Establishing a clinically relevant ct and associated radiomics pipeline for intracranial rodent tumour models
NEURO-ONCOLOGY(2022)
Abstract
Abstract While magnetic resonance imaging (MRI) is the predominant imaging modality for glioblastoma (GBM), pre-clinical MRI scanner availability is limited. As pre-clinical CT is more widely available and cost-effective, this study aimed to 1) establish preclinical-GBM CT and CT-radiomic workflows, 2) identify whether CE-CT could detect murine orthotopic GBM tumours on two CT instruments [TRIUMPHX-O-CT; IVISSPECTRUM-CT], 3) assess whether CT-radiomic features could distinguish tumour from normal tissue, and support earlier detection of tumours, 4) verify translation of pre-clinical CT-radiomic pipelines to, and assess pre-clinical CT-features in, clinical CE-CT scans.U87R-Luc2(n=25) and NFpp10a-Luc2(n=10) orthotopic GBM models were established and tumours monitored via bioluminescence imaging (BLI). Concurrently, mice underwent CE-CT (IV-iodine/300mg/mL/50kV-scan). Extracted radiomic features (PyRadiomics) underwent dimensionality reduction (Spearman correlation; >0.85). Remaining features were analysed (Recursive feature elimination (RFE)/RepeatedCV/randomforest) in normal and tumour tissue and across timepoints (TRIUMPHX-O-CT-Wk3vsWk6,Wk6vsWk9/12; IVISSPECTRUM-CT-Wk6vsWk9/12).CE-CT and radiomic pipelines were successfully established for orthotopic GBM models, using both CT-systems. On visual assessment of images, BLI was significantly more sensitive, with tumours detectable at Wk1 (BLI) vs Wk9 (CE-CT). However, RFE analysis identified CT-radiomic features (first_order&glcm) which differentiated tumour from normal tissue (TRIUMPHX-O-CT). A subsequent feature set (first_order,glcm,gldm&glzm) were identified (TRIUMPHX-O-CT/IVISSPECTRUM-CT), detecting tumours earlier (Wk3&Wk6) than possible by visual assessment of CTs. Preclinical radiomic methods were successfully applied to exploratory clinical CE-CT scans(n=10). Here, several preclinical CT-features (e.g. Zone_Entropy) showed increased intensity in tumour regions. Overall experimental BLI is the most sensitive method for pre-clinical intracranial tumour detection. However, analysis of clinically relevant CT-radiomic features may facilitate tumour identification and earlier tumour detection (Wk3/Wk6-TRIUMPHX-O-CT/Wk6-IVISSPECTRUM-CT) than possible by visual assessment of CT (Wk9). Clinically relevant CT-derived radiomic features may therefore support intracranial rodent tumour assessment. Importantly, preclinical radiomic methods successfully translate to clinical CT-radiomic analysis. Parallel trends in tumour-specific feature intensities across pre-clinical and clinical scans suggest species conservation of features.
MoreTranslated text
Key words
associated radiomics pipeline,ct
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined