MR imaging in pediatric neuro-oncology.
Authors: Tina Young Poussaint; Sridhar Vajapeyam; Kelsey I Ricci; Ashok Panigrahy; Mehmet Kocak; Larry E Kun; James M Boyett; Ian F Pollack; Maryam Fouladi Journal: Neuro Oncol Date: 2015-10-20 Impact factor: 12.300
Authors: H Zhou; R Hu; O Tang; C Hu; L Tang; K Chang; Q Shen; J Wu; B Zou; B Xiao; J Boxerman; W Chen; R Y Huang; L Yang; H X Bai; C Zhu Journal: AJNR Am J Neuroradiol Date: 2020-07 Impact factor: 3.825
Authors: M S Aboian; E Tong; D A Solomon; C Kline; A Gautam; A Vardapetyan; B Tamrazi; Y Li; C D Jordan; E Felton; B Weinberg; S Braunstein; S Mueller; S Cha Journal: AJNR Am J Neuroradiol Date: 2019-11-06 Impact factor: 3.825
Authors: Katherine A Zukotynski; Sridhar Vajapeyam; Frederic H Fahey; Mehmet Kocak; Douglas Brown; Kelsey I Ricci; Arzu Onar-Thomas; Maryam Fouladi; Tina Young Poussaint Journal: J Nucl Med Date: 2017-03-30 Impact factor: 10.057
Authors: J H Harreld; S N Hwang; I Qaddoumi; R G Tatevossian; X Li; J Dalton; K Haupfear; Y Li; D W Ellison Journal: AJNR Am J Neuroradiol Date: 2016-07-28 Impact factor: 3.825