| Traditional image-guided navigation systems establish a 3D model of patient's anatomical structure using 3D reconstruction and visualization of pre-operative images. For soft tissue navigation, due to organ deformation caused by respiration, body motion, etc, the model established pre-operatively could not reflect the change of underlying anatomical structure during the operation and result in inaccurate navigation, even the failure of navigation. Therefore, real-time tracking of soft tissue deformation accurately has significant research and clinical value for the navigation of chest/abdominal surgeries, which involve soft tissue navigation.In this study, with the integration of high quality CT/MRI pre-operative images, real-time ultrasound (US) imaging and electromagnetic tracking system, we have proposed and implemented a system framework for real time tracking of soft tissue deformation. Based on this framework, the 3D structure and real-time deformation of the region of interest (ROI) could be tracked effectively. This system enables surgeons to perform abdominal interventional therapies, minimally invasive surgeries, and open surgeries more accurately.The system hardware consists of the following components: an electromagnetic tracking system, a video capture system, and a standard PC. The video output of a 2D US imaging system was connected to the video capturing adapter inside the PC, while the sensor of the tracking system was attached to the US probe firmly. The major software modules of the tracking system include the rigid registration of CT/MRI images with the coordinate of the electromagnetic tracking system, real-time capturing and localization of ultrasonic video, post-processing of ultrasound frames and ROI extraction, and the visualization of multi-modality images. In the capturing and localization module, the US probe was used to scan all the ROI in sequence and the tracking sensor detected the spatial position and orientation of every frame simultaneously. All of the data were captured and transferred to the PC. In the post-processing module, based on the characteristics and analysis of acquired frames, speckle noise was filtered with a local statistics method and the shape of ROI (such as the kidney) was extracted with a FCM segmentation algorithm. All the processes could be repeated where is in need.The framework was evaluated with preliminary experiments. With a phantom experiment, it has shown that the operation of the US system and probe would not influence the performance of the electromagnetic tracking system significantly. With fiducial markers fixed on skin, the registration between pre-operative MRI imaging and the tracking system was carried out and the registration accuracy was between 1 and 3 mm, which is acceptable for soft tissue navigation. With the post-processing module, the outline of kidney in US frames was extracted and compared with a 3D ROI model extracted from pre-operative high-resolution images. The preliminary test indicates that real-time US imaging is of great benefit to soft tissue surgeries/interventions. However, poor image quality, lower spatial resolution and limited region of scanning restrict its further application. The integration of real-time US frames with the pre-operative 3D anatomical structure would improve its performance greatly for soft tissue navigation.In conclusion, this thesis has implemented a system framework for real-time tracking of soft tissue deformation based on US imaging. Phantom experiment and volunteer study have indicated its feasibility for clinical application. The system integrates advantage of high quality pre-operative images with real-time ultrasound imaging for better identification and tracking of soft tissue deformation, which is the key problem during soft tissue navigation. It would be helpful for the implementation of computer-aided surgery involving soft tissue navigation and enable more accurate abdominal surgeries/interventions. In addition, it also provides a possible means for the property study of soft tissue. |
