AutoPathNet rapidly generated patient-specific candidate trajectories for minimally invasive evacuation of hypertensive ICH, with algorithm-generated trajectories showing lower composite geometric planning metric values than surgeon-implemented catheter trajectories in 76.2% of paired cases.
AutoPathNet generated three candidate trajectories within approximately 31 seconds per case.
In 76.2% of paired cases, the best algorithm-generated candidate trajectory had a lower composite geometric planning metric (m value) than the surgeon-implemented catheter trajectory.
AutoPathNet uses a multi-step pipeline reconstructing patient-specific anatomical models from CT-derived inputs to generate and rank candidate trajectories.
The authors identified several limitations requiring resolution before clinical deployment of AutoPathNet.
This research describes AutoPathNet, a computer algorithm designed to help neurosurgeons plan the safest path for a small tube (catheter) to drain blood clots that form in the brain after a hypertensive stroke — a condition called intracerebral hemorrhage (ICH). Currently, surgeons must manually plan these paths using brain scans, which takes time and can vary between surgeons. AutoPathNet automates this by reading CT scans, building a 3D model of the patient's brain anatomy, generating possible surgical paths, filtering out unsafe options, and ranking the remaining paths using a geometric scoring system — all in about 31 seconds. The researchers tested AutoPathNet by comparing its suggested surgical paths against the actual paths surgeons used in 21 real patients, as measured from post-surgery CT scans. In 16 out of 21 cases (about 76%), the algorithm's top-ranked path scored better geometrically than what the surgeon actually did. However, the authors are careful to note that a better geometric score does not necessarily mean better patient outcomes — it simply means the algorithm's path fits geometric safety criteria more favorably within the model's constraints. This research suggests that AutoPathNet could serve as a helpful planning tool to assist neurosurgeons rather than replace their judgment, potentially making surgical path planning faster and more consistent. However, the authors emphasize that the system is not ready for clinical use without further testing in prospective studies that include real-world functional brain mapping data and actual patient outcome measurements. The study represents an early but promising step toward computer-assisted surgical planning for brain hemorrhage treatment.
Li W, Dong L, Yan H, Zhang Y, Liu X, Tan K. (2026). AutoPathNet: a patient-specific automated 3D trajectory planning framework for minimally invasive evacuation of hypertensive intracerebral hemorrhage.. Computer assisted surgery (Abingdon, England). https://doi.org/10.1080/24699322.2026.2726003