EAES Academy

Aims. Electrode positioning during radiofrequency ablation of internal organs neoplasms or needle positioning for biopsy is a challenging and time-consuming task on which the result of diagnosis and treatment depends. And in the case of using robotic-assisted systems, the technical complexity associated with intraoperative position determination of the tumor also manifests itself.
Methods. To solve this problem with the help of modern technologies it is proposed to use the existing techniques and methods of computer vision. This process is more complicated because of low image resolution, the absence of a color component and overall noisiness.
Based on the assumption that the tracked object occupies the main part of the image in the desired area, it becomes possible to highlight it on the image with a certain color, which should simplify the perception of the ultrasound image.
The specifics of working with ultrasonic video images also made it possible to apply several optimizations, which decreased frame processing time. Since the velocity of the neoplasm is limited from above by some reasonable value, it is possible to restrict the search area on the next frame. Several parameters like velocity, position, and frame rate are being considered.
Results. Thus, in the final system, it is proposed to use more than ten algorithms that allow tracking neoplasms and other structures of internal organs. Combining them is possible with different weighting factors. Additional testing of this system prototype showed the possibility of using it to track patterns that are visually distinguished on the ultrasound image, such as neoplasms, tumors, cysts, and vessels.
Conclusion. This technique can achieve a performance of up to 40-90 frames per second with a tracking accuracy exceeding 90%, which will allow real-time transmission of the coordinates of the neoplasm to the navigation system of the robotic-assisted systems while highlighting various patient structures with colors for convenience of medical staff.