Project Details
[Return to Previous Page]Instrumented Mouthguard for use with 3D Avatars
Company: PSU Kraft Lab
Major(s):
Primary: BME
Secondary: CMPSC
Optional: ME
Non-Disclosure Agreement: NO
Intellectual Property: NO
Traumatic brain injury is a significant health concern for many athletes and soldiers. Novel technology that can diagnose, track and aid in the treatment of brain injury is critically needed. The sponsor has been working to create a sensor-enabled, cloud-based computing platform that predicts brain injuries based on sensors implanted in mouthguards. The computing platform is built on Amazon Web Services (AWS) and uses finite element modeling to predict intracranial brain strain – a leading indicator of mild traumatic brain injury. Once a player experiences an impact, the sensor company calls the brain simulation Application Programming Interface (API) and sends the impact data (accelerations) for the player. Once the impact data arrives at the cloud, a finite element mesh is needed to compute the solution. Interestingly, the computing platform uses technology from the gaming industry to create individual-specific avatars which are then used to make the finite element mesh. The anatomical head shape and dimensions play an important role in the biomechanical response of the brain. The purpose of this capstone design project is to extend the study of the accuracy of the Avatar3D technology in representing the true shape of the human head. To make individual-specific finite element models, individuals are asked to take a profile or ‘selfie’ upon account creation. When the selfie is uploaded, it is sent to the Avatar3D Application Programming Interface (API) which transforms the two-dimensional image into a three-dimensional surface. Then, radial basis functions are used to scale a template finite element mesh of the skull and brain to a ‘target’ three-dimensional surface created from the avatar algorithm, which relies on machine learning and machine vision. The basic concept is to generate avatars using the machine learning/machine vision approach and compare that to ‘fully resolved’ approach where laser scanning is used to generate point clouds of the true head shape. This capstone design project will extend and apply the developed algorithms to develop a statistically based characterization of the methods. The team will need to: 1. Conduct background research on existing techniques to compare geometrical shapes and study the current algorithm implemented. The team will be required to make changes as needed. The methods should be reproducible and posted on Github for open access. 2. Review and update Penn State Institutional Review Board (IRB) approval to collect human data. Members of the team will need to become members on the IRB study team and may need to update the current IRB. 3. Collect Data. Obtain at least 30-point cloud scans using a Structure sensor mounted on an iPad. The team is encouraged to collect data from a diverse set of individuals (e.g., sex, ethnicity, hair styles, etc). Work with the sponsor to identify demographic distributions. 4. Provide statistical results on the comparison between laser scans and AI-derived head shapes. Important: comparison on shape AND absolute size are required. 5. (Optional) Development of a web-based tool, written in React.js for uploading Structure Sensor scan and selfie for comparison. For this your script that generates the similarity index should be hosted on a backend web server like, AWS and have frontend user interface. (The sponsor has the cloud-based accounts and web domains that can be used.) At the conclusion of the project, you should be able to provide a statistically based statement about the accuracy of the avatar3d method in representing the true shape of the human head.
