Researchers from New York have developed a new image-guidance technique that uses augmented reality (AR) to facilitate interventional procedures. They detail how their method may help improve the ease and accuracy of catheter placement into the heart in an article recently published online by PLOS One.
Traditional image guidance for interventional procedures involves looking back and forth between the patient and a 2D monitor displaying continuous x-ray images acquired via fluoroscopy.
This process offers a low-contrast visualization of the surgical site and poor depth perception, making it challenging for clinicians to position catheters precisely, noted first author Jun Liu, PhD, and colleagues from Weill Cornell Medicine in New York. Advanced visualization technologies such as AR and virtual reality (VR), however, have paved the way for more comfortable and effective methods for performing interventional procedures.
"The development of AR and VR technologies enables a more intuitive solution for 3D visualization and provides a transformative human-machine interface," they wrote. "[AR allows] the user to use both eyes to directly view the 3D models instead of looking at a 2D screen while constantly rotating the models."
In the present study, Liu and colleagues proposed a novel approach for image-guided interventional procedures that displays in real-time a hologram of the catheter alongside a virtual 3D model of a patient's heart and spine (PLOS One, July 1, 2019).
They generated the 3D anatomical model by segmenting and reconstructing CT scans of a patient's heart and spine. They also captured the real-time positioning and orientation of the catheter by processing live fluoroscopic images of the catheter, situated near the heart and spine, from two different angles. Then they used a series of mathematical equations to register the virtual 3D model with the 2D fluoroscopic images onto a single coordinate system and imported the combined imaging data into an AR device (HoloLens, Microsoft).
The group's proprietary registration algorithm allowed for accurate image registration between all 35 pairs of fluoroscopic images and CT scans that they examined, with an error of 0.42 mm. This average error was less than half of the 0.88-mm error reported for other commonly used image registration methods.
Finally, the researchers tested the feasibility of using their AR image-guidance system for catheter placement on a 3D-printed model of the heart and spine, which they created based on the same CT scans used for the AR models. When performing the procedure, they followed a predetermined trajectory for catheter insertion that they had plotted onto the virtual 3D model beforehand.
Creation process for a 3D anatomical heart and spine model, from a CT scan (A) to a virtual 3D reconstruction (B, C, D) to a 3D-printed model (E, F). Image courtesy of Liu et al. Licensed under CC BY 4.0.Overall, the clinicians were able to place the catheter with high accuracy at all five distinct puncture locations using the AR headset. On average, the placement of the catheter was 0.29 mm away from the designated entry point -- an error in placement that fell well within the 5-mm standard for safe insertion.
Beyond facilitating the planning and simulation of interventional procedures, the new AR guidance method and 3D-printed models can serve as learning and training tools for residents and fellows, the authors noted.
"Compared to standard interventional technology, the AR system enables the 3D visualization and more user-friendly interface for interventionalists to better understand the heart anatomy," they wrote. "Therefore, it may assist the development of new therapeutic procedures and lower the learning curve for existing procedures."
