MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Predicting Prognosis of Posterior Fossa Ependymoma Using MRI Radiomics

L Tam1, D Yecies1, M Han1, K Yeom1, S Mattonen2*, (1) Stanford University, Stanford, CA, USA (2) Western University, London, ON, CA

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: Posterior fossa ependymomas (PFE) are common pediatric brain tumors often assessed with magnetic resonance imaging (MRI) before surgery. Advanced image analysis with radiomics has shown promise in stratifying risk and outcome in other pediatric brain tumors. We aimed to extract high-dimensional MRI features to identify and compare prognostic imaging-based biomarkers of PFE to traditional clinical variables.


Methods: Children with PFE from five centers were retrospectively analyzed. Tumor volumes were manually contoured on T1-post contrast and T2-weighted MRI. A total of 900 features were extracted using Pyradiomics on each image series, including first-order statistics, size, shape, and texture metrics calculated on the original and filtered images. 10-fold cross-validation of a least absolute shrinkage and selection operator (LASSO) Cox regression was used to select the top radiomic features to predict progression free survival (PFS). Model performance was analyzed using the concordance index (C) and we compared models using clinical variables only, radiomics only, and radiomics plus clinical variables.


Results: A total of 93 children from five medical centers were included (median age = 3.3 years; 59 males; mean PFS = 50 months). Six radiomic features were selected all on T1 MRI, including 1 first-order kurtosis feature and 5 texture features. This model demonstrated significantly higher performance than the model with only clinical features (age at diagnosis and sex) [C: 0.69 vs 0.58, p<0.001]. Adding clinical features to the radiomic features didn’t improve prediction (p=0.67). However, for patients with molecular sub-typing (n=48), adding this feature to the clinical plus radiomics models significantly improved performance over only clinical features [C: 0.79 vs. 0.66, p=0.02]. Further validation and model refinement with additional datasets are ongoing.


Conclusion: This pilot study demonstrated the potential of MRI-based radiomic features to significantly improve risk stratification of PFE compared to traditional clinical features alone.

Keywords

Quantitative Imaging, MRI, Brain

Taxonomy

IM/TH- Image Analysis (Single Modality or Multi-Modality): Computer-aided decision support systems (detection, diagnosis, risk prediction, staging, treatment response assessment/monitoring, prognosis prediction)

Contact Email