Machine Vision-Based Surgical Lighting System:Design and Implementation

TL;DR

This work tackles the challenge of steady, ergonomic surgical lighting by replacing manual lamp adjustments with an automated, image-guided system. It leverages YOLOv11 to detect a blue marker above the target site and uses two tilt-pan servos controlled via MQTT on a Raspberry Pi to direct a high-power LED lamp toward the target. The approach demonstrates high object-detection performance and a practical workflow that reduces surgeon effort while maintaining illumination quality, with a modular ceiling-mounted design suitable for multi-unit collaboration. Real-world validation in operating rooms and scalability to multiple units are identified as key future steps to maximize clinical impact.

Abstract

Effortless and ergonomically designed surgical lighting is critical for precision and safety during procedures. However, traditional systems often rely on manual adjustments, leading to surgeon fatigue, neck strain, and inconsistent illumination due to drift and shadowing. To address these challenges, we propose a novel surgical lighting system that leverages the YOLOv11 object detection algorithm to identify a blue marker placed above the target surgical site. A high-power LED light source is then directed to the identified location using two servomotors equipped with tilt-pan brackets. The YOLO model achieves 96.7% mAP@50 on the validation set consisting of annotated images simulating surgical scenes with the blue spherical marker. By automating the lighting process, this machine vision-based solution reduces physical strain on surgeons, improves consistency in illumination, and supports improved surgical outcomes.

Figures (5)

• Figure 1: Overview of the proposed surgical lighting system. a) The surgeon holds the blue marker slightly above the TSS (the orange area) and presses a trigger. The on-device camera detect and localizes the marker position, and two servomotors move the lighting package to illuminate the TSS. b) Sequential Workflow of the system
• Figure 2: a) Outer view of the device, showing the camera, servomotors carrying the lighting package, and the 4 signal lights, along with the door lock and on/off switch. b) Inner view of the device, showcasing the internal components. c) The blue spherical marker with a 3 cm diameter used for detection. d) The trigger mechanism using a NodeMCU and a 9V battery.
• Figure 3: Localization schematic depicting the camera-captured image and the cropped area representing the surgical site where the blue marker will be positioned. The schematic illustrates how the motor rotation angles are determined based on the edges of the cropped image to direct the lighting package towards the location of the blue marker.
• Figure 4: a) The device is powered on and undergoes a 25-second configuration process, during which the red light is illuminated. b) Once the configuration is complete, the surgeon can hold the blue marker slightly above the TSS and presses the trigger. Upon activation of the trigger, the Raspberry Pi commands the camera to capture three images, a process that takes approximately 5 seconds. During this time, the yellow light is illuminated. c) The captured images are then processed by the YOLO algorithm to localize the blue marker. The detected location is mapped to the appropriate servo motor rotation angles, and the blue signal light turns on. d) The servo motors then adjust to the required angles, and once the motor positionin is complete, the green light is turned on. The device then waits for the next trigger press to repeat the process.
• Figure 5: This graph shows the performance of YOLO in detecting the blue marker over 25 epochs.a) Training box/loss reached 1.27 at the final epoch. b) Validation box/loss reached 1.52 c) mAP50 achieved 0.98 on test set and d) Recall reached 0.96