• needle detection
• embedded systems architecture
• computer vision-based control
• in-vitro simulators
• pneumothorax

Pneumothorax is the accumulation of air in the pleural cavity. Pigtail or chest tube is required for treatment of pneumothoraxes greater than 20-30%. It is essential to provide appropriate medical training that is in accordance with educational standards and ethical principles. For this reason, the use of simulation is growing rapidly. The aim of this master thesis was to develop computer vision algorithms needed to detect needle entry point, orientation, and tip tracking in a pigtail catheter medical simulator. This work aimed also to develop part of the architecture of this simulator. The computer vision algorithms were developed and tested in Visual Studio in C++. The architecture part was a stand camera support designed in Fusion 360 and printed through FDM technique. Three computer vision algorithms have been developed and they led to obtain entry point, orientation, and depth penetration of the needle inside the simulator. The final algorithm has been chosen considering the fully embedded system architecture, data elaboration constraint of microprocessor, and time for real time data elaboration. After being designed and realized, the architecture part can adapt at the whole architecture. This work developed a Computer Vision Control system, by ex-novo code, for a pneumothorax simulator to maintain the system portable with a fully embedded architecture, affordable costs, and rapid processing time of the images with immediate response to the operator.