Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Electrical and Computer Engineering


Dr. Abbas M. Samani

2nd Supervisor

Dr. Rajni V. Patel

Joint Supervisor


Various types of motion and deformation that the lung undergoes during minimally invasive tumor ablative procedures have been investigated and modeled in this dissertation. The lung frequently undergoes continuous large respiratory deformation, which can greatly affect the pre-planned outcome of the operation, hence deformation compensation becomes necessary. The first type of major deformation involved in a target lung throughout a tumor ablative procedure is the one encountered in procedures where the lung is totally deflated before starting the operation. A consequence of this deflation is that pre-operative images (acquired while the lung was partially inflated) become inaccurate for targeting the tumor. Another issue is that minimally invasive procedures usually employ intra-operative US imaging for guidance. However, US images of the deflated lung have very poor quality due to the small amount of air remaining in the deflated lung. To address the challenges associated with deflating the lung, a novel construction technique has been proposed in this thesis to obtain CT images of the totally deflated lung. This technique processes the lung’s 4D-CT respiratory image sequence acquired pre-operatively. It consists of a deformable registration/air volume estimation/extrapolation pipeline. The pipeline does not require any external marker as it is capable of estimating the lung’s air volume from the CT images automatically using a newly developed segmentation approach introduced in this thesis. To deal with poor quality issue of the US image, a novel registration strategy has been introduced to enhance the quality of the lung’s intra-operative US image by employing the constructed high quality CT image. The second major type of lung deformation tackled in this thesis is the one due to respiratory anatomical contact forces or needle insertion, which can be characterized using tissue biomechanical models. Two essential prerequisites of developing such models are realistic biomechanical parameters of the lung soft tissue, and proper lung tissue discretization, for which inevitable, yet reasonable, geometry simplification should be incorporated. These two critical necessities have been investigated in the last two parts of this thesis. The results reported in these parts have paved the way for accurate biomechanical modeling of the lung for predicting tissue deformation resulting from contact forces and needle insertion in future studies.