Truace
Subscribe

Automated surgical instrument counting with a YOLO computer vision model

Source article: Automatic Detection and Counting of Surgical Instruments in Surgery Using AI-Based Computer Vision Model

IntroductionCounting surgical instruments during surgical procedures is critical for ensuring patient safety. Our goal is to develop a machine learning-based computer vision algorithm to create a reliable, automated surgical instrument counting support system. We hypothesize that our machine learning algorithm, based on the YOLO (You Only Look Once) model, can effectively perform instrument counting in surgery videos, providing real-time feedback to the surgical team.MethodsYOLO is a real-time object detection a…

By the Truvace desk2 min read1 sourcePeer-reviewedTRV-2026-2109· cite & verify
Automatic Detection and Counting of Surgical Instruments in Surgery Using AI-Based Computer Vision Model
BLW Amputation Set in a wooden box by Jenny O'Donnell. CC BY-SA 2.0 uk · https://creativecommons.org/licenses/by-sa/2.0/uk/deed.en
Trace impact readingContested
P 67The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.

Both sides are scored from claims and sources, not community votes.

G 65The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.

In brief

Researchers built a real-time object detection system on the YOLO architecture to detect and count surgical instruments in video, with the aim of giving the surgical team live feedback. They trained it on simulated open surgery footage and annotated images, then tested it on a new surgical video.

Instrument counting is tied to patient safety, so automated support could in principle reduce reliance on manual counts. However, the reported success rate is modest, small tools were handled worse, and the system has not been shown to work in actual operating rooms.

Main points

  1. The model was built on YOLO, a real-time object detection algorithm based on a convolutional neural network.
  2. Training data included 100 minutes of simulated open surgery video with 600 instrument transfer events.
  3. Images covering 6 instrument types were collected, plus 500 more from online surgeries, and annotated by a human expert.
  4. Validation showed a 72.7% success rate, with weaker performance on small hemostats and needles.
  5. Authors call for better small-instrument detection and validation in real operating rooms.

The gain

A YOLO-based model detected and counted surgical instruments in simulated open surgery video, reaching a 72.7% success rate in validation and larger instruments were classified more reliably.

The problem

The model classified small instruments such as small hemostats and needles less successfully than large ones, and it was tested only in simulation, which limits confidence for real counting safeguards.

The rundown

The dataset combined 100 minutes of simulated open surgery video, including 600 instrument transfer events, with 500 images of 6 different surgical instruments and another 500 images from online surgeries. A human expert labeled the names and positions of instruments in all images.

Eighty percent of the data was used for training and 20 percent for validation. The trained model was then tested on a new surgical video to count instruments within a predefined region of interest.

The authors conclude the YOLOv11-based model can detect and count instruments in a simulated open surgical environment. They state future work should improve detection of small instruments and validate the system in real operating room environments.

What this doesn’t fix

The system was evaluated only in a simulated open surgical environment within a predefined region of interest, and it performed worse on small instruments, so real operating room performance remains unvalidated.

Sources

  1. Peer-reviewedSurgical Innovation2026-10-20

The debate