Integration of an optimized dynamic and radiomics analyses in 18F-FDOPA PET for the caracterization of gliomas in clinical routine practice
- Authors
- Publication Date
- Oct 27, 2021
- Source
- Hal-Diderot
- Keywords
- Language
- English
- License
- Unknown
- External links
Abstract
In the era of personalized medicine, having an efficient tool to provide additional information for the non-invasive characterization of gliomas is crucial. Such a tool would help clinicians make the best decisions to increase patient overall survival while preserving quality of life. In this context, medical imaging is becoming increasingly important especially amino-acid PET imaging, which is recommended by international guidelines as an adjunct to MRI. The 18F-FDOPA amino-acid radiotracer, which has been validated for PET in neuro-oncology, is the most widely used radiotracer for these indications in France. However, the interpretation of 18F-FDOPA PET results actually mainly relies on simple visual and semi-quantitative analyses of a static image that represents the spatial distribution of the radiotracer. This thesis seeks to better exploit each acquisition by adding dynamic and radiomics analyses. Dynamic analysis extract information about the temporal variation of tumoral metabolism and radiomics analysis quantify tumoral heterogeneity, a potential marker of tumor aggressiveness. This research work optimizes the methods for 18F-FDOPA PET dynamic analysis. It allows to provide a sound basis for drafting harmonization guidelines based on the simple and optimized semi quantitative model, which may be associated with normalization of tumor data with to the healthy brain, easily transposable to clinical routine practice and allowing to harmonize heterogeneously Carbidopa premedicated cases. The work then investigates the diagnostic performances of dynamic parameters for characterizing gliomas. Dynamic parameters are biomarkers of interest to non-invasively predict the IDH mutation even in the company of radiomics features, which may themselves help predict of the 1p/19q co-deletion. Although dynamic parameters are less important for the detection of glioma recurrences they deserve to be defined more precisely, particularly for high gr ade gliomas. In order to extend these results to the clinic, we developed a dedicated software for the advanced analysis of 18F-FDOPA PET in gliomas. In light of the results presented, dynamic and radiomics 18F-FDOPA PET analyses in neuro-oncology deserve to be further explored by integrating the temporal dimension in the computation of radiomics features. These analyses also need to be applied to other clinical questions through multicentric studies.
