Skip to main navigation Skip to search Skip to main content

Modélisation rénale 4D par approche volumétrique séquentielle et 2D ultrarapide en IRM

Translated title of the thesis: Renal 4D modeling by sequential volumetric approach and 2D ultrafast MRI
  • Mai Taing Wendy Chung

Student thesis: Master's thesis › Master in Engineering: Engineering

Abstract

Techniques of kidney tumor tracking during the delivery of a treatment of high intensity focused ultrasound (HIFU) are not quite explored yet because of the difficulties caused by the movements of the kidney during the treatment. The experimental approach of this project required a 3D MRI acquisition prior to treatment and 2D acquisitions during treatment. The objective was to register the 2D images with the 3D acquisition in order to propose a tracking method of a kidney and a tumor during a HIFU treatment by medical imaging with only 2D images. A kidney movement analysis has shown that the movement is decomposed into a significant translation in the cranio-caudal axis (compared to the two other axes) and a variable rotation from a subject to another. Then, a 2D/3D image registration by intensiy method has been proposed. To evaluate the efficiency of this method, two merit functions have been used : the cross correlation and the mutual information. Combining these two functions with the Downhill Simplex optimization function, the results showed that the correlation coefficients are greater with the cross correlation function compared to the mutual information function. This registration method has been tested with rotated images along the three axes to simulate small displacements of a subject during a treatment. The correlation coefficients were in average 96% with a calculating time of 11 seconds. Lastly, the coordinates of the center of the tumor have been located in real time from the 2D images by setting a reference point in these 2D images.
Date5 Apr 2016
Original languageFrench
Awarding Institution
  • École de technologie supérieure
SupervisorJacques A. de Guise (Supervisor) & Philippe Souteyrand (Co-supervisor)

Cite this

'