Machine-learning radiomics to predict bone marrow metastasis of neuroblastoma using magnetic resonance imaging

Abstract

Background: Neuroblastoma is one common pediatric malignancy notorious for high temporal and spatial heterogeneities. More than half of its patients develop distant metastases involving vascularized organs, especially the bone marrow. It is thus necessary to have an economical, noninvasive method without much radiation for follow-ups. Radiomics has been used in many cancers to assist accurate diagnosis but not yet in bone marrow metastasis in neuroblastoma.

Methods: A total of 182 patients with neuroblastoma were retrospectively collected and randomly divided into the training and validation sets. Five-hundred and seventy-two radiomics features were extracted from magnetic resonance imaging, among which 41 significant ones were selected via T-test for model development. We attempted 13 machine-learning algorithms and eventually chose three best-performed models. The integrative performance evaluations are based on the area under the curves (AUCs), calibration curves, risk deciles plots, and other indexes.

Results: Extreme gradient boosting, random forest (RF), and adaptive boosting were the top three to predict bone marrow metastases in neuroblastoma while RF was the most accurate one. Its AUC was 0.90 (0.86-0.93), F1 score was 0.82, sensitivity was 0.76, and negative predictive value was 0.79 in the training set. The values were 0.82 (0.71-0.93), 0.80, 0.75, and 0.92 in the validation set, respectively.

Conclusions: Radiomics models are likely to contribute more to metastatic diagnoses and the formulation of personalized healthcare strategies in clinics. It has great potential of being a revolutionary method to replace traditional interventions in the future.

Keywords: bone marrow metastasis; machine learning; magnetic resonance imaging; neuroblastoma; radiomics.

Figure 1

Study flow chart. SMOTE, Synthetic Minority Over‐Sampling Technique.

Figure 2

ROC curves of the top three models. (a) ROC curves of RF. (b) ROC curves of XGB. (c) ROC curves of ADB. ADB, adaptive boosting; AUC, area under the curve; CV, cross‐validation; RF, random forest; ROC, receiver operator characteristics; XGB, extreme gradient boosting.

Figure 3

Performances of the top three models. (a–f) Calibration curves of the top three models in the training and validation sets. (g–i) Radar plots of the six most significant features in each model. (j–o) Metastatic probability according to deciles of predicted risk in the top three models. ADB, adaptive boosting; RF, random forest; XGB, extreme gradient boosting.

Figure 4

T‐test for the top three models' probabilities of metastatic and nonmetastatic neuroblastoma in the training and validation set. (a) T‐test for the probabilities of metastatic and nonmetastatic neuroblastoma in RF. (b) T‐test for the probabilities of metastatic and nonmetastatic neuroblastoma in XGB. (c) T‐test for the probabilities of metastatic and nonmetastatic neuroblastoma in ADB. ADB, adaptive boosting; RF, random forest; XGB, extreme gradient boosting.