Automated Analysis of Ultrasound Images for the Assessment of Pelvic Floor Dysfunction

Pelvic organ prolapse affects half of the women over 50, with one in four having disturbing symptoms. Other pelvic floor dysfunctions are urinary incontinence and fecal incontinence, which is very debilitating. Transperineal 3D/4D ultrasound is increasingly used to gain insights into the biomechanical support of the pelvic floor. In clinical practice, standardized measurements are often made manually leaving a lot of room for interpretation while being labor intensive. As such, our team recently proposed a semi-automatic algorithm to make this process more practical. We are currently also documenting the direct relationship between vaginal delivery trauma and these dysfunctions, using ultrasound as well as magnetic resonance. Also, imaging is used to measure postoperative changes for repairs of these pelvic floor dysfunctions, both clinically and experimentally. The aim of this PhD position is to further automate ultrasound measurements using state-of-the-art machine learning, segmentation and image analysis techniques. Moreover, together with our industrial partner, the main goal is to translate and integrate these techniques into commercial software implemented on high end ultrasound scanners. Next to the technical developments to be made, the student is also expected to lead validation efforts in an already acquired large patient database as well as to improve the workflow and optimize the user experience with the software. Over time, also MR imaging may be included.