Current developments in surgical equipment and technologies are primarily aimed at making medical manipulation less invasive and shorter in time. This is also true for liver surgeries. Endoscopic tumour ablation is preferred to open surgery, however, the results of this intervention are highly dependent on the instrument placement precision. To track the instrument's motion this type of intervention is usually done under X-ray or ultrasound guidance. The first variant is not optimal for both patient and doctor due to the high radiation exposure, however, the second modality has a significant drawback of low spacial resolution, though a very high temporal.
Therefore, it would be advantageous to combine the high spatial resolution of MRI or CT with the temporal resolution of US. This can be done by registration of pre-interventional MRI or CT scans to the intraoperational US scans. One of the main problems of multimodal image registration of the liver is related to its continuous deformation and movement caused by respiration. A possible solution could be the creation of a motion model, which would predict changes in the liver position and transform preoperative images accordingly. However, as breathing patterns vary significantly between different people, one model will not be able to fit all the patients, therefore the model should be individualized.
The problem is that it should be trained on an imaging modality, which is going to be altered with the help of the model in the future. As CT images are going to be transformed, it is required to train each personalized model on a significant amount of CT images, which nullifies the initial attempt to decrease radiation exposure. A possible solution to this problem is to train the model on one imaging modality, such as 3D US, but apply it to the other (3D CT). To make it, a known transformation between these two modalities is required. It can be gained by performing 3D CT - 3D US registration.
Therefore, liver intervention would have a couple more steps:
1. Perform CT-US registration to obtain the transformation matrix;
2. Start the intervention under US guidance and apply the motion model to the preoperative CT scans to have the best out of two modalities.
The main aim of this work is to perform 3D US -3D CT image registration of the liver and alter the respiratory motion model to be used under a different modality.