Deep learning for real-time multi-class segmentation of artefacts in lung ultrasound.

Lung ultrasound (LUS) has emerged as a safe and cost-effective modality for assessing lung health, particularly during the COVID-19 pandemic. However, interpreting LUS images remains challenging due to its reliance on artefacts, leading to operator variability and limiting its practical uptake. To address this, we propose a deep learning pipeline for multi-class segmentation of objects (ribs, pleural line) and artefacts (A-lines, B-lines, B-line confluence) in ultrasound images of a lung training phantom. Lightweight models achieved a mean Dice Similarity Coefficient (DSC) of 0.74, requiring fewer than 500 training images. Applying this method in real-time, at up to 33.4 frames per second in inference, allows enhanced visualisation of these features in LUS images. This could be useful in providing LUS training and helping to address the skill gap. Moreover, the segmentation masks obtained from this model enable the development of explainable measures of disease severity, which have the potential to assist in the triage and management of patients. We suggest one such semi-quantitative measure called the B-line Artefact Score, which is related to the percentage of an intercostal space occupied by B-lines and in turn may be associated with the severity of a number of lung conditions. Moreover, we show how transfer learning could be used to train models for small datasets of clinical LUS images, identifying pathologies such as simple pleural effusions and lung consolidation with DSC values of 0.48 and 0.32 respectively. Finally, we demonstrate how such DL models could be translated into clinical practice, implementing the phantom model alongside a portable point-of-care ultrasound system, facilitating bedside assessment and improving the accessibility of LUS.

Diagnostic performance of sonographic activity scores for adult terminal ileal Crohn's disease compared to magnetic resonance and histological reference standards: experience from the METRIC trial.

OBJECTIVES: The simple ultrasound activity score for Crohn's disease (SUS-CD) and bowel ultrasound score (BUSS) are promising intestinal ultrasound (IUS) indices of CD, but studied mainly in small settings with few sonographers. We compared SUS-CD and BUSS against histological and magnetic resonance enterography (MRE) reference standards in a post hoc analysis of a prospective multicentre, multireader trial. METHODS: Participants recruited to the METRIC trial (ISRCTN03982913) were studied, including those with available terminal ileal (TI) biopsies. Sensitivity and specificity of SUS-CD and BUSS for TI CD activity were calculated with 95% confidence intervals (CI), from the prospective observations of the original METRIC trial sonographers against the histological activity index (HAI) and the simplified magnetic resonance index of activity (sMARIA). RESULTS: We included 284 patients (median 31.5 years, IQR 23-46) from 8 centres, who underwent IUS and MRE. Of these, 111 patients had available terminal ileal biopsies with HAI scoring. Against histology, sensitivity and specificity for active disease were 79% (95% CI 69-86%) and 50% (31-69%) for SUS-CD, and 66% (56-75%) and 68% (47-84%) for BUSS, respectively. Compared to sMARIA, the sensitivity and specificity for active CD were 81% (74-86%) and 75% (66-83%) for SUS-CD, and 68% (61-74%) and 85% (76-91%) for BUSS, respectively. The sensitivity of SUS-CD was significantly greater than that of BUSS against HAI and sMARIA (p < 0.001), but its specificity was significantly lower than of BUSS against the MRE reference standard (p = 0.003). CONCLUSIONS: Particularly when compared to MRE activity scoring, SUS-CD and BUSS are promising tools in a real-world clinical setting. CLINICAL RELEVANCE STATEMENT: When tested using data from a multicentre, multireader diagnostic accuracy trial, the simple ultrasound activity score for Crohn's disease (SUS-CD) and bowel ultrasound score (BUSS) were clinically viable intestinal ultrasound indices that were reasonably sensitive and specific for terminal ileal Crohn's disease, especially when compared to a magnetic resonance reference standard. KEY POINTS: The simple ultrasound activity score for Crohn's disease and bowel ultrasound score are promising intestinal ultrasound indices of Crohn's disease but to date studied mainly in small settings with few sonographers. Compared to histology and the magnetic resonance reference standard in a multicentre, multireader setting, the sensitivity of simple ultrasound activity score for Crohn's disease is significantly greater than that of bowel ultrasound score. The specificity of simple ultrasound activity score for Crohn's disease was significantly lower than that of bowel ultrasound score compared to the magnetic resonance enterography reference standard. The specificity of both indices was numerically higher when the magnetic resonance enterography reference standard was adopted.

Observer agreement for small bowel ultrasound in Crohn's disease: results from the METRIC trial.

PURPOSE: To prospectively evaluate interobserver agreement for small bowel ultrasound (SBUS) in newly diagnosed and relapsing Crohn's disease. METHODS: A subset of patients recruited to a prospective trial comparing the diagnostic accuracy of MR enterography and SBUS underwent a second SBUS performed by one of a pool of six practitioners, who recorded the presence, activity and location of small bowel and colonic disease. Detailed segmental mural and extra-mural observations were also scored. Interobserver variability was expressed as percentage agreement with a construct reference standard, split by patient cohort, grouping disease as present or absent. Prevalence adjusted bias adjusted kappa (PABAK), and simple percentage agreement between practitioners, irrespective of the reference standard, were calculated. RESULTS: Thirty-eight patients (11 new diagnosis, 27 relapse) were recruited from two sites. Overall percentage agreement for small bowel disease presence against the consensus reference was 82% (52-95% (95%CI)), kappa coefficient (κ) 0.64, (substantial agreement) for new diagnosis and 81%, κ 0.63 (substantial agreement) for the relapsing cohort. Agreement for colonic disease presence was 64%, κ 0.27 (fair agreement) in new diagnosis and 78%,κ 0.56 (moderate agreement) in the relapsing cohort. Simple agreement between practitioners was 84% and 87% for small bowel and colonic disease presence respectively. Practitioners agreed on small bowel disease activity in 24/27 (89%) where both identified disease. Kappa agreement for detailed mural observations ranged from κ 0.00 to 1.00. CONCLUSION: There is substantial practitioner agreement for small bowel disease presence in newly diagnosed and relapsing CD patients, supporting wider dissemination of enteric US.