Laser Precision: Russian Researchers Develop an AI Assistant for Surgeons
Russian developers have trained artificial intelligence to identify laser fibers and stones in video recordings of surgical procedures. The technology could provide an additional layer of monitoring during surgery and help doctors review complex clinical cases.
Kidney stones are a common problem that can cause excruciating pain. Treatment often involves removing the stone, also known as a urinary calculus. Several methods are available, including laser lithotripsy, which breaks up the stone with laser energy. This approach is particularly useful for large stones that cannot be removed endoscopically through the urethra.
During laser procedures, surgeons must aim the instrument precisely at the stone without damaging surrounding tissue. They rely on images from an endoscopic camera to guide their movements, making visual monitoring a critical part of the procedure.
A team from MIREA – Russian Technological University and the Kulakov National Medical Research Center for Obstetrics, Gynecology and Perinatology has developed a prototype AI-powered service to analyze these procedures.
Every Millimeter Matters
The system uses a neural network to identify the surgical field, locate the stone and laser fiber in the video image, determine their positions and direction, and classify the target of the laser.
To build the system, the developers had to solve several computer vision challenges. The AI assistant must distinguish the relevant objects in actual surgical recordings, determine where they are and analyze the direction of the laser emitter. During testing, the system achieved accuracy above 90%.
The team also improved processing speed. Initially, analyzing a recording of a single procedure took 102 seconds. After the software was optimized, processing time fell to 21 seconds, nearly five times faster. This matters because clinical use depends not only on recognition accuracy but also on the ability to process large volumes of video quickly. The recordings can help clinicians review complex episodes, assess a surgeon’s actions and prepare training materials.
Putting Patient Safety First
As Vadim Kirillov, the project’s research lead and a lecturer in the Department of Artificial Intelligence Technologies at MIREA’s Institute of Artificial Intelligence, emphasized, the technology was developed above all to help protect patients during surgery.
“Our service helps automatically annotate video, identify complex episodes for specialist training and analyze the quality of laser targeting. Importantly, we worked alongside a practicing physician, which allowed us to account for the clinic’s actual needs rather than focusing solely on the algorithm’s technical capabilities,” Kirillov said.
Collaboration with the Kulakov Center helped the young researchers connect their algorithms to real-world surgical needs. “We are proud that our team has turned an idea into a working prototype with high accuracy. Projects like this show that our students are ready to tackle real healthcare challenges, and that an interdisciplinary approach is becoming the foundation for developing technologies that meet practical needs,” said Stanislav Kudzh, rector of MIREA – Russian Technological University.
The next step is clinical validation and an assessment of how the algorithm performs under different conditions.
Toward Clinical Adoption
In the future, the technology could serve as the basis for an AI assistant that highlights relevant objects on the surgical display and alerts physicians to potentially dangerous situations. For example, it could flag cases in which the laser fiber is aimed at surrounding tissue rather than the stone, or when the target is difficult to see. Further testing will be needed. Researchers must evaluate not only correct detections but also potential errors, including false alarms and instances in which the system fails to detect a dangerous event.
Patients stand to benefit from the additional layer of monitoring during surgery. Surgeons, in turn, could analyze procedures more quickly and review difficult cases more efficiently. The service could also prove useful in training medical professionals.
The technology may also have potential beyond the Russian market. AI-based medical solutions developed by Russian software engineers are already reaching international users. For example, medical imaging analysis systems developed by Celsus have begun to be used in Pakistan.
