Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Giulia Zari has a total of 8 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Accuracy of Video Optical See-Through Augmented Reality Surgical System (VOSTARS) to Guide Craniotomy: A Phantom and a Human Cadaver Study

    by Nicola Montemurro, Marina Carbone, Rosanna Maria Viglialoro, Francesco Simi, Giulia Zari, Sara Condino, Francesco Acerbi

    Background: Augmented reality (AR) may support neuronavigation by displaying patient-specific anatomical information and surgical plans directly within the surgeon’s view of the operative field. Reliable registration between virtual content and patient anatomy remains a critical requirement. Patient-specific templates offer a non-invasive registration strategy, but positioning errors at the facial support may propagate to the surgical target. Methods: We performed a preclinical feasibility evaluation of the Video and Optical See-Through Augmented Reality Surgical System (VOSTARS), a head-mounted AR platform for surgical navigation, using patient-specific template-based registration. Six craniotomies were planned using CT datasets from two fresh-frozen human cadaver heads and two corresponding cranial phantoms were generated from the same datasets. Visualization accuracy, reported as target visualization error (TVE3D), and trajectory-tracing performance were first assessed on the phantoms. The complete AR-guided planning-to-execution workflow was subsequently evaluated on the corresponding cadaveric specimens by assessing containment of the AR-traced craniotomy trajectories within patient-specific inspection windows corresponding to a tolerance of ±1.5 mm. Results: In the phantom experiments, the mean TVE3D calculated across five of the six planned craniotomies was 1.64 mm, with individual mean values ranging from 1.33 to 1.91 mm. Craniotomy 1 had a mean TVE3D of 6.92 mm; the higher error was attributed to incorrect phantom assembly. Four of the six craniotomy trajectories traced on the phantoms under AR guidance showed complete overlap with the corresponding virtual trajectories. For the other two, trajectory overlap was 62.5% and 67.0%, with a mean deviation from the planned trajectory of 0.79 mm in both cases. In the cadaveric experiments, the complete AR-guided planning-to-execution workflow was completed for all six planned craniotomies. Two of the six AR-traced craniotomy trajectories were completely contained within the inspection windows corresponding to the predefined ±1.5-mm tolerance. Of the remaining four trajectories, two were partially contained, one was minimally contained and one was not contained within the inspection window. Conclusions: This study supports the technical feasibility of a patient-specific AR workflow for craniotomy guidance across anatomically different cranial approaches. In this exploratory preclinical study, VOSTARS enabled the translation of patient-specific preoperative planning directly to the operative field and provided AR guidance during the initial stages of cranial surgery. The cadaveric findings also highlight template fit and positioning as important determinants of workflow performance. Further quantitative and comparative validation is required before clinical implementation.

  • Open Access

    Magic Leap 1 versus Microsoft HoloLens 2 for the Visualization of 3D Content Obtained from Radiological Images

    by Giulia Zari, Sara Condino, Fabrizio Cutolo, Vincenzo Ferrari

    The adoption of extended reality solutions is growing rapidly in the healthcare world. Augmented reality (AR) and virtual reality (VR) interfaces can bring advantages in various medical-health sectors; it is thus not surprising that the medical MR market is among the fastest-growing ones. The present study reports on a comparison between two of the most popular MR head-mounted displays, Magic Leap 1 and Microsoft HoloLens 2, for the visualization of 3D medical imaging data. We evaluate the functionalities and performance of both devices through a user-study in which surgeons and residents assessed the visualization of 3D computer-generated anatomical models. The digital content is obtained through a dedicated medical imaging suite (Verima imaging suite) developed by the Italian start-up company (Witapp s.r.l.). According to our performance analysis in terms of frame rate, there are no significant differences between the two devices. The surgical staff expressed a clear preference for Magic Leap 1, particularly for the better visualization quality and the ease of interaction with the 3D virtual content. Nonetheless, even though the results of the questionnaire were slightly more positive for Magic Leap 1, the spatial understanding of the 3D anatomical model in terms of depth relations and spatial arrangement was positively evaluated for both devices.