Fusion of magnetic resonance and ultrasound images for endometriosis detection ; La fusion de l'IRM et de l'échographie

Endometriosis is a gynecologic disorder that typically affects women in their reproductive age and is associated with chronic pelvic pain and infertility. In the context of pre-operative diagnosis and guided surgery, endometriosis is a typical example of pathology that requires the use of both magnetic resonance (MR) and ultrasound (US) modalities. These modalities are used side by sidebecause they contain complementary information. However, MRI and US images have different spatial resolutions, fields of view and contrasts and are corrupted by different kinds of noise, which results in important challenges related to their analysis by radiologists. The fusion of MR and US images is a way of facilitating the task of medical experts and improve the pre-operative diagnosis and the surgery mapping. The object of this PhD thesis is to propose a new automatic fusion method for MRI and US images. First, we assume that the MR and US images to be fused are aligned, i.e., there is no geometric distortion between these images. We propose a fusion method for MR and US images, which aims at combining the advantages of each modality, i.e., good contrast and signal to noise ratio for the MR image and good spatial resolution for the US image. The proposed algorithm is based on an inverse problem, performing a super-resolution of the MR image and a denoising of the US image. A polynomial function is introduced to modelthe relationships between the gray levels of the MR and US images. However, the proposed fusion method is very sensitive to registration errors. Thus, in a second step, we introduce a joint fusion and registration method for MR and US images. Registration is a complicated task in practical applications. The proposed MR/US image fusion performs jointly super-resolution of the MR image and despeckling of the US image, and is able to automatically account for registration errors. A polynomial function is used to link ultrasound and MR images in the fusion process while an appropriate similarity measure is introduced ...