MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Multi-Block Discriminant Analysis of Integrative 18F-FDG-PET/CT Radiomics for Predicting Circulating Tumor Cells in Early Stage Non-Small Cell Lung Cancer Treated with Stereotactic Body Radiation Therapy

S Lee*, G Kao, S Feigenberg, J Dorsey, M Frick, S Jean-baptiste, C Uche, Y Fan, Y Xiao, University of Pennsylvania, Philadelphia, PA

Presentations

(Sunday, 7/12/2020) 1:00 PM - 3:00 PM [Eastern Time (GMT-4)]

Room: Track 2

Purpose: To integrate ¹8F-FDG-PET/CT radiomics with multi-block discriminant analysis for predicting circulating tumor cells (CTCs) in early stage non-small cell lung cancer (NSCLC) treated with stereotactic body radiation therapy (SBRT).


Methods: Fifty-six patients with stage I NSCLC treated with SBRT underwent ¹8F-FDG-PET/CT imaging pre-SBRT and post-SBRT (median, 5 months; range, 3-10 months). For each patient, CTCs were assessed via a telomerase-based assay before and within 3 months after SBRT, and dichotomized at 5 and 1.3 CTCs/mL. Pre-SBRT, post-SBRT and delta radiomics features were extracted from the gross tumor volume of PET/CT images, each of which included 1,548/1,562 PET/CT radiomics features. Two (PET/CT)-block radiomics data integration was performed using block sparse partial least squares-discriminant analysis (sPLS-DA) referred as DIABLO that maximized common information between PET/CT radiomics datasets and identified key radiomics signatures, while discriminating CTC levels. The predictive performance of the DIABLO model for pre-SBRT and post-SBRT CTCs was evaluated using the combined AUC (averaged across different blocks) for each latent component with 20×5-fold cross-validation (CV), and compared with that of concatenation-based sPLS-DA that consisted of combining all radiomics features into one block. CV prediction scores between one class vs the other were compared using the Wilcoxon rank sum test.


Results: For predicting pre-SBRT CTCs, pre-SBRT radiomics-derived DIABLO model selected 11 features (PET/CT: 3/2, 1/1, 1/1 and 1/1) with four components, achieving CV-AUC of 0.841 (p=0.018) in the fourth component. For predicting post-SBRT CTCs, post-SBRT and delta radiomics-derived DIABLO models selected 11 and 12 features (post-SBRT PET/CT: 1/1, 3/2 and 1/3; delta PET/CT: 1/2, 1/4 and 3/1) with three components, achieving CV-AUCs of 0.794 (p=0.018) and 0.799 (p=0.013) in the third component. All corresponding single-block sPLS-DA models couldn’t attain CV-AUCs higher than 0.6.


Conclusion: Multi-block integration with discriminant analysis of ¹8F-FDG-PET/CT radiomics has the potential for predicting pre-SBRT and post-SBRT CTCs.

Funding Support, Disclosures, and Conflict of Interest: This project was supported by NCI grants, U24CA180803(IROC) and U10CA180868(NRG). Drs. Kao and Dorsey are co-founders and have equity in Liquid Biotech USA, Inc., a University of Pennsylvania PCI-developed company through the UPStart program.

Keywords

Image Analysis, FDG PET, CT

Taxonomy

IM/TH- Image Analysis (Single Modality or Multi-Modality): Imaging biomarkers and radiomics

Contact Email