Automated Biomechanical Pipelines for Personalized Diaphragmatic Assessment
Project Overview
This project is building a patient specific digital anatomy pipeline for soft robotic breathing support. It transforms CT scans into 3D diaphragm interface models by combining AI segmentation with anatomical constraints and evidence weighted reconstruction, creating geometry that can be used for personalised actuator design, biomechanical simulation and future surgical planning.
What were the key results of your research project?
1) Developed an automated hybrid AI geometry pipeline that converts thoracic CT data into patient specific diaphragm interface models using lung, organ, skeletal and CT image constraints rather than relying on direct diaphragm segmentation.
2) The pipeline produced three practical outputs: a CT aligned surface for anatomical review, a full reconstructed anatomical surface, and a conservative evidence filtered planning patch for downstream modelling and export.
3) Preliminary reconstructed surfaces had a mean area of approximately 900.2 cm^2, with a mean planning patch of 358.4 cm^2, meaning around 39.4% of the reconstructed surface was retained after confidence filtering.
4) Exported planning geometry was structurally clean in the evaluated outputs, with 0 zero area faces and 0 detected non manifold edges.
5) Overall, the project demonstrated that patient specific CT anatomy can be converted into a measurable, quality controlled 3D interface suitable for further anatomical validation and future finite element and soft robotic actuator modelling.
How do you feel you have benefitted from completing this internship and has it made you consider future career paths?
Completing this internship has given me a much better understanding of how computational modelling, medical imaging and software engineering can be applied to real clinical problems. As a medical student, I am used to thinking about anatomy and physiology from a clinical perspective, but this project pushed me to think about the same problems quantitatively in terms of geometry, modelling assumptions, validation and how patient-specific data can inform device design.
It has also made me much more interested in working at the interface between medicine, engineering and medical technology. I can see myself pursuing a clinical career while remaining involved in medtech research, particularly in areas such as computational biomechanics, surgical planning and medical device development. The internship has shown me that clinicians can contribute meaningfully to technical projects by helping connect engineering methods with clinically relevant questions and constraints.