Predicting future surgical steps during MCA aneurysm clipping using a multimodal transformer - Takeaways - MDSpire

Predicting future surgical steps during MCA aneurysm clipping using a multimodal transformer

  • By

  • Thomas J. On

  • Jonathan A. Tangsrivimol

  • Jiuxu Chen

  • Yuan Xu

  • Baoxin Li

  • Michael T. Lawton

  • Mark C. Preul

  • June 2, 2026

  • 0 min

Share

  • 1

    This study evaluated a transformer-based model for predicting surgical steps in MCA aneurysm clipping using recorded microscope videos.

  • 2

    The multimodal model achieved the highest accuracy of 0.683 and a weighted F1 score of 0.673, outperforming video-only and annotation-only models.

  • 3

    A total of 25 uncomplicated MCA bifurcation aneurysm surgeries were analyzed, with 18 used for training and 7 for independent testing.

  • 4

    The study demonstrated the feasibility of fixed-horizon surgical step prediction under controlled conditions, highlighting the potential of AI in neurosurgery.

  • 5

    Findings indicate that multimodal input combining video features and prior annotations enhances predictive performance for surgical procedures.

Original Source(s)

Predicting future surgical steps during MCA aneurysm clipping using a multimodal transformer

  • by Thomas J. On, Jonathan A. Tangsrivimol, Jiuxu Chen, Yuan Xu, Baoxin Li, Michael T. Lawton, Mark C. Preul

  • June 2, 2026

Related Content

Efficacy of Perampanel as a Monotherapy for Seizure Control During Awake Craniotomy in Glioma Patients

Management approaches, associated complications, and patient outcomes in spontaneous cerebellar hemorrhage: findings from a Swedish single-center observational study

Case Report: “Damage control” in obstetric neurosurgery: staged management of ruptured arteriovenous malformation with herniation