Gianluca Smerilli

Dr. Gianluca Smerilli

orcid

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

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

Representative Publications

View All Publications
8 February 2022 in Arthritis Research & Therapy
Verified
Journal Article
Open access
PDF file
GSGianluca Smerilli
Gianluca Smerilli
ECedoardo cipolletta
edoardo cipolletta
GSGianmarco Sartini
Gianmarco Sartini
EMErica Moscioni
Erica Moscioni
MCMariachiara Di Cosmo
Mariachiara Di Cosmo
EFEmilio Filippucci
Emilio Filippucci
Background: Deep learning applied to ultrasound (US) can provide a feedback to the sonographer about the correct identification of scanned tissues and allows for faster and standardized measurements. The most frequently adopted parameter for US diagnosis of carpal tunnel syndrome is the increasing of the cross-sectional area (CSA) of the median nerve. Our aim was to develop a deep learning algorithm, relying on convolutional neural networks (CNNs), for the localization and segmentation of the median nerve and the automatic measurement of its CSA on US images acquired at the proximal inlet of the carpal tunnel.Methods: Consecutive patients with rheumatic and musculoskeletal disorders were recruited. Transverse US images were acquired at the carpal tunnel inlet, and the CSA was manually measured. Anatomical variants were registered. The dataset consisted of 246 images (157 for training, 40 for validation, and 49 for testing) from 103 patients each associated with manual annotations of the nerve boundary. A Mask R-CNN, state-of-the-art CNN for image semantic segmentation, was trained on this dataset to accurately localize and segment the median nerve section. To evaluate the performances on the testing set, precision (Prec), recall (Rec), mean average precision (mAP), and Dice similarity coefficient (DSC) were computed. A sub-analysis excluding anatomical variants was performed. The CSA was automatically measured by the algorithm.Results: The algorithm correctly identified the median nerve in 41/49 images (83.7%) and in 41/43 images (95.3%) excluding anatomical variants. The following metrics were obtained (with and without anatomical variants, respectively): Prec 0.86 ± 0.33 and 0.96 ± 0.18, Rec 0.88 ± 0.33 and 0.98 ± 0.15, mAP 0.88 ± 0.33 and 0.98 ± 0.15, and DSC 0.86 ± 0.19 and 0.88 ± 0.19. The agreement between the algorithm and the sonographer CSA measurements was excellent [ICC 0.97 (0.94–0.98)].Conclusions: The developed algorithm has shown excellent performances, especially if excluding anatomical variants. Future research should aim at expanding the US image dataset including a wider spectrum of normal anatomy and pathology. This deep learning approach has shown very high potentiality for a fully automatic support for US assessment of carpal tunnel syndrome.

23 August 2022 in Rheumatology
Verified
Journal Article
Closed access
PDF file
ECedoardo cipolletta
edoardo cipolletta
EFEmilio Filippucci
Emilio Filippucci
AAAbhishek Abhishek
Abhishek Abhishek
JBJacopo Di Battista
Jacopo Di Battista
GSGianluca Smerilli
Gianluca Smerilli
AMAndrea Di Matteo
Andrea Di Matteo
Objectives: To determine an US scanning protocol with the best accuracy for the diagnosis of gout and CPPD in patients with acute mono/oligo-arthritis of unknown origin.Methods: Patients with acute mono/oligo-arthritis in whom a joint aspiration at the most clinically involved joint (target joint) was requested were consecutively enrolled. US was performed in each patient before the arthrocentesis. The accuracy of different US findings and scanning protocols for the diagnosis of gout and CPPD was calculated.Results: A total of 161 subjects were included (32 gout patients, 30 CPPD patients and 99 disease-controls). US findings had a high specificity for gout (0.92–0.96) and CPPD (0.90–0.97), while the sensitivity ranged from 0.73 to 0.85 in gout (double contour sign and tophi, respectively) and from 0.60 to 0.90 in CPPD (hyaline and fibrocartilage deposits, respectively). The US assessment of two joints bilaterally (gout: knees, MTP1 joints; CPPD: knees, wrists) plus the target joint had an excellent diagnostic sensitivity (gout: 0.91, CPPD: 0.93) and specificity (gout: 0.91, CPPD: 0.89). This targeted US scanning protocol yielded to higher diagnostic accuracy compared with the US evaluation of the target joint [gout area under the curve (AUC) 0.91 vs 0.84, P = 0.03; CPPD AUC 0.93 vs 0.84, P = 0.04] unless the target joint was the knee or the MTP1 joint in gout and the knee or the wrist in CPPD.Conclusions: A targeted US scanning protocol of two joints bilaterally plus the target joint showed an excellent accuracy (>90%) for the diagnosis of crystal arthritis in patients with acute mono/oligoarthritis. : A video abstract is available for this article and can be viewed at https://doi.org/10.1093/rheumatology/keac479.

21 July 2020 in Clinical Rheumatology
Verified
Journal Article
Open access
PDF file
GSGianluca Smerilli
Gianluca Smerilli
AMAndrea Di Matteo
Andrea Di Matteo
ECedoardo cipolletta
edoardo cipolletta
SCSergio Carloni
Sergio Carloni
AIAntonella Incorvaia
Antonella Incorvaia
EFEmilio Filippucci
Emilio Filippucci
Objectives: To comparatively assess the sonographic spectrum of carpal tunnel syndrome (CTS) in patients with rheumatoid arthritis (RA) and in patients with idiopathic CTS. Methods: Fifty-seven RA patients and 25 idiopathic CTS patients were consecutively enrolled. The diagnosis of CTS in RA patients was made according to clinical history and examination. The following sonographic findings were assessed at carpal tunnel level: median nerve cross-sectional area (CSA) at the carpal tunnel proximal inlet, finger flexor tendons tenosynovitis, radio-carpal synovitis and intraneural power Doppler (PD) signal. Results: CTS was diagnosed in 15/57 RA patients (26.3%). Twenty-three RA wrists with CTS, 84 RA wrists without CTS and 34 idiopathic CTS wrists were evaluated. The average CSA of the median nerve was higher in idiopathic CTS than in RA wrists with CTS (17.7 mm2 vs 10.6 mm2, p < 0.01). A higher rate of inflammation of synovial structures (flexor tendons sheath and/or radio-carpal joint) was found in RA wrists with CTS compared with those without CTS (p = 0.04) and idiopathic CTS (p = 0.02). Intraneural PD signal was more common in CTS (in both RA and idiopathic CTS) wrists compared with wrists without CTS (p < 0.01). Conclusion: The sonographic spectrum of CTS in RA patients is characterized by an inflammatory pattern, defined by the presence of finger flexor tendons tenosynovitis and/or radio-carpal joint synovitis. Conversely, a marked median nerve swelling is the dominant feature in idiopathic CTS. Intraneural PD signal is a frequent finding in both conditions. Key Points• Carpal tunnel syndrome (CTS) associated with rheumatoid arthritis (RA) and idiopathic CTS have distinct ultrasound patterns.• The most characteristic sonographic features of CTS in RA patients are those indicative of synovial tissue inflammation at carpal tunnel level. Conversely, marked median nerve swelling is the dominant finding in idiopathic CTS.• Intraneural power Doppler signal is a frequent finding in both conditions.• In patients with CTS, differently from electrophysiology, US can provide clues prompting a rheumatology referral in case of prominent inflammatory findings at carpal tunnel level.

24 August 2021 in Current Rheumatology Reports
Verified
Journal Article
Open access
PDF file
GSGianluca Smerilli
Gianluca Smerilli
AMAndrea Di Matteo
Andrea Di Matteo
ECedoardo cipolletta
edoardo cipolletta
WGWalter Grassi
Walter Grassi
EFEmilio Filippucci
Emilio Filippucci
Purpose of Review: To provide an overview of the ultrasound (US) studies focusing on enthesitis in psoriatic arthritis (PsA).Recent Findings: Last-generation US equipment has demonstrated the ability to detect subtle morphostructural and vascular abnormalities at entheseal level. US is able to identify pathologic changes in both “classical” (i.e., the site of attachment of tendons, ligaments, and joint capsules into the bone) and “functional” entheses (i.e., anatomical regions where tendons or ligaments wrap around bony pulleys).Summary: US has the potential to be the first-line method in the assessment of enthesitis. In the present review we critically discussed the current definitions of US enthesitis, the scoring systems, and the main fields of application (i.e., the detection of enthesitis in PsA and psoriasis, the identification of different disease subsets, and the assessment of response to treatment).

16 March 2022 in Rheumatology
Verified
Journal Article
Closed access
PDF file
AMAndrea Di Matteo
Andrea Di Matteo
ECedoardo cipolletta
edoardo cipolletta
GCGiulia Maria Destro Castaniti
Giulia Maria Destro Castaniti
GSGianluca Smerilli
Gianluca Smerilli
CACarla Airoldi
Carla Airoldi
EFEmilio Filippucci
Emilio Filippucci
Objectives: To investigate the reliability of the OMERACT US Task Force definition of US enthesitis in SpA.Methods: In this web exercise, based on the evaluation of 101 images and 39 clips of the main entheses of the lower limbs, the elementary components included in the OMERACT definition of US enthesitis in SpA (hypoechoic areas, entheseal thickening, power Doppler signal at the enthesis, enthesophytes/calcifications, bone erosions) were assessed by 47 rheumatologists from 37 rheumatology centres in 15 countries. Inter- and intra-observer reliability of the US components of enthesitis was calculated using Light’s kappa, Cohen’s kappa, Prevalence And Bias Adjusted Kappa (PABAK) and their 95% CIs.Results: Bone erosions and power Doppler signal at the enthesis showed the highest overall inter-reliability [Light’s kappa: 0.77 (0.76–0.78), 0.72 (0.71–0.73), respectively; PABAK: 0.86 (0.86–0.87), 0.73 (0.73–0.74), respectively], followed by enthesophytes/calcifications [Light’s kappa: 0.65 (0.64–0.65), PABAK: 0.67 (0.67–0.68)]. This was moderate for entheseal thickening [Light’s kappa: 0.41 (0.41–0.42), PABAK: 0.41 (0.40–0.42)], and fair for hypoechoic areas [Light’s kappa: 0.37 (0.36–0.38); PABAK: 0.37 (0.37–0.38)]. A similar trend was observed in the intra-reliability exercise, although this was characterized by an overall higher degree of reliability for all US elementary components compared with the inter-observer evaluation.Conclusions: The results of this multicentre, international, web-based study show a good reliability of the OMERACT US definition of bone erosions, power Doppler signal at the enthesis and enthesophytes/calcifications. The low reliability of entheseal thickening and hypoechoic areas raises questions about the opportunity to revise the definition of these two major components for the US diagnosis of enthesitis.

25 March 2022 in Rheumatology
Verified
Journal Article
Closed access
PDF file
AMAndrea Di Matteo
Andrea Di Matteo
GSGianluca Smerilli
Gianluca Smerilli
ECedoardo cipolletta
edoardo cipolletta
RWRichard J Wakefield
Richard J Wakefield
RARossella De Angelis
Rossella De Angelis
WGWalter Grassi
Walter Grassi
Objectives: The objectives of this study were (1) to explore US findings for muscle mass, muscle quality and muscle stiffness in SLE patients and healthy subjects; (2) to investigate the relationship between the US muscle findings and physical performance in SLE patients and healthy subjects.Methods: Quadriceps muscle thickness was used for assessment of muscle mass, muscle echogenicity (using a visual semi-quantitative scale and grayscale analysis with histograms) for assessment of muscle quality, and point shear-wave elastography (SWE) for assessment of muscle stiffness in 30 SLE patients (without previous/current myositis or neuromuscular disorders) and 15 age-, sex- and BMI-matched healthy subjects. Hand grip strength tests and short physical performance battery (SPPB) tests were carried out in the same populations.Results: No difference was observed between SLE patients and healthy subjects for quadriceps muscle thickness (35.2 mm ±s.d. 6.8 vs 34.8 mm ± s.d. 6.0, respectively, P = 0.79). Conversely, muscle echogenicity was significantly increased in SLE patients (visual semi-quantitative scale: 1.7 ± s.d. 1.0 vs 0.3 ± s.d. 0.5, respectively, P < 0.01; grayscale analysis with histograms: 87.4 mean pixels ± s.d. 18.8 vs 70.1 mean pixels ± s.d. 14.0, respectively, P < 0.01). Similarly, SWE was significantly lower in SLE patients compared with healthy subjects {1.5 m/s [interquartile range (IQR) 0.3] vs 1.6 m/s (IQR 0.2), respectively, P = 0.01}. Muscle echogenicity was inversely correlated with grip strength (visual semi-quantitative scale, Rho: –0.47, P = 0.01; grayscale analysis with histograms, Rho: –0.41, p < 0.01) and SPPB (visual semi-quantitative scale, Rho: –0.50, P < 0.01; grayscale analysis with histograms Rho: –0,46, P < 0.01).Conclusions: US assessment of muscle echogenicity and stiffness is useful for the early detection of muscle involvement in SLE patients.