C-arms are mobile x-ray imaging devices frequently used in the operating room (OR). Their maneuverability enables surgeons to acquire images in different angles, and potentially recover three dimensional structures of anatomy. In particular, C-arms can be used for tomographic reconstruction that resembles CT. This use, however, is limited due to multiple factors, including hardware cost and physical constraints on the scan protocol.
In this talk, I present an emerging methodology of fusing intra-operative x-rays with pre-operative anatomical data, which may be based on a patient-specific CT scan or on statistical anatomical models, to compensate for information loss due to the scan constraints. The reconstructed images cover the range between a “best guess” of the 3D structure, a complete cone-beam reconstruction, and a fused volume that’s part one and part the other. Importantly, the method allows us to significantly improve the reconstruction from the observed x-ray images while not losing information in the gap between the x-rays and the prior model.
As a result, we can perform plausible reconstructions using a relatively low-end, ubiquitous C-arm rather than the high-end systems which are currently in clinical use.
The talk surveys various concepts and elements used in the implementation of the method, including image registration, volumetric modeling, visualization algorithms, and calibration of x-ray systems, all required in the final goal which we name “hybrid reconstruction.”
Main collaborators who contributed to this research include: Gouthami Chintalapani, Lotta Ellingsen, Junghoon Lee, Prof. Jerry Prince, Krishna Ramampurthi, and Prof. Russell H. Taylor.