Required Hardware and Data
Designing this solution required a variety of data and hardware. First, to create a transformation between the trackers and the bones of interest, our team needed to use some data format that would allow this to be possible. Because a pre-surgery CT scan is already a standard process, we found it useful to ask for a CT scan that includes both the bones and the trackers. From here, we could iterate through the slices of the scan to find the bones and trackers.
The first piece of data needed was a CT scan of the desired operation area. More specifically, it is crucial that this CT scan was from an operation where these 3D trackers are in place. Without these in place, it would be impossible to create a registration between the trackers and the bones of interest. This data type was limited, as this tool has never been used in a real application. To bypass this issue, our team requested a CT scan of a model spine with the trackers in place. Since it was also important to accurately register the bone in the CT scans, our team found it useful to acquire additional CT scans, even though they do not have the 3D markers in place. This data can be used to test our bone registration algorithm’s robustness. Although these were not useful for creating a relationship between the bone and the tracker, the abundance of this data type was critical for refining the registration algorithm. One frame of such a CT scan is displayed below in Figure 1. Like the other CT scans, these were provided by Dr. Pallotta.
The NDI Cygna-6D camera, along with the 3D trackers are the devices that are responsible for feeding position and rotation data into our program, such that it can be transformed to model the bones. These two pieces of hardware are displayed below in Figure 2.
These pieces of hardware, in union with the prepackaged software, output data in the format R^6, where each element of V is a positional or rotation measurement. An additional vector is included in the output for every tracker in the frame of the camera. This data type was easily collected by our team, as we had all the hardware and software needed for collection.
The main algorithms involved in the completion of the project deliverables were separated into three parts: Tracker Classification Algorithm (TCA), Bone Classification Algorithm (BCA), and Combining the TCA and BCA.