1329 Synthetic [18F]FET-PET Generation and Hotspot Prediction Via Preoperative MRI of Glioma Lacking Radiographic High-Grade Characteristics

INTRODUCTION: Limited amino acid Positron Emission Tomography (PET) availability hinders therapeutic decision-making for gliomas without typical high-grade imaging features. METHODS: We trained a deep learning (DL)-based model to segment tumors in MRI and extract radiomic features using the pyradiomics package and classification with a Random Forest classifier. To generate [18F]FET-PET images, we employed a Generative Adversarial Network (GAN) framework and utilized a split-input fusion module for processing different MRI sequences through feature extraction, concatenation, and self-attention. RESULTS: We included MRI and PET images from 215 studies for the hotspot classification task and 211 studies for the synthetic PET generation task. The top-performing radiomic features achieved 80% accuracy for hotspot prediction. From the synthetic [18F]FET-PET, 85% were classified as clinically useful by senior physicians. Peak Signal-to-Noise Ratio analysis indicated high signal fidelity with a peak at 40 dB, while Structural Similarity Index values showed structural congruence. Root Mean Square Error analysis demonstrated lower values below 5.6. Most Visual Information Fidelity scores ranged between 0.6 to 0.7. This indicates that synthetic PET images retain the essential informational content required for clinical assessment and diagnosis. CONCLUSIONS: For the first time, we demonstrate that predicting high [18F]FET-uptake and generating synthetic PET images out of preoperative MRI in LGG and HGG is feasible. Advanced MRI modalities and other generative AI models will be used to improve the algorithm further in future studies.