Guest talk by Dr. Mohammad I. Daoud

MAS guest researcher Dr. Mohammad I. Daoud from the German-Jordanian University / Departement of Computer Engineering in Amman is giving a talk about an applied topic in image data modelling, acquisition and interpretation: “Three-Dimensional Computational Modelling of Preclinical Ultrasound Cancer Imaging”

• Date: 02.10.2014 (Thursday)
• Time: 14:00 – 15:00
• Room: A102 (Bonn-Rhein-Sieg University, Sankt Augustin)

 
Abstract 
Substantial progress has been made toward using high-frequency ultrasound imaging to track tumour growth, but the relationships between high-frequency ultrasound images and tissue microanatomy are incompletely understood. A parallel three-dimensional (3D) ultrasound simulator and a 3D tissue microanatomical model are developed to investigate these relationships. The ultrasound simulator uses a 3D formulation of a k-space numerical method to compute wavefront propagation and runs on distributed-memory computer clusters to enable imaging simulations with short running time. The accuracy of the simulator is demonstrated by computing scattering from fluid spheres and comparing the results with matching analytical solutions. The microanatomical model treats tissue as a population of stochastically positioned cells, where each cell is represented as a spherical nucleus surrounded by cytoplasm. The model is employed to represent the microstructure of healthy mouse liver and an experimental liver metastasis. Normal and cancerous tissue specimens stained with DAPI and H&E are digitized at 20x magnification and analyzed to specify values of the model parameters. For each simulated tissue, the spatial organization of cells is controlled by a Gibbs-Markov point process. The parameters of the Gibbs-Markov process are tuned to reproduce the number density and distribution of center-to-center spacing of nuclei in the DAPI-stained slides of the corresponding experimental tissue specimen. The ultrasound simulator is used to synthesize B-mode images of the simulated healthy and tumour tissues. The first-order speckle statistics of the images of each simulated tissue are compared with corresponding experimental images. The simulations show good matching between the images of the simulated healthy tissue and images of healthy liver. Moreover, good matching is achieved between the images of the simulated tumour and matching experimental images when acoustic properties are used that are different from the values assumed for healthy tissue. These simulations suggest that changes in the first-order speckle statistics that accompany tumour progression are related to variations in tissue acoustic and microstructural properties.