Abstract
Magnetic resonance imaging (MRI) robot can work in MRI environment for performing image-guided surgery or interventions. However, due to the limitations of magnetic field compatibility and rapid imaging methods, MRI robotics still have some challenges for clinical applications. For example, real-time imaging is required to improve accuracy and safety of intraoperative image-guided intervention. To accelerate imaging speed, sensitivity encoding (SENSE) has been widely used in clinical MRI. However, due to aliasing artifact and noise, SENSE requires regularization (e.g. Tikhonov regularization) for optimal reconstruction quality. Manually tuned regularization parameters needs human assistance and a long time processing, therefore it is not suitable for automatic robotic operations. In this project, an automatic tuning algorithm of Tikhonov regularization parameter is proposed to adapt automation operations in MRI robotics. The algorithm uses genetic algorithm for searching optimal regularization parameter automatically. Experimental results demonstrate that the proposed method is able to find optimal regularization parameter without manual tuning. It is desirable for automatic MRI robotics preoperative or intraoperative surgery or interventions.