top of page

This Robot Removed 8 Kidney Tumors With Zero Positive Margins

14 hours ago
2 min read

A robot performed eight autonomous tumor resections with 77 electrosurgical cuts, and every resection ended with negative margins.


The kidneys were hydrogel models cast from real patient CT scans, which makes the result promising but not yet a patient milestone.


Partial nephrectomy depends on one hard judgment, which is removing the whole tumor while leaving as much healthy kidney as possible.


What the Robot Actually Did


The system first cut around the tumor surface and then dissected deeper in stages.


One arm made the electrosurgical cuts while the other held the tissue with vacuum retraction.


Between stages, the robot scanned the scene again and replanned its next move.


The average margin error was 1.61 millimeters.


Why Looking Again Made the Difference


A kidney changes shape as it is cut, so a map drawn before surgery goes stale within minutes.


The team used networks trained in simulation to infer the full 3D anatomy from a single partial camera view.


Tumor position error during deep dissection was 2.13 millimeters, compared with 6.29 millimeters for a baseline that assumed the kidney stayed rigid.


A version that planned once and never replanned succeeded 0 percent of the time and punctured the tumor three times.


A Human Still Approved Every Move


A supervisor selected the tumor on screen and approved every cutting plan before the robot moved.


The supervisor could also request a new plan or halt the system at any point.


The robot planned and executed the cuts, and the human held the authority to say yes.


What the Result Does Not Show


The robot averaged about 12 minutes per resection, while the surgeon comparison averaged under 4 minutes.


The surgeon comparison covered only four resections, and the surgeon's minimum margin of 0.57 millimeters was thinner than the robot's 2.17 millimeters.


The phantoms did not bleed and had no blood flow.


The study covered open surgery only, and the authors state the sample was too small for strong statistical comparisons.


What to Watch Next


The next test is living tissue with blood flow and motion, which is where results from phantoms are hardest to carry over.


Until then, the achievement is a working method for seeing, replanning and cutting, not yet proof that a robot can do it safely in a person.


For verified evidence on autonomy claims across systems, read The 2026 Surgical Technology Intelligence Report.



Sources



Related reading



Weekly Newsletter

Surgical Tech Radar

The innovations reshaping the operating room, curated and verified, delivered to your inbox.

  • New surgical AI tools and robotics, tracked weekly
  • Verified claims separated from hype
  • Concise briefings built for busy surgeons

Free forever. No spam. Unsubscribe anytime.

bottom of page