Computer Vision
Two Webcams ($60 each)
ArUco Marker (negligible)
Augmented Reality
Meta Quest 3 ($600)
Conclusions
Our project demonstrated the potential viability of low-cost AR navigation systems in minimally invasive spinal surgeries. Using a simplified spinal model, a detection feedback system, and 3D objects generated through CAD modeling and Unity packages, we assessed the accuracy and error deviation of our AR surgical navigation system. Our proposed $500 AR setup holds promise as an alternative to existing spinal surgery navigation systems, with a 20% increase in average accuracy using AR navigation system as shown in Table 1 and the noteworthy decrease in error deviation in unsuccessful attempts with the AR navigation system compared to the baseline measurement shown in Figure 11. In addition, our ArUco tracking system showed promising results with an error less than 3 mm which indicates its viability for usage in our AR surgical navigation system.
One limitation in our project is the usage of webcams that need to be in line-of-sight at all times to properly track the ArUco markers. People or objects blocking the webcam can hinder the ArUco tracking ability of the surgical navigation system.Therefore, the ArUco marker tracking system may increase higher than needed. Furthermore, our surgical navigation system can be less reliable than those that use infrared based tracking systems which do not rely on visible light which increases reliability. Since our system uses regular webcams, a well-lit room is needed to transmit enough visible light to the cameras and accurately track the ArUco markers. Another limitation is the usage of a simple 3D cylinder as opposed to an accurate spine model generated from a real MRI scan. Since we were testing the viability and accuracy of the augmented reality overlay in our surgical navigation system, a simple cylinder model was used to easily pinpoint and measure the accuracy of the augmented reality overlay. Using a realistic spine model generated from a real MRI scan may produce different results in accuracy and error deviation due to the nonuniform structure of the vertebrae. For instance, the vertebrae have protrusions such as the spinous and transverse process which can interfere with the needle when it is inserted into the patient.