Processing 3D Scans Using Statistical Shape Analysis and Automatic Pose Correction for Subsequent Orthosis Fitting

In the medical field, 3D-technology enables the creation of individualized medical devices that are tailored to perfectly fit the patient's anatomy. After the acquisition of the patient’s 3D-scan, the data needs to be processed before it can be used to design medical devices. Two of the biggest challenges in processing the 3D-data are patient posture and scan quality, where surface information is distorted by noise or foreign bodies. Automatic patient posture correction can be done in numerous ways, but utilizing a generic template model has several advantages. First of all, the template posture can be set to a particular position by the user, reflecting the therapy administered beforehand. The patient scan will then simply match the posture of the model. Additionally, the position of anatomical features of the patient scan can easily be identified with the help of the template model. Another issue needed to overcome is alternating scan quality, which can dramatically decrease the ability to closely fit an orthopedic aid to the patient scan. With the help of machine learning via statistical shape models (SSM), an algorithm can be trained from a dataset of 3D-scans to reconstruct the mesh without affecting the geometrical features of the patient. Afterwards, the repaired and corrected scan can be used to design and print a custom-made orthopedic aid such as an ankle-foot orthosis (AFO).