3D models derived from medical imaging can be used, for example, for 3D visualization and preoperative surgical planning.
A statistical shape model (SSM) is a mathematical model that identifies anatomical features of complex 3D shapes and describes the variation in shapes of patients in a population. This model can help determine the suitability of a standard treatment or the need for patient-specific care.
The first part of this thesis describes the development of an SSM of the abdominal aortic aneurysms (AAA). By combining conventional CT-measurements, such as aortic neck length and diameter, with SSM-derived features, a model was created that best discriminated patients with and without complications. Machine learning based models demonstrated that treatment outcome could be accurately predicted using preoperative SSM features.
The second part of the thesis focuses on an SSM of the hemipelvis, describing anatomical differences between men and women both qualitatively and quantitatively. This SSM was used for virtual reconstruction of acetabular fractures, which is particularly valuable when the intact contralateral side is unavailable for preoperative surgical planning. An analysis of the fit of standard implants showed that they were more often suboptimal for the female pelvis.
This thesis demonstrated the versality and clinical relevance of SSMs in vascular and trauma surgery.
