The field of medical image processing has made tremendous progress in the past few years thanks to the continuous improvement of imaging technologies as well as the computing power nowadays available on modern computers. These developments benefited various fields of medicine, notably liver surgery, which evolved into a distinct specialty involving multidisciplinary teams composed of hepato-biliary surgeons, oncologists, interventional radiologists and computer vision engineers. Collaboration between these specialists enables the development of innovating solutions which allow surgeons to perform ambitious surgeries while minimizing surgical risks. These procedures can be achieved in particular thanks to better visualization and surgery planning tools. By precisely modeling the anatomy of a patient and analyzing its hepatic profile, it possible to virtually simulate surgical procedures and evaluate their outcome.
A fundamental step in using these virtual surgery planning tools is the precise modeling of the anatomical structures of the liver from radiological images such as CT-Scans and magnetic resonance images (MRI). This step, known as 3D segmentation, consists of delimiting the boundary of organs in medical images in order to produce an accurate 3D representation. These 3D models are essential to evaluate relevant clinical parameters required for surgical planning, such as the volume of the liver or the proximity of malignant lesions to sensible vascular structures of the organ.
The work reported in this thesis deal with the topic of liver segmentation, which even after many years of research remains an open problem. A first part proposes a generic interactive segmentation tool and presents its application to various 2D segmentation tasks. The second part proposes a whole liver segmentation tool which is thoroughly evaluated on CT-Scan and MRI. Finally, a hepatic vascular modeling methodology is presented, allowing to delimit the different perfusion and draining areas of the liver. Throughout these chapters, a special attention is given to MRI, to efficiency of the proposed tools and to the clinical adoption potential.
| Date | 7 Feb 2017 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Jacques A. de Guise (Supervisor) & An Tang (Co-supervisor) |
|---|
Chartrand, G. (Author),
De Guise, J. (Supervisor) & Tang, A. (Co-supervisor),
7 Feb 2017Student thesis: Doctoral thesis › Doctorate in Engineering: Engineering