Purpose or Case Report:
Neuroblastoma is a clinically heterogeneous pediatric malignancy, varying in location, histopathologic appearance, and biologic characteristics. Genetics plays an important role in the prognosis. Amplification of the MYC family member, MYCN, is found in 25% of cases and correlates with high-risk disease and poor prognosis. However, genetic information can only be obtained via surgery or biopsy with concurrent morbidity and sampling variability associated with biopsy. The ability to detect MYCN amplification from routine pre-operative imaging can stratify neuroblastoma risk groups and affect clinical decision making. The purpose of this study was to predict the patient's MYCN status based on radiomics analysis of the magnetic resonance imaging (MRI) characteristics in patients with neuroblastoma.
Methods & Materials:
In this study, a cohort of 130 patients (72 females and 58 males) diagnosed with neuroblastoma from the Children’s Hospital of Philadelphia was included and divided into two groups (“Train” and “Test”) randomly in a 7:3 split. Of these, 27 were MYCN amplified and 103 were MYCN non-amplified. The average age of patients was 3 years and 8 months. Their pre-operative images were manually segmented using 3D Slicer software. An automatic machine learning Tree-Based Pipeline Optimization Tool (TPOT) was run on the dataset and tested on the “Test” group.
Results: The automatic TPOT exported pipeline[ RandomForestClassifier( bootstrap = True, criterion = “entropy”, max_features = 0.35000000000000003, min_samples_leaf = 15, min_samples_split = 12, n_estimators = 100)] based on the T2WI “Train” group (84 cases) achieved an accuracy of 75.7% (AUC = 0.78) in the T2WI “Test” group (37 cases) with 75.0% specificity and 80.0% sensitivity in predicting the presence of MYCN amplification. The accuracy of the prediction for T1C “Test” group (31 cases) based on T1C “Train” group (70 cases) was 87.1% (AUC = 0.78) with 89.3% specificity and 66.7% sensitivity. Finally, the images of T1C and T2WI were combined and divided into “Train” (70 cases) and “Test” (30 cases) groups, on which the accuracy of the prediction for “Test” group was 76.7% (AUC = 0.76) with 76.9% specificity and 75.0% sensitivity.

Conclusions:
MRI radiomics can predict the MYCN status of neuroblastoma with good accuracy, which assist oncologists stratifying neuroblastoma risk groups and guiding clinical decision making.