4711879_monash_158647.pdf (139.9 MB)
Download file

Phase contrast x-ray imaging for quantifying pulmonary form and function

Download (139.9 MB)
posted on 02.03.2017, 01:09 authored by Leong, Andrew Fwu Tay
This thesis presents newly developed propagation-based phase contrast x-ray (PB-PCX) imaging-based methods for studying lung form and function. The structure of the lungs is highly complex and arguably even more so its respiratory behavior. Whilst many imaging-based advances have been made to provide insight into the structure and mechanics of the lung, none have yet possessed the capabilities to render highly detailed images of the lungs and provide real-time tracking of its behavior. Achieving greater insight into the structure-function relationship of the lungs can potentially lead to more accurate and sensitive diagnostic tools for respiratory diseases. Moreover, it can help design safer and more effective ventilation strategies for patients in respiratory distress. PB-PCX imaging provides strong soft tissue contrast to enable the fine features of the lungs visible, including the intricate network of the airways, and provide real-time imaging. The radiation dose per image is no more and potentially less than conventional x-ray imaging. These properties motivate the work presented here in utilizing PB-PCX imaging for developing methods to study the lungs. In many other lung imaging techniques, a contrast agent or a large radiation dose are often required to visualize lung tissue. Chapter 3 presents a method that involves aligning two PB-PCX chest images to segment the bony anatomy and isolate the lungs before applying the single image phase retrieval algorithm (SIPRA) to regionally measure the relative change in lung air volume. This expands on a previous method that utilizes only SIPRA, which was found to be accurate only for measuring regional lung air volume across large areas of the lungs. From PB-PCX chest images of rabbit kittens being ventilated while immersed in a water-filled tube (this is required to implement SIPRA), regional lung air volume is found to be less accurately measured using the previous method (SIPRA only) than the improved method (bone segmentation and SIPRA). This justifies the importance of segmenting the bones to perform local measures of lung air volume and validates the bone segmentation algorithm developed here. Applying the improved method to a mechanically ventilating rabbit, volumetric maps show lung aeration can be highly heterogeneous. The drawback of the new method described above is that as the lung air volume increases, the bones are less accurately aligned and segmented, resulting in errors in regional lung volume measures. A different approach is taken by representing the PB-PCX chest image in Fourier space. Since the bones and airways of the lungs are of different length scales, the latter being much smaller in dimensions than the former, they occupy different bands of spatial frequencies. The signal corresponding to the airways is known as lung speckle since they appear as a spatially random distribution of bright and dark intensity spots in PB-PCX chest images. Focusing only on the spatial frequencies belonging to the lung, chapter 4 presents a theoretical model of the lung speckle power spectrum based on the solution to Helmholtz equation while treating the lung as a random distribution of spherical voids embedded in soft tissue. This model is validated in simulated PB-PCX lung images using the angular spectrum formulation of scalar diffraction integrals. It shows that the integral over the domain of spatial frequencies occupied by the lung is dependent on lung air volume. This fact has enabled a relationship to be determined between these two parameters by calibrating them using PB-PCX images of mechanically ventilated rabbit kittens in water-filled tubes. The calibration curve is used to measure lung air volumes from PB-PCX chest images of rabbit kittens without having to immerse the animals in water-filled tubes, and shows strong agreement with that measured from a gold standard technique (flowmeter). Besides avoiding needing to align the bones to segment them from the lungs, a higher signal-to-noise ratio is achieved due to removing x-ray attenuation from water in the tube. In a final study, this thesis shows that the integral of the lung speckle power spectrum encodes information about the number and size of alveoli in the lung. Chapter 5 presents a method that extracts this information based on the theoretical model developed in chapter 4. That model assumes the alveoli are uniformly randomly distributed, but at increasing lung air volume the alveoli become closely packed and give rise to short-range ordering; hence the underlying theory is generalized to account for short-range-ordering. However, additional information on the radial distribution of the alveoli is required to adopt this more general theoretical model into this method. To determine whether it was necessary to account for short-range-ordering, PB-PCX imaging experiments were performed on samples of glass microspheres of known size. It is shown that the method using the original model is robust against the effects of short-range-ordering, thereby it can be used to accurately measure the number and dimensions of alveoli in the lungs over a large range of lung air volumes. The reason is found to be that short-range-order affects the shape of the lung speckle power spectrum but not its integral, thus avoiding needing to use the theoretical model that accounts for short-range-order. This method is applied to rabbit kittens and shows the presence of alveolar recruitment/de-recruitment, highlighting that alveoli may open/collapse instead of just varying in size to accommodate the flow of air. Findings such as this will help shape how diagnosis respiratory diseases and ventilation strategies are improved. 

Awards: Vice-Chancellor’s Commendation for Doctoral Thesis Excellence in 2015.


Campus location


Principal supervisor

Marcus John Kitchen

Year of Award


Department, School or Centre

Physics and Astronomy


Faculty of Science