Giuseppe Loggia

Mr. Giuseppe Loggia


Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 June 2025 in Computers in Biology and Medicine
Verified
Journal Article
Open access
LWLuohong Wu
Luohong Wu
NCNicola A. Cavalcanti
Nicola A. Cavalcanti
MSMatthias Seibold
Matthias Seibold
GLGiuseppe Loggia
Giuseppe Loggia
LRLisa Reissner
Lisa Reissner
PFPhilipp Fürnstahl
Philipp Fürnstahl
Ultrasound-based bone surface segmentation is crucial in computer-assisted orthopedic surgery. However, ultrasound images have limitations, including a low signal-to-noise ratio, acoustic shadowing, and speckle noise, which make interpretation difficult. Existing deep learning models for bone segmentation rely primarily on costly manual labeling by experts, limiting dataset size and model generalizability. Additionally, the complexity of ultrasound physics and acoustic shadow makes the images difficult for humans to interpret, leading to incomplete labels in low-intensity and anechoic regions and limiting model performance. To advance the state-of-the-art in ultrasound bone segmentation and establish effective model benchmarks, larger and higher-quality datasets are needed. We propose a methodology for collecting ex-vivo ultrasound datasets with automatically generated bone labels, including anechoic regions. The proposed labels are derived by accurately superimposing tracked bone Computed Tomography (CT) models onto the tracked ultrasound images. These initial labels are refined to account for ultrasound physics. To clinically evaluate the proposed method, an expert physician from our university hospital specialized in orthopedic sonography assessed the quality of the generated bone labels. A neural network for bone segmentation is trained on the collected dataset and its predictions are compared to expert manual labels, evaluating accuracy, completeness, and F1-score. We collected UltraBones100k, the largest known dataset comprising 100k ex-vivo ultrasound images of human lower limbs with bone annotations, specifically targeting the fibula, tibia, and foot bones. A Wilcoxon signed-rank test with Bonferroni correction confirmed that the bone alignment after our optimization pipeline significantly improved the quality of bone labeling (p<0.001). The model trained on UltraBones100k consistently outperforms manual labeling in all metrics, particularly in low-intensity regions (at a distance threshold of 0.5 mm: 320% improvement in completeness, 27.4% improvement in accuracy, and 197% improvement in F1 score) CONCLUSION:: This work is promising to facilitate research and clinical translation of ultrasound imaging in computer-assisted interventions, particularly for applications such as 2D bone segmentation, 3D bone surface reconstruction, and multi-modality bone registration.

6 April 2025 in The Spine Journal
Verified
Journal Article
Open access
GLGiuseppe Loggia
Giuseppe Loggia
MFMazda Farshad
Mazda Farshad
MJMoritz Jokeit
Moritz Jokeit
JWJonas Widmer
Jonas Widmer
SDStefani Dossi
Stefani Dossi
Marco D. BurkhardMarco D. Burkhard
Marco D. Burkhard
Adjacent segment disease (ASDis) and degeneration (ASDeg) are common complications following lumbar fusion, ranging from asymptomatic radiographic changes to debilitating symptoms requiring revision surgery. While the impact of spinopelvic alignment (SPA) on postoperative outcomes and ASDis prevention is well-studied in long-construct fusions, its role in short-segment lumbosacral fusions, particularly over the long term, remains unclear. This study aimed to investigate the association between global and distal lumbar SPA with the development of ASDis and ASDeg in the long-term follow-up of patients undergoing short-segment lumbosacral fusion. Secondary outcomes were patient reported outcome measures (PROMs) in relation to adjacent segment changes. Retrospective single-center cohort study with minimum follow-up of 5 years. A total of 86 patients who underwent L4-S1 spinal fusion between 2003 and 2015, with a mean follow-up of 12±4 years. The primary outcome was the development of adjacent segment changes, classified into 2 groups: (1) surgically-treated ASDis, and (2) ASDeg, defined as radiographic evidence of adjacent segment changes without surgical intervention. Secondary outcomes included PROMs: the Oswestry Disability Index (ODI) and the European Quality of Life 5 Dimensions 5 Level (EQ5D5L). Pre- and postoperative standing radiographs were annotated and the following global lumbar SPA parameters analyzed: Pelvic incidence (PI), sacral slope (SS), pelvic tilt (PT), lumbar lordosis (LL), PI-LL-mismatch, lumbar pelvic angle (LPA). Additionally, the following distal lumbar SPA parameters were analyzed: Distal lordosis (DL) between L4-S1, lordosis distribution index (LDI=DL/LL), PI-DL-difference, DL-PI-ratio, adjacent segment lordosis (ASL). Postoperative ODI and EQ5D5L were assessed via a telephone interview. Among 86 patients with a mean follow-up of 12±4 years the incidence of ASDis was 27.9% (n=24), while 7.0% (n=6) showed ASDeg. No significant differences in pre- and postoperative global lumbar SPA were found in patients who developed ASDis or ASDeg compared to patients who did not. PI (53.6° vs 59.6°) and preoperative SS (34.9° vs 40.3°) trended to be lower in the ASDis group compared to controls, although nonsignificant. Pre- and postoperative distal lumbar SPA parameters did not differ between groups. Patients with adjacent segment changes, regardless of whether they underwent revision surgery, reported worse PROMs. ODI scores were elevated in both the ASDis group (28, IQR 15-42) and the ASDeg group (34, IQR 14-47) compared to controls (10, IQR 2-24; p=.005). EQ5D5L scores were lower in both the ASDis (70, IQR 53-83) and the ASDeg groups (60, IQR 55-85) compared to controls (85, IQR 75-90; p=.025). In this long-term follow-up study of patients undergoing short-segment lumbosacral fusion, neither preoperative nor postoperative SPA was significantly associated with the development of ASDis or ASDeg. These findings suggest that other factors may have a greater influence on ASDis risk. Patients who developed adjacent segment changes, whether or not they required revision surgery, experienced worse long-term postoperative PROMs, indicating the importance of ASDis prevention.

5 June 2025 in Journal of the American Academy of Orthopaedic Surgeons
Verified
Journal Article
Closed access
GLGiuseppe Loggia
Giuseppe Loggia
FAFranziska C. S. Altorfer
Franziska C. S. Altorfer
FAFedan Avrumova
Fedan Avrumova
DLDarren R. Lebl
Darren R. Lebl
Robotic-assisted (RA) spine surgery has markedly improved accuracy and consistency across a wide range of spinal procedures. However, successful implementation requires understanding and addressing potential complications. This review systematically analyzes potential pitfalls and their prevention strategies in RA spine surgery and proposes a structured approach to navigate these complexities. Key challenges identified include registration errors, which may be reduced through high-resolution imaging and stable patient positioning; trajectory inaccuracies, which can be addressed by ensuring robust mounting and optimizing robotic arm movements; and anatomical variations, which benefit from thorough preoperative planning and tailored instrumentation. Strategies to minimize skiving include the use of specialized tools and enhanced haptic feedback to improve trajectory control. Soft-tissue interference can be managed through careful approach selection and retraction techniques, whereas intraoperative motion may be minimized with secure reference arrays and structured team workflows. In addition, technical failures highlight the importance of systematic software testing and regular hardware maintenance. RA spine surgery holds enormous future potential, but its broader adoption may benefit from a comprehensive understanding of these challenges. By systematically assessing these factors, this framework may help surgeons refine the safety and efficiency of RA spine procedures while reducing potential risks.

Verified
Journal Article
Open access
PDF file
GLGiuseppe Loggia
Giuseppe Loggia
FAFedan Avrumova
Fedan Avrumova
DLDarren R. Lebl
Darren R. Lebl
Introduction: Robotic-assisted (RA) spine surgery enhances pedicle screw placement accuracy through real-time navigation and trajectory guidance. However, the absence of traditional direct haptic feedback by freehand instrumentation remains a concern for some, particularly in minimally invasive (MIS) procedures where direct visual confirmation is limited. During RA spine surgery, navigation systems display three-dimensional data, but factors such as registration errors, intraoperative motion, and anatomical variability may compromise accuracy. This technical note describes a visuohaptic intraoperative phenomenon observed during RA spine surgery, its underlying mechanical principles, and its utility. During pedicle screw insertion with a slow-speed automated drill in RA spine procedures, a subtle and rhythmic variation in resistance has been observed both visually on the navigation interface and haptically through the handheld drill. This intraoperative pattern is referred to in this report as a cyclical insertional torque (CIT) pattern and has been noted across multiple cases. The CIT pattern is hypothesized to result from localized stick–slip dynamics, where alternating phases of resistance and release at the bone–screw interface generate periodic torque fluctuations. The pattern is most pronounced at low insertion speeds and diminishes with increasing drill velocity. CIT is a newly described intraoperative observation that may provide visuohaptic feedback during pedicle screw insertion in RA spine surgery. Through slow-speed automated drilling, CIT offers a cue for bone engagement, which could support intraoperative awareness in scenarios where tactile feedback is reduced or visual confirmation is indirect. While CIT may enhance surgeon confidence during screw advancement, its clinical relevance, reproducibility, and impact on placement accuracy have yet to be validated.

2 October 2025 in HSS Journal®: The Musculoskeletal Journal of Hospital for Special Surgery
Verified
Journal Article
Closed access
GLGiuseppe Loggia
Giuseppe Loggia
FAFranziska C. S. Altorfer
Franziska C. S. Altorfer
FAFedan Avrumova
Fedan Avrumova
CACeleste Abjornson
Celeste Abjornson
JZJiaqi Zhu
Jiaqi Zhu
DLDarren R. Lebl
Darren R. Lebl
Background: Cervical disc arthroplasty (CDA) has gained widespread recognition as a motion-preserving alternative to anterior cervical discectomy and fusion for treating degenerative cervical disc disease. Over the past decade, the rate of CDA procedures has increased by a remarkable 654%. However, CDA complications can lead to a repeat procedure or can be converted to fusion. Purpose: We sought to evaluate the time of onset (early, intermediate, or late) of complications following CDA as well as the progression leading to revision surgeries. Methods: A retrospective review of the Food and Drug Administration Manufacturer and User Facility Device Experience (MAUDE) database examined complications across 9 CDA devices from January 1, 2005, to September 30, 2023. Complications were categorized by postoperative time to occurrence: early (18 months). The revision procedures were categorized as either a repeat CDA or conversion to fusion. Results: A total of 688 CDA revision cases were reviewed, with 265 cases (38.5%) reporting revision onset time (median: 12 months, interquartile range: 3-36). Migration was the most frequent complication in both the early (n = 44, 53.0%) and intermediate (n = 25, 32.1%) periods. Subsidence peaked during the intermediate period (11.5%, n = 9), while neck pain and heterotopic ossification (HO) occurred most in the late postoperative phase, at 22.5% (n = 23) and 7.8% (n = 8), respectively. The specific revision type was recorded in 509 of the revision cases; of these, 8 devices were predominantly associated with fusion (>69%), while Mobi-C showed a lower proportion of fusion revisions (16.2%). Conclusions: This retrospective review of the MAUDE database showed the timing of complications leading to CDA revision. Migration predominated in the early and intermediate phases, while subsidence peaked in the intermediate phase. In the late phase, neck pain, osteolysis, and HO become more prominent. Conversion to fusion was common across most devices. These data provide insight for clinicians to monitor patients after CDA while considering the relative frequencies of reported complications. Level of Evidence: Level IV: retrospective analysis of a prospectively maintained government database.

12 November 2025 in Global Spine Journal
Verified
Journal Article
Open access
GLGiuseppe Loggia
Giuseppe Loggia
FAFedan Avrumova
Fedan Avrumova
Marco D. BurkhardMarco D. Burkhard
Marco D. Burkhard
MKMichael J. Kelly
Michael J. Kelly
FAFranziska C. S. Altorfer
Franziska C. S. Altorfer
DLDarren R. Lebl
Darren R. Lebl
Study Design: Comparative cadaveric validation. Objectives: To evaluate the influence of prior surgical experience on the accuracy and efficiency of robotic-assisted muscle-preserving (RAMP) decompression using human specimens. Methods: RAMP decompressions were performed from T8 - L5 in 8 cadavers by a non-fellowship-trained surgeon (NFTS) and a fellowship-trained surgeon (FTS). Computed tomography (CT) images were used for preoperative robotic planning and registration. Operative durations were documented, postoperative CT imaging was used to evaluate decompression accuracy and quantified anterior cortical bone removal (ACBR). Results: A total of 80 levels underwent RAMP decompression (40 thoracic, 40 lumbar). Median single-level operative times were 2.6 min (IQR 1.9-3.7) for NFTS and 2.1 min (IQR 1.6-3.0) for FTS, with no significant difference ( P = 0.109). Decompression accuracy demonstrated submillimeter deviation from planned margins, with posterior deviation measuring 0.6 mm (IQR 0.5-1.5) for NFTS and 0.7 mm (IQR 0.4-1.0) for FTS ( P = 0.573), and anterior deviation measuring 0.3 mm (IQR 0.1-0.5) in both ( P = 0.938). ACBR did not differ significantly between NFTS and FTS. Operative times declined with experience in both surgeons. Initial single-level RAMP decompressions showed greater variability (NFTS: 3.13-8.10 min; FTS: 1.39-10.56 min), while later cases converged to narrower ranges (NFTS: 1.18-2.46 min; FTS: 1.16-2.11 min). Conclusions: This first comparative cadaveric study evaluating surgeon experience in the novel RAMP decompression demonstrates high accuracy and incrementally improving efficiency regardless of prior robotic experience. These findings support the integration of decompression into robotic workflows and pave the way for a broader applicability of high-precision, muscle-preserving techniques in spine surgery.