{
  "abstract": "Background Accurate endovascular delivery is critical as intra-arterial therapies are increasingly explored for malignant brain tumors. We previously described a perfusion-guided super-selective infusion technique; however, single-pedicle approaches may result in incomplete tumor coverage.Methods We developed an AI-assisted, perfusion-guided, multi-pedicle super-selective infusion strategy to improve delivery accuracy and precision. An AI-based segmentation algorithm identified tumor-feeding arterial pedicles, which were confirmed with microcatheter perfusion imaging overlaid with brain MRI. MSC-D24 was infused through multiple distal arterial branches (M2, P2 and beyond), with dose allocation proportional to each pedicle’s contribution. The procedure was repeated at 14 days to assess reproducibility. Tumor coverage was quantified and compared to conventional single-pedicle infusion.Results Three patients with malignant brain tumors underwent treatment with MSC-D24 using this approach. AI-assisted mapping identified multiple tumor-feeding pedicles in all cases, all of which were successfully catheterized (technical success: 100%). Multi-pedicle infusion achieved >85% tumor coverage in all cases, compared to <65% with single-pedicle proximal infusion. This approach enabled consistent multi-territory delivery from distal branches while minimizing off-target perfusion. No procedural complications or complications of off-target embolization were observed. Repeat infusion at 14 days was feasible and reproducible in all patients.Conclusion AI-assisted multi-pedicle super-selective infusion significantly improves tumor coverage compared to conventional single-pedicle techniques and represents a feasible, reproducible, and scalable workflow for intra-arterial therapy delivery. This strategy may enhance therapeutic precision in neurointerventional oncology and warrants further evaluation of clinical efficacy.Disclosures C. Young: None. F. Lang: None. S. Chen: None.",
  "authors": [
    {
      "affiliations": [
        "Neurosurgery, University of Texas MD Anderson Cancer Center, Houston, TX"
      ],
      "name": "C Young"
    },
    {
      "affiliations": [
        "UTMB, Houston, TX"
      ],
      "name": "P Kan"
    },
    {
      "affiliations": [
        "Neurosurgery, University of Texas MD Anderson Cancer Center, Houston, TX"
      ],
      "name": "F Lang"
    },
    {
      "affiliations": [
        "Interventional Radiology, University of Texas MD Anderson Cancer Center, Houston, TX"
      ],
      "name": "S Chen"
    }
  ],
  "title": "O-033 From single-pedicle to whole tumor coverage: AI-guided multi-territory super-selective endovascular infusion for brain tumors",
  "uid": "aca918cd-8cf4-5b12-9a2c-752602c1017e"
}