32105 Virtual preoperative planning of lung segmentectomies: development of a segment-predicting algorithm
Begeleider(s): Saar Vermijs en dr. Bram Cornette [UZ Gent]

Richtingen: Master of Science in Biomedical Engineering, International Master of Science in Biomedical Engineering, Master of Science in Computer Science Engineering

Probleemstelling:

The goal of developing virtual computer models of organs or organ systems in the human body (i.e. ‘virtual organs’ or ‘digital twins’) is to increase understanding of human anatomy and physiology, and assist in therapy planning. In this thesis, the aim is to develop a so-called digital twin of the lungs, to assist in surgical treatment planning of lung cancer.

In patients with lung cancer who are eligible for surgical treatment, the tumour is resected. Whereas a few year ago this was mostly done by resecting the entire affected lobe (‘lobectomy’), more recently a shift has been noticed towards lung-sparing resection. In this case, not the entire affected lobe is resected, but only the affected segment of that lobe (‘segmentectomy’). The standard segments are shown in Fig. 1. However, this lung-sparing technique comes with new challenges, mainly during preoperative planning.

Figure 1: The lung segments can be standardized (left lung: 7 segments; right lung: 8 segments). However, patient-specific variations often occur.

Figure 1: The lung segments can be standardized (left lung: 7 segments; right lung: 8 segments). However, patient-specific variations often occur.

 

Preoperative planning of the surgery is most often done using a CT-scan of the patient. On a CT-scan, the tumour is identified and the lobes can be distinguished. This makes it clear which lobe should be resected during a lobectomy. For a segmentectomy, however, this is less clear, as the contours of a segment are not visible on the CT-scan. Keeping in mind that these segments can differ between patients, the need for a patient-specific digital twin of the affected lung showing its segments is evident.


Doelstelling:

In this thesis, the student will start from medical images and process these to create patient-specific 3D models (in collaboration with master thesis students from medicine). Then, the student will optimize and extend an existing Python algorithm to divide these 3D models of the lungs into segments (Fig. 2). The student will also work on visualisation options to make the digital twin user-friendly in the clinic practice.

Figure 2: The results of the current Python algorithm dividing the lungs into segments. As can be seen here, there is room for improvement.

Figure 2: The results of the current Python algorithm dividing the lungs into segments. As can be seen here, there is room for improvement.