Statistical Chest Shape Modeling: Application to the Evaluation of Pectus Excavatum Treatment Outcomes

Monitoring the progress of medical treatment is paramount to its success and the patient's quality-of-life. An efficient, cost-effective, and accurate assessment of the treatment's progress that illustrates its tangible impact can motivate patients and guide the procedure. This work discusses such an assessment tool for both surgical and nonsurgical interventions to correct Pectus Excavatum(PE) - a chest wall deformity characterized by depression of the sternum. We previously proposed a novel approach for measuring and evaluating the severity of pectus deformities. Our method uses 3-D optical scans which are analyzed to compute surface deformation via registration-based point comparisons, and a distance map is generated, providing a quantitative representation of chest wall improvements. The work presented in this paper expands on this effort by adding a module to objectively evaluate the aesthetic outcomes. In this study, we developed a statistical 3D chest model based on the Gaussian Process Morphable Model (GPMM) using a limited number of optical surface scans (n=20) of healthy male subjects. The model builds an average chest shape which can be used to evaluate the aesthetic results of PE treatment. We analyzed the impact of different kernel functions in the modeling of the chest variation and report on the models generalization, specificity, and compactness.