Influence of inter-subject correspondences on liver motion predictions from population models

Evaluation of 2D and 3D ultrasound tracking algorithms and impact on ultrasound‐guided liver radiotherapy margins

“…Therefore we did not include 3D forecasting in this work. 3D guidance from 2D motion predictions can be achieved by carefully aligning the US transducer with the main direction of the respiratory motion, calibrating the US coordinate system with the treatment coordinate system, and employing a 4D motion model …”

“…Such an approach achieved in‐vivo for the right liver lobe of eight subjects a spatio‐temporal 3D mean prediction accuracy of 2.4 (2.7) mm for a system latency of 150 (400) ms based on 2D US tracking (mean accuracy 0.9 mm) and a population 4D motion model . Note that improved performance was observed for the central liver region, which is closer to the typical US plane, and by adapting the model to the subject using few breath‐hold observations …”

“…3D guidance from 2D motion predictions can be achieved by carefully aligning the US transducer with the main direction of the respiratory motion, calibrating the US coordinate system with the treatment coordinate system, and employing a 4D motion model. 38,40 Figure 3 compares the mean prediction error (MPE) of the linear adaptive filter and decision fusion prediction. Predictions are obtained from leave-one-subject-out cross-validation on…”

“…Instead, the motion of other visible anatomical structures (e.g. vessels) can be estimated and used as input to 4D liver motion models to spatially predict the tumor position …”

“…vessels) can be estimated 35,36 and used as input to 4D liver motion models to spatially predict the tumor position. [37][38][39][40] Although US probes are typically operated by hands, either passive arms or robotic arms can be used to hold the probe and therefore operators are not required to be in the treatment room. A robotic arm offers the additional advantage of moving with the target organ or helping inexperienced users to find it through cooperative control.…”