A “groundbreaking” artificial intelligence (AI)-guided technique may improve the accuracy of delivering experimental therapies directly to malignant brain tumours, according to research presented today at the Society of NeuroInterventional Surgery (SNIS) annual meeting (20–24 July, Seattle, USA).
The approach uses AI to identify all of the blood vessels feeding a tumour, allowing physicians to deliver therapy to more of the tumour while reducing off-target delivery.
Minimally invasive intra-arterial therapy allows for highly targeted dosing, minimising side-effects on the rest of the body. Traditionally, physicians have delivered these therapies through a single artery—but, because many tumours receive blood from multiple vessels, it is believed that a single-artery approach may result in incomplete tumour coverage.
In the present study, researchers at the University of Texas MD Anderson Cancer Center (Houston, USA) developed a new AI-assisted technique that identifies tumour-feeding arterial pedicles before delivering treatment. After confirming the AI findings with advanced imaging, the physicians then delivered a specific amount of therapy, tailored to each artery based on the amount of blood the vessel supplied.
Three patients with malignant brain tumours subsequently underwent treatment using this approach. The AI-assisted mapping successfully identified multiple tumour-feeding arteries in every patient, allowing physicians to treat all of these pedicles. The multi-pedicle approach covered more than 85% of each tumour across all cases—compared to less than 65% coverage with single-pedicle proximal infusion.
By creating a patient-specific map of the tumour’s blood supply, the physicians were therefore able to deliver therapy more precisely to the areas that needed it, reducing delivery outside of the intended treatment area. The researchers also found the procedure could be safely repeated two weeks later.
“One of the biggest challenges in treating malignant brain tumours is that every patient’s anatomy is different,” said Christopher Young (University of Texas Medical Branch, Galveston, USA), the study’s primary author. “AI gives us another tool to personalise treatment based on each patient’s unique blood supply, with the goal of delivering therapy more precisely. While more research is needed, this approach has the potential to improve patient care and marks an exciting advancement in neurointerventional oncology.”
Additional studies will be needed to determine whether improved tumour coverage leads to better outcomes for patients, according to the researchers.
