Update on Molecular Imaging Combined with Artificial Intelligence in Brain Tumors

Purpose: To assess the current and future clinical applications, benefits, and limitations of molecular imaging (MI) and artificial intelligence (AI) in the setting of diagnosing and treating brain tumors.

Methods/Materials: A literature review was performed to review published studies on the current applications of MI and AI including the use of AI-assisted models and positron emission tomography (PET) with radiotracers in patients with suspected brain tumors. After review, information was compiled to construct a presentation including basic and current concepts. Case samples from our institution were included to illustrate related imaging modalities and patterns.

Results: This presentation showcases the current knowledge about PET modalities and AI integration to provide more accurate diagnoses and develop patient-tailored treatment plans. Current studies exhibit the advantages of using PET with a variety of tracers such as 68Ga-DOTATATE and amino acid tracer (AAT) 18F-FET to identify neoplastic tissue and delineate tumor extent. These methods can also be used to evaluate anticancer therapy response by allowing discrimination of tumor recurrence, pseudoprogression, and radionecrosis. Other studies highlight the role of AI in improving patient outcomes in brain tumor cases in both the diagnostic and therapeutic stages. Namely, AI-assisted segmentation of brain tumors can reduce segmentation time while increasing efficiency. AI models and computer-aided detection (CAD) tools also allow characterization of tumors that may not be easily assessed by the human eye.

Conclusions: The use of new MI agents combined with AI in neuro-oncology have progressed significantly as modalities to supplement previously standardized imaging processes. This presentation details the new advancements in each of these modalities in neuro-oncology including their clinical indications, advantages, and limitations. AAT are promising in that they are more sensitive in imaging brain tumors and may be useful in distinguishing recurrent tumors from radiation necrosis, although further studies are needed.