Automatic registration and precise tumour localization method for robot-assisted puncture procedure under inconsistent breath-holding conditions.

Background: During percutaneous puncture procedure, breath holding is subjectively controlled by patients, and it is difficult to ensure consistent tumour position between the preoperative CT scanning phase and the intraoperative puncture phase. In addition, the manual registration process is time-consuming and has low accuracy.

Methods: We have proposed an automatic registration method using optical markers and a tumour breath-holding position estimation model based on the support vector regression algorithm. A robot system and a tumour respiratory motion simulation platform are built to perform puncture tests under different breath-holding states.

Results: The experimental results show that automatic registration has higher accuracy than manual registration, and with the tumour breath-holding position estimation model, the targeting accuracy of puncture under inconsistent breath-holding conditions is greatly improved.

Conclusions: The proposed automatic registration and tumour breath-holding position estimation model can improve the accuracy and efficiency of puncture under inconsistent breath-holding conditions.