Semi-Automatic Segmentation via Ultrasound Imaging: For AR-Guided Laparoscopic Liver Tumor Resections

Abstract

Liver cancer is one of the most prevalent and deadly forms of cancer. The primary treatment for liver cancer is the surgical removal of the tumor, known as resection. While traditionally performed via open surgery, there has been a significant shift towards minimally invasive surgery, which involves smaller incisions and offers advantages such as faster recovery and shorter hospital stays. Despite its benefits, minimally invasive surgery presents challenges in distinguishing between tumor tissue and healthy liver tissue. An effective solution to these challenges is an augmented reality guiding tool that assists surgeons by overlaying a 3D model of the tumor onto the laparoscopic camera view during surgery. This master’s thesis, conducted in collaboration with Navari Surgical, aims to develop a semi-automatic segmentation method based on ultrasound imaging for creating such a 3D tumor model. The method involves preprocessing techniques and a segmentation algorithm implemented in Python. The evaluation, based on relevant metrics, demonstrates that the method performs well with images of high tumor visibility, though challenges remain with images containing multiple tumor areas and those requiring individualized preprocessing parameters. Future work should focus on refining these aspects and incorporating machine learning models to enhance accuracy and usability. This study establishes the feasibility of using ultrasound imaging for liver tumor segmentation and provides a foundation for further research in this area.