Herein a propagation-based phase contrast x-ray imaging technique for measuring particle size and number is presented. This is achieved with an algorithm that utilizes the Fourier space signature of the speckle pattern associated with the images of particles. We validate this algorithm using soda-lime glass particles, demonstrating its effectiveness on random and non-randomly packed particles. This technique is then applied to characterise lung alveoli, which are difficult to measure dynamically in vivo with current imaging modalities due to inadequate temporal resolution and/or depth of penetration and field-of-view. We obtain an important result in that our algorithm is able to measure changes in alveolar size on the micron scale during ventilation and shows the presence of alveolar recruitment/de-recruitment in newborn rabbit kittens. This technique will be useful for ventilation management and lung diagnostic procedures.