Speaker
Description
In this talk, we introduce a patient-specific digital twin framework termed COMPASS (COMprehensive Personalized ASsessment System) for adaptive radiotherapy of non-small cell lung cancer (NSCLC) patients based on fractional PET/KVCT imaging, radiomics, dosiomics, and biologically equivalent dose (BED) kinetics. Specifically, eight NSCLC patients treated with biology-guided radiotherapy (BGRT) were modeled with 99 organ-fraction observations across 24 organ trajectories. Organ specific time-series features were derived to preserve spatial dose heterogeneity and biological response. A gated recurrent unit (GRU) autoencoder was used to learn compact latent representations of evolving dose–response trajectories for critical organs, which were subsequently classified using logistic regression to predict eventual CTCAE grade ≥1 toxicity. Despite the limited cohort size, COMPASS achieved an AUC of 0.90 with 80% sensitivity and 78% specificity. Importantly, elevated toxicity risks were predicted several fractions prior to clinical symptom onset, defining an actionable window for early intervention for patients. Incorporation of BED kinetics and spatial dose-texture features sensitively captured transient metabolic and dosimetric perturbations not reflected by regular metrics. COMPASS provides a physics- and biology-informed framework for adaptive radiotherapy where organ toxicity and tumor dynamics are continuously updated based on delivered dose and images to guide radiotherapy.
