Individual estimation of X ray dose received by patient during digital mammography

PhD mammography

On October 22th, Sophie Ribes has received her Ph.D. degree in radiophysics and medical imaging from Paul Sabatier University of Toulouse. Her work was founded and co-supervised by Magellium. The purpose of the thesis was the individual estimation of the X-ray dose received by patients during mammography.

Sophie has developed a methodology to estimate the 3D configuration of breast components (skin, gland, muscle and fat) during the mammography exam and estimate by Monte-Carlo simulation the radiation received by the gland.

The 3D volume is estimated using MRI imaging of the breast that is automatically segmented into the four anatomic components that are then numerically distorted using finite elements to bring the MRI configuration into the mammography geometric configuration. Her work shows the limits of the current official protocols for dose estimation and opens the door to individual follow-up. In particular, she has shown on a limited set of cases that such protocols underestimate the dose. This is a serious issue (risk-benefice balance) for asymptomatic patients with high-risk who need to go through mammography exams on a regular basis.

Automatic MRI segmentation into the four anatomic components has been published in IEEE Transactions on Medical Imaging (Y-MI). Parts of this work have been performed with INGRID Magellium software to implement the processing chains and to benchmark the developed image processing algorithms.

Keywords: breast, senology, numerical mammography, dosimetry, individualisation, MRI, automatic segmentation, Markov random fields, anisotropic diffusion, K-means, denoising, biomechanical modelling, finite elements, non-linear deformation, Monte-Carlo simulation, X-ray.

Ribes, S., “Development of a method for the determination of individual absorbed dose in digital mammography using multimodal imaging”, Ph.D. thesis in radiophysics and medical imaging, supervised by Olivier Caselles, Paul Sabatier University of Toulouse, 2014.

Ribes, S., Didierlaurent, D., Decoster, N., Gonneau, E., Risser, L., Feillel, V. & Caselles, O. (2014). “Automatic segmentation of breast MR images through a Markov random field statistical model.” IEEE Trans Med Imaging (