Computed tomography and echocardiography image fusion technique for cardiac images

Ultrasound is used in minimally invasive cardiac procedures widely, because of its convenience and noninvasive nature. However, the low quality of ultrasound images usually limits their usefulness as a tool to guide cardiac procedures; it is often complicated to relate images to their anatomical context in the heart. For improving the interpretability of ultrasound images, where keeping ultrasound as a flexible real time imaging and functional modality, there is a need for some registration techniques that integrate them with their correspond context in high quality pre-operative models, such as Computed Tomography images or Magnetic Resonance Imaging. In this study, a fusion system which integrates the knowledge of segmentation and intensity into registration is presented in Computed Tomography and Echocardiography images of heart. The goal of this thesis is integrating detected features, segmentation result information, and intensity information from two mentioned images, into a non-rigid registration framework, and achieve a high quality spatial mapping. The fusion system is developed as following: First, multiple Echocardiography images are compounded to get a better quality image with wider field of view. A fusion method is presented which particularly intends to increase the segment-ability of echocardiography features such as ventricle contours and improving their contrast. The presented method is also capable of enhancing the contrast, decreasing the impact of echo artifacts, expanding the field of view and improving the signal to noise ratio. Then, a segmentation approach based on a constrained Level set method is developed to identify the feature from Echocardiography images. It is a new geometrically level Set algorithm for the segmentation of the endocardial contours in echocardiographic images in presence of intensity non-uniformity. It will present an accurate and robust segmentation technique, which its results are going to use as input for fusion system in the following. In last stage, non-rigid registration is applied using segmentation result information plus intensity information from two images and a consistent transformation to match these features together is calculated. The proposed fusion system can use for medical interventions, for better physiological understanding, effective image guidance surgery, treatment, monitoring and diagnostic purposes, through finding spatial mapping between two images, to observe the changes of anatomical structure and to merge the information from multiple modalities. As it will be discussed in detail in the thesis, for input image, the proposed technique is unable in accurate segmentation in many instances at end diastole (87.3%) and over half the time at end-systole (61.7%). However, for fused images, it is unable to detect accurate segmentation 24.6% of times at end diastole, whilst there was just one failing at end systole (3.1%). It means fusion results in enhanced image quality consequently leads to effective ventricles segmentation. For evaluation, beside uncertainty estimation and visually evaluation by experts, quantitative and qualitative evaluations are conducted. For measuring the accuracy quantitatively, target registration error (TRE) is calculated before and after the registration, then a comparison is made. Also, different performance metrics are implemented to examine the performance of the proposed fusion system. For further studies, the combined navigation system can be designed for real-time surgery guidance. Furthermore, integrating virtual models and echocardiographic images will provide a potential means for giving image-guidance for processes which include both functional and anatomical imaging. Another direction for further study will be doing the registration for whole cardiac cycle: applying temporal synchronization between CT and echocardiography which is achieved by using ECG signals. Visualization of the result can be investigated further, as well.

Preview Text FSKTM 2016 45 - IR.pdf Download (906kB) | Preview

Actions (login required)

View Item