Aim of the study was the evaluation of the efficacy of the use of a new wearable AR video see-throught system based on Head Mounted Displays (HMDs) to guide the position of a working cannula into the vertebral body through a transpedicular approach without the use X-Ray images guidance.

We describe a head mounted stereoscopic video see-through display that allows the augmentation of video frames acquired by two cameras with the rendering of patient specific 3D models obtained on the basis of pre-operative radiological volumetric images. The system does not employ any external tracker to detect movements of the user or of the patient. User's head movements and the consistent alignment of the virtual patient with the real one, are accomplished through machine vision methods applied on pairs of live images.

Our system has been tested on an experimental setup that simulate the reaching of lumbar pedicle as in a vertebral augmentation procedure avoiding the employment of ionizing radiation. Aim of the study is to evaluate the ergonomics and the accurancy of the systems to guide the procedure. We performed 4 test sessions with a total of 32 kirschner wire implanted by a single operator wearing the HMD with the AR guide. The system accurancy was evaluated by a post-operative CT scan.

The most ergonomic AR visualization comprise the use of a pair of virtual viewfinders (one at the level of the skin entry point and one at the level of the trocar's bottom) aligned according to the planned direction of the trocar insertion. With such AR guide the surgeon must align the tip of the needle to the center of the first viewfinder placed on the patient's skin. indeed the viewfinder barycenter provides a 2 degrees of freedom (DoFs) positioning guide corresponding to the point of insertion preoperatively planned over the external surface of the model. The second viewfinder is used by the surgeon to rotate and align the trocar according to the planned direction of insertion (2 rotational DOFs). After the first test series a clamping arm has been introduced to maintain the reached trocar's trajectory.

The post-operative CT scan was registered to the preoperative one and the trajectories obtained with the AR guide were compared to the planned one. The overal results obtained in the 4 test session show a medium error of 1.18+/−0.16 mm.

In the last year there was a growing interest to the use of Augmented Reality systems in which the real scene watched by the surgeon is merged with virtual informations extracted from the patient's medical dataset (medical data, patient anatomy, preoperative plannig). Wearable Augmented Reality (WAR) with the use of HDMs allows the surgeon to have a “natural point of view” of the surgical field and of the patient's anatomy avoiding the problems related to eye-hand coordination.

Results of the in vitro tests are encouraging in terms of precision, system usability and ergonomics proving our system to be worthy of more extensive tests.

Pedicle screws fixation to stabilize lumbar spinal fusion has become the gold standard for posterior stabilization. However their positioning remain difficult due to variation in anatomical shape, dimensions and orientation, which can determine the inefficacy of treatment or severe damages to close neurologic structures. Image guided navigation allows to drastically decrease errors in screw placement but it is used only by few surgeons due to its cost and troubles related to its using, like the need of a localizer in the surgical scenario and the need of a registration procedure. An alternative image guided approach, less expensive and less complex, is the using of patient specific templates similar to the ones used for dental implants or knee prosthesis.

Like proposed by other authors we decided to design the templates using CT scans. (slice thickness of 2.0 mm). Template developing is done, for each vertebra, using a modified version of ITK-SNAP 1.5 segmentation software, which allow to insert cylinders (full or empty) in the segmented images. At first we segment the spine bone and then the surgeon chose screw axes using the same software. We design each template with two hollow cylinders aligned with the axes, to guide the insertion in the pedicle, adding contact points that fit on the vertebra, to obtain a template right positioning. Finally we realize the templates in ABS using rapid prototyping. After same in-vitro tests, using a synthetic spine (by Sawbones), we studied a solution to guarantee template stability with simple positioning and minimizing intervention invasiveness. Preliminary ex-vivo animal testing on porcine specimens has been conducted to evaluate template performance in presence of soft-tissue in place, simulating dissection and vertebra exposure. For verification, the surgeon examined post-operative CT-scans to evaluate Kirschner wires positioning.

During the ex-vivo animal test sessions, template alignment resulted easy thanks to the spinous process contact point. Their insertion required no additional tissue removal respect to the traditional approach. The positioning of contact points on vertebra's lamina and articular processes required just to shift the soft tissue under the cylinders bases. The surgeon in some cases evaluated false stable template positions since not each of the 4 contact points were actually in contact with the bone surface and tried the right position. CT evaluation demonstrate a positive results in 96.5% of the Kirschner wires implanted.

Our approach allows to obtain patient specific templates that does not require the complete removal of soft tissue around vertebra. Guide positioning is facilitated thanks to the using of the spinous processes contact point, while false stable positions can be avoided using four redundant contact points. The templates can be used to guide the drill, the insertion of Kirschner in case of use of cannulated screws or to guide directly the screw. After these preliminary ex-vivo animal tests we obtained the authorization of the Italian Health Ministry to start the human study.