Lung Ultrasound Data Research Articles

Deep learning approaches for automated classification of neonatal lung ultrasound with assessment of human-to-AI interrater agreement

Neonatal respiratory disorders pose significant challenges in clinical settings, often requiring rapid and accurate diagnostic solutions for effective management. Lung ultrasound (LUS) has emerged as a promising tool to evaluate respiratory conditions in neonates. This evaluation is mainly based on the interpretation of visual patterns (horizontal artifacts, vertical artifacts, and consolidations). Automated interpretation of these patterns can assist clinicians in their evaluations. However, developing AI-based solutions for this purpose is challenging, primarily due to the lack of annotated data and inherent subjectivity in expert interpretations. This study aims to propose an automated solution for the reliable interpretation of patterns in LUS videos of newborns. We employed two distinct strategies. The first strategy is a frame-to-video-level approach that computes frame-level predictions from deep learning (DL) models trained from scratch (F2V-TS) along with fine-tuning pre-trained models (F2V-FT) followed by aggregation of those predictions for video-level evaluation. The second strategy is a direct video classification approach (DV) for evaluating LUS data. To evaluate our methods, we used LUS data from 34 neonatal patients comprising of 70 exams with annotations provided by three expert human operators (3HOs). Results show that within the frame-to-video-level approach, F2V-FT achieved the best performance with an accuracy of 77% showing moderate agreement with the 3HOs. while the direct video classification approach resulted in an accuracy of 72%, showing substantial agreement with the 3HOs, our proposed study lays down the foundation for reliable AI-based solutions for newborn LUS data evaluation.

FLUEnT: Transformer for detecting lung consolidations in videos using fused lung ultrasound encodings

Pneumonia is the leading cause of death among children around the world. According to WHO, a total of 740,180 lives under the age of five were lost due to pneumonia in 2019. Lung ultrasound (LUS) has been shown to be particularly useful for supporting the diagnosis of pneumonia in children and reducing mortality in resource-limited settings. The wide application of point-of-care ultrasound at the bedside is limited mainly due to a lack of training for data acquisition and interpretation. Artificial Intelligence can serve as a potential tool to automate and improve the LUS data interpretation process, which mainly involves analysis of hyper-echoic horizontal and vertical artifacts, and hypo-echoic small to large consolidations. This paper presents, Fused Lung Ultrasound Encoding-based Transformer (FLUEnT), a novel pediatric LUS video scoring framework for detecting lung consolidations using fused LUS encodings. Frame-level embeddings from a variational autoencoder, features from a spatially attentive ResNet-18, and encoded patient information as metadata combiningly form the fused encodings. These encodings are then passed on to the transformer for binary classification of the presence or absence of consolidations in the video. The video-level analysis using fused encodings resulted in a mean balanced accuracy of 89.3 %, giving an average improvement of 4.7 % points in comparison to when using these encodings individually. In conclusion, outperforming the state-of-the-art models by an average margin of 8 % points, our proposed FLUEnT framework serves as a benchmark for detecting lung consolidations in LUS videos from pediatric pneumonia patients.

A machine learning-based lung ultrasound algorithm for the diagnosis of acute heart failure.

Lung ultrasound (LUS) is an effective tool for diagnosing acute heart failure (AHF). However, several imaging protocols currently exist and how to best use LUS remains undefined. We aimed at developing a lung ultrasound-based model for AHF diagnosis using machine learning. Random forest and decision trees were generated using the LUS data (via an 8-zone scanning protocol) in patients with acute dyspnea admitted to the Emergency Department (PLUME study, N = 117) and subsequently validated in an external dataset (80 controls from the REMI study, 50 cases from the Nancy AHF cohort). Using the random forest model, total B-line sum (i.e., in both hemithoraces) was the most significant variable for identifying AHF, followed by the difference in B-line sum between the superior and inferior lung areas. The decision tree algorithm had a good diagnostic accuracy [area under the curve (AUC) = 0.865] and identified three risk groups (i.e., low 24%, high 70%, and very high-risk 96%) for AHF. The very high-risk group was defined by the presence of 14 or more B-lines in both hemithoraces while the high-risk group was described as having either B-lines mostly localized in superior points or in the right hemithorax. Accuracy in the validation cohort was excellent (AUC = 0.906). Importantly, adding the algorithm on top of a validated clinical score and classical definition of positive LUS scanning for AHF resulted in a significant improvement in diagnostic accuracy (continuous net reclassification improvement = 1.21, P < 0.001). Our simple lung ultrasound-based machine learning algorithm features an excellent performance and may constitute a validated strategy to diagnose AHF.

Improving the Generalizability and Performance of an Ultrasound Deep Learning Model Using Limited Multicenter Data for Lung Sliding Artifact Identification.

Deep learning (DL) models for medical image classification frequently struggle to generalize to data from outside institutions. Additional clinical data are also rarely collected to comprehensively assess and understand model performance amongst subgroups. Following the development of a single-center model to identify the lung sliding artifact on lung ultrasound (LUS), we pursued a validation strategy using external LUS data. As annotated LUS data are relatively scarce-compared to other medical imaging data-we adopted a novel technique to optimize the use of limited external data to improve model generalizability. Externally acquired LUS data from three tertiary care centers, totaling 641 clips from 238 patients, were used to assess the baseline generalizability of our lung sliding model. We then employed our novel Threshold-Aware Accumulative Fine-Tuning (TAAFT) method to fine-tune the baseline model and determine the minimum amount of data required to achieve predefined performance goals. A subgroup analysis was also performed and Grad-CAM++ explanations were examined. The final model was fine-tuned on one-third of the external dataset to achieve 0.917 sensitivity, 0.817 specificity, and 0.920 area under the receiver operator characteristic curve (AUC) on the external validation dataset, exceeding our predefined performance goals. Subgroup analyses identified LUS characteristics that most greatly challenged the model's performance. Grad-CAM++ saliency maps highlighted clinically relevant regions on M-mode images. We report a multicenter study that exploits limited available external data to improve the generalizability and performance of our lung sliding model while identifying poorly performing subgroups to inform future iterative improvements. This approach may contribute to efficiencies for DL researchers working with smaller quantities of external validation data.

Low-complexity lung ultrasound video scoring by means of intensity projection-based video compression

Since the outbreak of COVID-19, efforts have been made towards semi-quantitative analysis of lung ultrasound (LUS) data to assess the patient’s condition. Several methods have been proposed in this regard, with a focus on frame-level analysis, which was then used to assess the condition at the video and prognostic levels. However, no extensive work has been done to analyze lung conditions directly at the video level. This study proposes a novel method for video-level scoring based on compression of LUS video data into a single image and automatic classification to assess patient’s condition. The method utilizes maximum, mean, and minimum intensity projection-based compression of LUS video data over time. This enables to preserve hyper- and hypo-echoic data regions, while compressing the video down to a maximum of three images. The resulting images are then classified using a convolutional neural network (CNN). Finally, the worst predicted score given among the images is assigned to the corresponding video. The results show that this compression technique can achieve a promising agreement at the prognostic level (81.62%), while the video-level agreement remains comparable with the state-of-the-art (46.19%). Conclusively, the suggested method lays down the foundation for LUS video compression, shifting from frame-level to direct video-level analysis of LUS data.

Predicting heart failure readmission and all-cause mortality in patients with acute heart failure: a simple predischarge risk score

Abstract Introduction Predischarge risk stratification of patients hospitalized for acute heart failure (AHF) could facilitate tailored treatment and follow-up, however, simple scores to predict short-term risk for HF readmission or death are lacking. Purpose We sought to develop a congestion-focused risk score to predict HF readmission and all-cause mortality after discharge in patients hospitalized for AHF. Methods We used data from a prospective, two-center observational study in adults hospitalized for AHF. Laboratory data were collected on admission. Patients underwent a physical examination, 4-zone, and in a subset 8-zone, lung ultrasound (LUS), and conventional echocardiography at baseline (within a median of 1 day after admission). A second LUS was performed before discharge in a subset of patients. The primary endpoint was the composite of HF rehospitalization or all-cause death. HF hospitalizations were adjudicated blinded to LUS findings. Results Among 297 patients with complete data (median age 74 years, 43% women, mean left ventricular ejection fraction 39%), 76 patients (26%) were hospitalized for HF or died within 90 days after discharge. A stepwise Cox regression model selected four significant independent predictors of the composite outcome and each was assigned points proportional to its regression coefficient: NT-proBNP ≥2000 pg/mL (admission) (3 points), systolic blood pressure &amp;lt;120 mmHg (baseline) (2 points), left atrial volume index ≥60 mL/m² (baseline) (1 point) and ≥9 B-lines on predischarge 4-zone LUS (3 points). The weighted risk score provided adequate risk discrimination for the composite outcome (HR 1.48 per 1 point increase, 95% confidence interval: 1.32-1.67, p&amp;lt;0.001, C-statistic: 0.70; Figure 1). In a subset of patients with 8-zone LUS data (n=176), results were similar (C-statistic: 0.72). Conclusion A four-variable, weighted risk score integrating clinical, laboratory and ultrasound imaging data provided adequate risk discrimination for 90-day adverse outcomes in patients hospitalized for AHF. If validated in a larger dataset, this score could be used to identify patients needing the closest monitoring, more intensive follow-up and early treatment adjustment.Figure 1.

Coronavirus disease 2019 patients prognostic stratification based on low complex lung ultrasound video compression

In the last years, efforts have been made towards automating semi-quantitative analysis of lung ultrasound (LUS) data. To this end, several methods have been proposed with a focus on frame-level classification. However, no extensive work has been done to evaluate LUS data directly at the video level. This study proposes an effective video compression and classification technique for assessing LUS data. This technique is based on maximum, mean, and minimum intensity projection (with respect to the temporal dimension) of LUS video data. This compression allows preserving hyper- and hypo-echoic regions and results in compressing a LUS video down to three frames, which are then classified using a convolutional neural network (CNN). Results show that this compression not only preserves visual artifacts appearance in the reduced data, but also achieves a promising agreement of 81.61% at the prognostic level. Conclusively, the suggested method reduces the amount of frames needed to assess LUS video down to 3. Note that on average a LUS videos consists of a few hundreds frames. At the same time, state-of-the-art performance at video and prognostic levels are achieved, while significantly reducing the computational cost.

Effectiveness of transferring ultrasound deep learning models from adults to pediatrics for frame based pneumonia classification

Hand-held ultrasound devices are emerging as a promising intervention to aid in diagnosing deadly early childhood pneumonia in the developing world. Lung ultrasound (LUS) data, however, can be difficult to read and interpret accurately, and thus require trained professionals. A variety of deep learning (DL) models have been developed to aid professionals in this task, but the difficulty curating quality training datasets limits the generalization capabilities of these models. To combat this data scarcity, we utilized a variety of DL models trained on LUS data collected from adult COVID-19 patients in Italy and were able to apply transfer learning to increase pneumonia related imaging pattern classification in LUS data from children and infants in rural Zambia. In this regard, we found that transfer learning significantly increased frame classification F1 performance as compared to training models from scratch. These findings indicate the promising potential for developing generalizable AI for LUS classification in diverse contexts given limited data. This analysis demonstrates that DL models effectively transfer their capabilities when fine tuned on a new population demographics, allowing the possibility to clinically deploy these models without the need to acquire large amounts of new data.

Trajectory and correlates of pulmonary congestion by lung ultrasound in patients with acute myocardial infarction: insights from PARADISE-MI.

PARADISE-MI examined the efficacy of sacubitril/valsartan in acute myocardial infarction (AMI) complicated by reduced left ventricular ejection fraction (LVEF), pulmonary congestion, or both. We sought to assess the trajectory of pulmonary congestion using lung ultrasound (LUS) and its association with cardiac structure and function in a pre-specified substudy. Patients without prior heart failure (HF) underwent eight-zone LUS and echocardiography at baseline (±2 days of randomization) and after 8 months. B-lines were quantified offline, blinded to treatment, clinical findings, time point, and outcomes. Among 152 patients (median age 65, 32% women, mean LVEF 41%), B-lines were detectable in 87% at baseline [median B-line count: 4 (interquartile range 2-8)]. Among 115 patients with LUS data at baseline and follow-up, B-lines decreased significantly from baseline (mean ± standard deviation: -1.6 ± 7.3; P = 0.018). The proportion of patients without pulmonary congestion at follow-up was significantly higher in those with fewer B-lines at baseline. Adjusted for baseline, B-lines at follow-up were on average 6 (95% confidence interval: 3-9) higher in patients who experienced an intercurrent HF event vs. those who did not (P = 0.001). A greater number of B-lines at baseline was associated with larger left atrial size, higher E/e' and E/A ratios, greater degree of mitral regurgitation, worse right ventricular systolic function, and higher tricuspid regurgitation velocity (P-trend <0.05 for all). In this AMI cohort, B-lines, indicating pulmonary congestion, were common at baseline and, on average, decreased significantly from baseline to follow-up. Worse pulmonary congestion was associated with prognostically important echocardiographic markers.

Relationship between lung ultrasound and electrical impedance tomography as regional assessment tools during PEEP titration in acute respiratory distress syndrome caused by multi-lobar pneumonia: a pilot study.

Acute respiratory distress syndrome (ARDS) caused by multilobar pneumonia (MLP) is markedly different from typical ARDS in pathology, imaging characteristics, and lung mechanics. Regional lung assessment is required. We aimed to analyze the relationship between two regional assessment tools, lung ultrasound (LUS) and electrical impedance tomography (EIT) during positive end-expiratory pressure (PEEP) titration, and determine an appropriate PEEP level. We conducted a prospective study of patients with ARDS caused by MLP with PaO2/FiO2 < 150mmHg. All subjects were equipped with two EIT belts connected with a single EIT machine to measure upper and lower hemithorax impedance change alternatingly at each PEEP level. LUS score was simultaneously determined in chest wall regions corresponding to the EIT regions during PEEP titration. We acquired EIT and LUS data in eight regions of interest at seven PEEP levels in 12 subjects. Therefore, 672 pairs of data were obtained for analysis. There were significant relationships between LUS score and tidal impedance variation and pixel compliance (Cpix). The Spearman's rho between LUS score vs. tidal impedance variation and LUS score vs. the Cpix were -0.142, P < 0.001, and -0.195, P < 0.001, respectively. The relationship between the LUS score and Cpix remained the same at every PEEP level but did not reach statistical significance. The individual's mean expected PEEP by LUS was similar to the EIT [10.33(± 1.67) vs. 10.33(± 1.44) cm H2O, P = 0.15]. Regarding the MLP, the LUS scores were associated with EIT parameters, and LUS scores might proof helpful for finding individual PEEP settings in MLP.

Trajectory of pulmonary congestion by lung ultrasound in patients with acute myocardial infarction and association with cardiac structure and function

Abstract Background The PARADISE-MI trial examined the efficacy of sacubitril/valsartan in patients with acute myocardial infarction (AMI) complicated by reduced left ventricular ejection fraction (LVEF), pulmonary congestion or both. Little is known about the trajectory and echocardiographic correlates of pulmonary congestion in this population. Purpose We sought to assess the trajectory of pulmonary congestion using lung ultrasound (LUS) and its association with cardiac structure and function in a subset of patients enrolled in PARADISE- MI. Methods Participants underwent 8-zone LUS at baseline and 8 months. B-lines were quantified offline, blinded to treatment group, clinical findings, timepoint and outcomes by a core laboratory. Paired t-tests, chi-squared tests, and linear regression analyses were conducted. Results Among 152 patients (median age 65 years, 32% women, 35% obese, mean LVEF 41%), any B-lines were detectable in 87%, the median sum of B-lines in 8 zones was 4 [IQR 2–8], and 67% had ≥3 B-lines indicative of congestion. Greater number of B-lines at baseline was associated with larger left atrial (LA) size, higher E/e' and E/A ratios, greater degree of mitral regurgitation, worse right ventricular (RV) systolic function, and higher tricuspid regurgitation velocity (P trend &amp;lt;0.05 for all) (Figure 1). Among 115 patients with 8-month LUS data, there was a significant decline in number of B-lines from baseline (mean ± SD: −1.6±7.3; p=0.018). Adjusted for baseline, B-lines at follow-up were on average 6 (95% CI: 3, 9) higher in a patient who experienced an intercurrent heart failure (HF) event than a non-HF patient (p=0.001). Among 75 patients with ≥3 B-lines at baseline, a decrease in B-lines to &amp;lt;3, indicating decongestion, occurred in 37% and was similar in the sacubitril/valsartan and ramipril groups (36% vs. 39%, p=0.83). Conclusions In this post-AMI cohort, sonographic B-lines, indicating pulmonary congestion, were common at baseline and were significantly higher at follow-up in those who developed HF. Worse pulmonary congestion at baseline was associated with prognostically important echocardiographic markers of LV filling pressure, pulmonary pressure, and RV function. Funding Acknowledgement Type of funding sources: Private company. Main funding source(s): Novartis

State of the Art in Lung Ultrasound, Shifting from Qualitative to Quantitative Analyses.

Lung ultrasound (LUS) has been increasingly expanding since the 1990s, when the clinical relevance of vertical artifacts was first reported. However, the massive spread of LUS is only recent and is associated with the coronavirus disease 2019 (COVID-19) pandemic, during which semi-quantitative computer-aided techniques were proposed to automatically classify LUS data. In this review, we discuss the state of the art in LUS, from semi-quantitative image analysis approaches to quantitative techniques involving the analysis of radiofrequency data. We also discuss recent in vitro and in silico studies, as well as research on LUS safety. Finally, conclusions are drawn highlighting the potential future of LUS.

Are Lung Ultrasound Features More Severe in Children Diagnosed with Bronchiolitis after the COVID-19 Lockdown Period?

The non-pharmacological measures implemented during the SARS-CoV-2 pandemic disrupted the usual bronchiolitis seasonality. Some authors have speculated that, after the lock down period, there would be an increase in the number and severity of respiratory infections due to the re-introduction of respiratory viruses. We collected clinical, microbiological and lung ultrasound data using the classification of the Italian Society of Thoracic Ultrasound (ADET) in children with bronchiolitis during the pandemic compared to the pre-pandemic period, with the aim of assessing whether the epidemic of bronchiolitis during the pandemic was characterized by a more severe lung involvement documented by lung ultrasound. We enrolled 108 children with bronchiolitis (52 pre-pandemic and 56 COVID-19 period), with a median age of 1.74 months (interquartile range, IQR 1–3.68) and 39.8% were females. Rhinovirus detection and high-flow nasal cannula usage were both increased during the COVID-19 period, although overall need of hospitalization and pediatric intensive care unit admissions did not change during the two periods. Lung ultrasound scores were similar in the two cohorts evaluated. Conclusions: our study suggests that, despite changes in microbiology and treatments performed, lung ultrasound severity scores were similar, suggesting that that bronchiolitis during the pandemic period was no more severe than pre-pandemic period, despite children diagnosed during the pandemic had a higher, but it was not statistically significant, probably, due to small sample size, probability of being admitted.

Transfer Learning for Automated COVID-19 B-Line Classification in Lung Ultrasound.

Lung ultrasound (LUS) as a diagnostic tool is gaining support for its role in the diagnosis and management of COVID-19 and a number of other lung pathologies. B-lines are a predominant feature in COVID-19, however LUS requires a skilled clinician to interpret findings. To facilitate the interpretation, our main objective was to develop automated methods to classify B-lines as pathologic vs. normal. We developed transfer learning models based on ResNet networks to classify B-lines as pathologic (at least 3 B-lines per lung field) vs. normal using COVID-19 LUS data. Assessment of B-line severity on a 0-4 multi-class scale was also explored. For binary B-line classification, at the frame-level, all ResNet models pretrained with ImageNet yielded higher performance than the baseline nonpretrained ResNet-18. Pretrained ResNet-18 has the best Equal Error Rate (EER) of 9.1% vs the baseline of 11.9%. At the clip-level, all pretrained network models resulted in better Cohen's kappa agreement (linear-weighted) and clip score accuracy, with the pretrained ResNet-18 having the best Cohen's kappa of 0.815 [95% CI: 0.804-0.826], and ResNet-101 the best clip scoring accuracy of 93.6%. Similar results were shown for multi-class scoring, where pretrained network models outperformed the baseline model. A class activation map is also presented to guide clinicians in interpreting LUS findings. Future work aims to further improve the multi-class assessment for severity of B-lines with a more diverse LUS dataset.

Lung Ultrasound to Assist ICU Admission Decision-Making Process of COVID-19 Patients With Acute Respiratory Failure.

OBJECTIVES:There is only low-certainty evidence on the use of predictive models to assist COVID-19 patient’s ICU admission decision-making process. Accumulative evidence suggests that lung ultrasound (LUS) assessment of COVID-19 patients allows accurate bedside evaluation of lung integrity, with the added advantage of repeatability, absence of radiation exposure, reduced risk of virus dissemination, and low cost. Our goal is to assess the performance of a quantified indicator resulting from LUS data compared with standard clinical practice model to predict critical respiratory illness in the 24 hours following hospital admission.DESIGN:Prospective cohort study.SETTING:Critical Care Unit from University Hospital Purpan (Toulouse, France) between July 2020 and March 2021.PATIENTS:Adult patients for COVID-19 who were in acute respiratory failure (ARF), defined as blood oxygen saturation as measured by pulse oximetry less than 90% while breathing room air or respiratory rate greater than or equal to 30 breaths/min at hospital admission. Linear multivariate models were used to identify factors associated with critical respiratory illness, defined as death or mild/severe acute respiratory distress syndrome (Pao2/Fio2 < 200) in the 24 hours after patient’s hospital admission.INTERVENTION:LUS assessment.MEASUREMENTS AND MAIN RESULTS:One hundred and forty COVID-19 patients with ARF were studied. This cohort was split into two independent groups: learning sample (first 70 patients) and validation sample (last 70 patients). Interstitial lung water, thickening of the pleural line, and alveolar consolidation detection were strongly associated with patient’s outcome. The LUS model predicted more accurately patient’s outcomes than the standard clinical practice model (DeLong test: Testing: z score = 2.50, p value = 0.01; Validation: z score = 2.11, p value = 0.03).CONCLUSIONS:LUS assessment of COVID-19 patients with ARF at hospital admission allows a more accurate prediction of the risk of critical respiratory illness than standard clinical practice. These results hold the promise of improving ICU resource allocation process, particularly in the case of massive influx of patients or limited resources, both now and in future anticipated pandemics.

Deep Learning-Based Classification of Reduced Lung Ultrasound Data From COVID-19 Patients.

The application of lung ultrasound (LUS) imaging for the diagnosis of lung diseases has recently captured significant interest within the research community. With the ongoing COVID-19 pandemic, many efforts have been made to evaluate LUS data. A four-level scoring system has been introduced to semiquantitatively assess the state of the lung, classifying the patients. Various deep learning (DL) algorithms supported with clinical validations have been proposed to automate the stratification process. However, no work has been done to evaluate the impact on the automated decision by varying pixel resolution and bit depth, leading to the reduction in size of overall data. This article evaluates the performance of DL algorithm over LUS data with varying pixel and gray-level resolution. The algorithm is evaluated over a dataset of 448 LUS videos captured from 34 examinations of 20 patients. All videos are resampled by a factor of 2, 3, and 4 of original resolution, and quantized to 128, 64, and 32 levels, followed by score prediction. The results indicate that the automated scoring shows negligible variation in accuracy when it comes to the quantization of intensity levels only. Combined effect of intensity quantization with spatial down-sampling resulted in a prognostic agreement ranging from 73.5% to 82.3%.These results also suggest that such level of prognostic agreement can be achieved over evaluation of data reduced to 32 times of its original size. Thus, laying foundation to efficient processing of data in resource constrained environments.

A standardised and evidenced-based lung ultrasound acquisition protocol and scoring system for the monitoring and stratification of COVID-19 and post-COVID-19 patients

Lung ultrasound (LUS) is currently utilized worldwide to assess COVID-19 patients. However, imaging protocols are often defined arbitrarily, and studies on post-COVID-19 are lacking. In this work, we report on the capabilities of standardized LUS to monitor and stratify COVID-19 and post-COVID-19 patients. A validated and standardized imaging protocol based on 14 scanning-areas and a 4-level scoring system were utilized to collect and analyze data from 220 patients, 100 COVID-19 positive, and 120 post-COVID-19. Next, the capability of five imaging protocols (based on 4, 8, 10, 12, and 14 scanning-areas) to intercept the most significant LUS findings was compared. Moreover, a longitudinal-study was conducted aiming at investigating the possibility to simplify the protocol during follow-up. Results on the agreement between AI-models and LUS experts with respect to LUS data evaluation are also reported. In conclusion, a 12-areas protocol emerges as the optimal trade-off between a time-efficient and an accurate LUS examination. However, it appears not to be possible to reduce further the number of scanning-areas during follow-up. Finally, COVID-19 and post-COVID-19 data seem to show differences capable to confuse AI models that were not trained on post-COVID-19 data, supporting the hypothesis of the existence of LUS patterns specific to post-COVID-19.

Lung Ultrasound in COVID-19 and Post-COVID-19 Patients, an Evidence-Based Approach.

ObjectivesWorldwide, lung ultrasound (LUS) was utilized to assess coronavirus disease 2019 (COVID‐19) patients. Often, imaging protocols were however defined arbitrarily and not following an evidence‐based approach. Moreover, extensive studies on LUS in post‐COVID‐19 patients are currently lacking. This study analyses the impact of different LUS imaging protocols on the evaluation of COVID‐19 and post‐COVID‐19 LUS data.MethodsLUS data from 220 patients were collected, 100 COVID‐19 positive and 120 post‐COVID‐19. A validated and standardized imaging protocol based on 14 scanning areas and a 4‐level scoring system was implemented. We utilized this dataset to compare the capability of 5 imaging protocols, respectively based on 4, 8, 10, 12, and 14 scanning areas, to intercept the most important LUS findings. This to evaluate the optimal trade‐off between a time‐efficient imaging protocol and an accurate LUS examination. We also performed a longitudinal study, aimed at investigating how to eventually simplify the protocol during follow‐up. Additionally, we present results on the agreement between AI models and LUS experts with respect to LUS data evaluation.ResultsA 12‐areas protocol emerges as the optimal trade‐off, for both COVID‐19 and post‐COVID‐19 patients. For what concerns follow‐up studies, it appears not to be possible to reduce the number of scanning areas. Finally, COVID‐19 and post‐COVID‐19 LUS data seem to show differences capable to confuse AI models that were not trained on post‐COVID‐19 data, supporting the hypothesis of the existence of LUS patterns specific to post‐COVID‐19 patients.ConclusionsA 12‐areas acquisition protocol is recommended for both COVID‐19 and post‐COVID‐19 patients, also during follow‐up.

Lung Ultrasound Findings Associated With COVID-19 ARDS, ICU Admission, and All-Cause Mortality.

As lung ultrasound (LUS) has emerged as a diagnostic tool in patients with COVID-19, we sought to investigate the association between LUS findings and the composite in-hospital outcome of ARDS incidence, ICU admission, and all-cause mortality. In this prospective, multi-center, observational study, adults with laboratory-confirmed SARS-CoV-2 infection were enrolled from non-ICU in-patient units. Subjects underwent an LUS evaluating a total of 8 zones. Images were analyzed off-line, blinded to clinical variables and outcomes. A LUS score was developed to integrate LUS findings: ≥ 3 B-lines corresponded to a score of 1, confluent B-lines to a score of 2, and subpleural or lobar consolidation to a score of 3. The total LUS score ranged from 0-24 per subject. Among 215 enrolled subjects, 168 with LUS data and no current signs of ARDS or ICU admission (mean age 59 y, 56% male) were included. One hundred thirty-six (81%) subjects had pathologic LUS findings in ≥ 1 zone (≥ 3 B-lines, confluent B-lines, or consolidations). Markers of disease severity at baseline were higher in subjects with the composite outcome (n = 31, 18%), including higher median C-reactive protein (90 mg/L vs 55, P < .001) and procalcitonin levels (0.35 μg/L vs 0.13, P = .033) and higher supplemental oxygen requirements (median 4 L/min vs 2, P = .001). However, LUS findings and score did not differ significantly between subjects with the composite outcome and those without, and were not associated with outcomes in unadjusted and adjusted logistic regression analyses. Pathologic findings on LUS were common a median of 3 d after admission in this cohort of non-ICU hospitalized subjects with COVID-19 and did not differ among subjects who experienced the composite outcome of incident ARDS, ICU admission, and all-cause mortality compared to subjects who did not. These findings should be confirmed in future investigations. The study is registered at Clinicaltrials.gov (NCT04377035).

Multi-feature Multi-Scale CNN-Derived COVID-19 Classification from Lung Ultrasound Data.

The global pandemic of the novel coronavirus disease 2019 (COVID-19) has put tremendous pressure on the medical system. Imaging plays a complementary role in the management of patients with COVID-19. Computed tomography (CT) and chest X-ray (CXR) are the two dominant screening tools. However, difficulty in eliminating the risk of disease transmission, radiation exposure and not being cost-effective are some of the challenges for CT and CXR imaging. This fact induces the implementation of lung ultrasound (LUS) for evaluating COVID-19 due to its practical advantages of noninvasiveness, repeatability, and sensitive bedside property. In this paper, we utilize a deep learning model to perform the classification of COVID-19 from LUS data, which could produce objective diagnostic information for clinicians. Specifically, all LUS images are processed to obtain their corresponding local phase filtered images and radial symmetry transformed images before fed into the multi-scale residual convolutional neural network (CNN). Secondly, image combination as the input of the network is used to explore rich and reliable features. Feature fusion strategy at different levels is adopted to investigate the relationship between the depth of feature aggregation and the classification accuracy. Our proposed method is evaluated on the point-of-care US (POCUS) dataset together with the Italian COVID-19 Lung US database (ICLUS-DB) and shows promising performance for COVID-19 prediction.