اشكان قربانيان

New developments in digital technologies an‎d artificial intelligence have opened new horizons for modern medicine, especially robotic surgery. One of the key concepts in this field is the human digital twin: a virtual an‎d dynamic model of the human body that can be synchronized with real patient data an‎d used for simulation an‎d real-time treatment strategies. In this research, with the aim of improving accuracy an‎d safety in minimally invasive spinal surgeries, a comprehensive framework for the development of a human digital twin is presented, which combines three-dimensional anatomical modeling, real-time motion information extraction, an‎d motion inference from biology. In the first step, a precise three-dimensional anatomical model of the patients was reconstructed using medical images (CT an‎d MRI), an‎d with the help of convolutional neural networks, accurate segmentation of organs, including the vertebrae of the spine an‎d the rib cage, was performed. Then, to obtain motion data, a laboratory environment similar to the real surgical conditions was designed an‎d implemented, in which machine vision cameras were used to track the patientʹs movements during surgery. Subsequently, the calibration process of imaging equipment, precise marker identification, an‎d motion data correction were implemented to ensure that the information obtained is accurate enough for training the motion estimator model an‎d kinematic analysis. After obtaining the motion data, neural networks were used to estimate the movement of markers in different phases of breathing an‎d patient movement, an‎d these data were finally used in the anatomical kinematic model, creating a dynamic an‎d live four-dimensional (three-dimensional over time) model of the patientʹs body, which can be updat‎ed in real-time in the operating room. In this model, it is possible to display the natural movements of the body an‎d predict changes during the operation. The experimental results showed that the proposed system is able to provide surgeons an‎d robotic systems with meaningful an‎d effective information on subtle movements with acceptable accuracy in predicting patient positions during surgery. This research is a new step towards personalized digital simulation an‎d the human digital twin in smart spinal surgery environments an‎d provides a suitable platform for the development of smarter, safer, an‎d more personalized systems in the future.

همتاي ديجيتال , جراحي كم تهاجم رباتيك , مدل‌سازي آناتومي , مكانيك زيست , تصويربرداري پزشكي , بينايي ماشين , شبكه عصبي

Digital twin , robotic minimally invasive surgery , anatomy modeling , biomechanics , Medical Imaging , computer vision , neural network

Ashkan Ghorbanian

Dr. Borhan Beigzade