Abstract

It is well known that high-dose radiation results in brain injury, the long-term effects of low-dose radiation, however, is uncertain. The purpose of this study was to investigate the role of low-dose bath on MRI changes, on both clinical and micro-architectural levels, using machine learning and computational methods. Six hundreds and fifty-six temporal lobe and 45 whole brain DVHs from patients treated for nasopharyngeal cancer (NPC) with IMRT or proton were included in this study. Temporal lobe injury (TLI) was defined as development of new T1 enhancement on MRI with or without surrounding T2 edema. Patients were divided randomly into train (50%) and test (50%) sets. Accuracy and AUC were used to evaluate the model performance. Minimum redundancy maximum relevance (MRMR) or SHAP algorithms was employed for feature selection. Support vector machine or random forest was used for classification. Automated cortical region segmentation using FreeSurfer v6 was performed in 33 patients with a minimum follow-up of 4 years. Architectural and biological MRI changes were determined in 34 different brain regions for each individual patient. The top-ranked temporal lobe features predicting TLI were V66/V38 for IMRT patients and V10 for proton patients with an AUC of 0.95 and 0.74, respectively. For whole brain, the top features were V16 and V13 with an AUC of 0.70. The rates of TLI at 5 years for V10-20(whole brain) ≥ 180cc and V10-20(whole brain) < 180 were 39.5% and 6.2%, respectively (HR = 5.5, 95% CI 1.4-22.0, p = 0.02). There were global changes in gray matter thickness, with most pronounced changes occurred in parietal lobe (-4.79%, p = 0.007) and occipital lobe (-5.68%, p = 0.03). Similarly, there were diffuse changes in white matter and subcortical volume. After radiation, the frontal lobe increased by 17.5% (p = 0.04), lateral ventricle 41% (p = 0.03), and choroid plexus 34.3% (p = 0.03). Low-dose radiation bath is associated with increased risk of temporal lobe injury and global disruption in brain architecture in NPC survivors.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call