Lung ultrasound for etiological diagnosis of pneumonia in the emergency department: correlation with bronchoalveolar lavage results

This study aims to develop a deep learning algorithm, Pneumonia-Plus, based on computed tomography (CT) images for accurate classification of bacterial, fungal, and viral pneumonia. A total of 2763 participants with chest CT images and definitepathogen diagnosis were included to train and validate an algorithm. Pneumonia-Plus was prospectively tested on a nonoverlapping dataset of 173 patients. The algorithm's performance in classifying three types of pneumonia was compared to that of three radiologists using the McNemar test to verify its clinical usefulness. Among the 173 patients, area under the curve (AUC) values for viral, fungal, and bacterial pneumonia were 0.816, 0.715, and 0.934, respectively. Viral pneumonia was accurately classified with sensitivity, specificity, and accuracy of 0.847, 0.919, and 0.873. Three radiologists also showed good consistency with Pneumonia-Plus. The AUC values of bacterial, fungal, and viral pneumonia were 0.480, 0.541, and 0.580 (radiologist 1: 3-year experience); 0.637, 0.693, and 0.730 (radiologist 2: 7-year experience); and 0.734, 0.757, and 0.847 (radiologist 3: 12-year experience), respectively. The McNemar test results for sensitivity showed that the diagnostic performance of the algorithm was significantly better than that of radiologist 1 and radiologist 2 (p < 0.05) in differentiating bacterial and viral pneumonia. Radiologist 3 had a higher diagnostic accuracy than the algorithm. The Pneumonia-Plus algorithm is used to differentiate between bacterial, fungal, and viral pneumonia, which has reached the level of an attending radiologist and reduce the risk of misdiagnosis. The Pneumonia-Plus is important for appropriate treatment and avoiding the use of unnecessary antibiotics, and provide timely information to guide clinical decision-making and improve patient outcomes. Pneumonia-Plus algorithm could assist in the accurate classification of pneumonia based on CT images, which has great clinical value in avoiding the use of unnecessary antibiotics, and providing timely information to guide clinical decision-making and improve patient outcomes. • The Pneumonia-Plus algorithm trained from data collected from multiple centers can accurately identify bacterial, fungal, and viral pneumonia. • The Pneumonia-Plus algorithm was found to have better sensitivity in classifying viral and bacterial pneumonia in comparison to radiologist 1 (5-year experience) and radiologist 2 (7-year experience). • The Pneumonia-Plus algorithm is used to differentiate between bacterial, fungal, and viral pneumonia, which has reached the level of an attending radiologist.

The diagnosis of pediatric pneumonia and determination of the likely pathogen are complicated as the clinical picture overlaps with other respiratory illnesses, interpretation of radiographs is subjective, and laboratory results are rarely diagnostic. This study was designed to describe the relative rates of bacterial and viral pneumonia in the pediatric Emergency Department (ED), determine the accuracy of pediatric ED physicians' ability to diagnose pneumonia and distinguish bacterial from viral etiology, and to determine clinical and laboratory predictors of bacterial pneumonia. Children 3 months to 16 years of age presenting to seven Canadian pediatric EDs before the COVID-19 pandemic with fever and cough who had a chest radiograph performed for possible pneumonia were enrolled and underwent standardized clinical investigations. An expert panel was convened and reached a Consensus Diagnosis of typical or atypical bacterial pneumonia, viral pneumonia or not pneumonia for each case. The expert panel assessed 247 cases with the Consensus Diagnosis being typical bacterial pneumonia (N = 44(18%)), atypical bacterial pneumonia (N = 18(7%)), viral pneumonia (N = 46(19%)) and no pneumonia (N = 139(56%)). Treating ED physician diagnoses were typical bacterial pneumonia (N = 126(51%)), atypical bacterial pneumonia (N = 3(1%)), viral pneumonia (N = 10(4%)) and no pneumonia (N = 108(44%)) with low agreement between a diagnosis of bacterial pneumonia by the ED physician and the panel's Consensus Diagnosis (Kappa 0.15 (95% CI 0.08, 0.21)). Cut off values that predicted bacterial pneumonia as the Consensus Diagnosis were ESR ≥ 47 mm/hour, CRP ≥ 42 mg/L and procalcitonin ≥0.85 ng/m. Age greater than 5 years and cough for 5 or more days also predict bacterial pneumonia. In this cohort, pediatric ED physicians over-diagnosed typical bacterial pneumonia and underdiagnosed viral and atypical bacterial pneumonia. Bacterial pneumonia is most likely in children over 5 years of age, with cough for 5 or more days and/or with elevated inflammatory markers.

We initiated a prospective study with a group of practitioners to assess the etiology, clinical presentation, and outcome of community-acquired pneumonia in patients diagnosed in the outpatient setting. All patients with signs and symptoms suggestive of pneumonia and an infiltrate on chest X-ray underwent an extensive standard workup and were followed over 4 weeks. Over a 4-year period, 184 patients were eligible, of whom 170 (age range, 15-96 yr; median, 43 yr) were included and analyzed. In 78 (46%), no etiologic agent could be demonstrated. In the remaining 92 patients, 107 etiologic agents were implicated: 43 were due to "pyogenic" bacteria (39 Streptococcus pneumoniae, 3 Haemophilus spp., 1 Streptococcus spp.), 39 were due to "atypical" bacteria (24 Mycoplasma pneumoniae, 9 Chlamydia pneumoniae, 4 Coxiella burnetii, 2 Legionella spp.), and 25 were due to viruses (20 influenza viruses and 5 other respiratory viruses). There were only a few statistically significant clinical differences between the different etiologic categories (higher age and comorbidities in viral or in episodes of undetermined etiology, higher neutrophil counts in "pyogenic" episodes, more frequent bilateral and interstitial infiltrates in viral episodes). There were 2 deaths, both in patients with advanced age (83 and 86 years old), and several comorbidities. Only 14 patients (8.2%) required hospitalization. In 6 patients (3.4%), the pneumonia episode uncovered a local neoplasia. This study shows that most cases of community-acquired pneumonia have a favorable outcome and can be successfully managed in an outpatient setting. Moreover, in the absence of rapid and reliable clinical or laboratory tests to establish a definite etiologic diagnosis at presentation, the spectrum of the etiologic agents suggest that initial antibiotic therapy should cover both S. pneumoniae and atypical bacteria, as well as possible influenza viruses during the epidemic season.

BackgroundEmergency department visits quadrupled with the initial onset and surge during the 2009 H1N1 influenza pandemic in New York City from April to June 2009. This time period was unique in that >90% of the circulating virus was surveyed to be the novel 2009 H1N1 influenza A according to the New York City Department of Health. We describe our experience using lung ultrasound in a case series of patients with respiratory symptoms requiring chest X-ray during the initial onset and surge of the 2009 H1N1 influenza pandemic.MethodsWe describe a case series of patients from a prospective observational cohort study of lung ultrasound, enrolling patients requiring chest X-ray for suspected pneumonia that coincided with the onset and surge of the 2009 H1N1 influenza pandemic.ResultsTwenty pandemic 2009 H1N1 influenza patients requiring chest X-ray were enrolled during this time period. Median age was 6.7 years. Lung ultrasound via modified Bedside Lung Ultrasound in Emergency protocol assisted in the identification of viral pneumonia (n = 15; 75%), viral pneumonia with superimposed bacterial pneumonia (n = 7; 35%), isolated bacterial pneumonia only (n = 1; 5%), and no findings of viral or bacterial pneumonia (n = 4; 20%) in this cohort of patients. Based on 54 observations, interobserver agreement for distinguishing viral from bacterial pneumonia using lung ultrasound was ĸ = 0.82 (0.63 to 0.99).ConclusionsLung ultrasound may be used to distinguish viral from bacterial pneumonia. Lung ultrasound may be useful during epidemics or pandemics of acute respiratory illnesses for rapid point-of-care triage and management of patients.

This study is aimed at exploring the ability to use heparin-binding protein (HBP) in bronchoalveolar lavage fluid (BALF) to differentially diagnose bacterial infection from viral infection for severe community-acquired pneumonia (CAP) in critically ill children. A total of 181 children with severe CAP admitted to the intensive care unit (ICU) were included in this study. BALF and blood samples were collected within the first 24 hours of admission. BALF HBP and interleukin-6 (IL-6) concentrations and neutrophil percentage (N%) as well as blood HBP, IL-6, procalcitonin (PCT), C-reactive protein, white blood cell concentrations and N% were measured. Of the enrolled children, 126 were confirmed to have bacterial pneumonia, and 55 were confirmed to have viral pneumonia. Blood HBP and PCT concentrations and N% and BALF HBP and IL-6 concentrations and N% were significantly higher in bacterial pneumonia than in viral pneumonia (P < 0.05). In the bacterial pneumonia group, HBP and IL-6 concentrations and N% in BALF samples were all significantly higher than those in blood samples (P < 0.001), and BALF HBP and IL-6 concentrations and N% were correlated with blood HBP and IL-6 concentrations and N%, respectively (r = 0.439, 0.250, and 0.235, P < 0.01). BALF N% and blood N% were both correlated with BALF HBP concentrations and blood HBP concentrations, respectively (r = 0.622 and 0.346, P < 0.001). ROC analysis revealed that BALF HBP showed the best ability to predict bacterial pneumonia, with an area under the curve of 0.994, a sensitivity of 95.24%, and a specificity of 100.00% at its optimal cutoff value of 74.05 ng/mL. BALF HBP might be a promising biomarker for the early discrimination of bacterial infection from viral infection in critically ill children with severe CAP.

BackgroundThe usefulness of bronchoalveolar lavage (BAL) fluid cellular analysis in pneumonia has not been adequately evaluated. This study investigated the ability of cellular analysis of BAL fluid to differentially diagnose bacterial pneumonia from viral pneumonia in adult patients who are admitted to intensive care unit.MethodsBAL fluid cellular analysis was evaluated in 47 adult patients who underwent bronchoscopic BAL following less than 24 hours of antimicrobial agent exposure. The abilities of BAL fluid total white blood cell (WBC) counts and differential cell counts to differentiate between bacterial and viral pneumonia were evaluated using receiver operating characteristic (ROC) curve analysis.ResultsBacterial pneumonia (n = 24) and viral pneumonia (n = 23) were frequently associated with neutrophilic pleocytosis in BAL fluid. BAL fluid median total WBC count (2,815/µL vs. 300/µL, P<0.001) and percentage of neutrophils (80.5% vs. 54.0%, P = 0.02) were significantly higher in the bacterial pneumonia group than in the viral pneumonia group. In ROC curve analysis, BAL fluid total WBC count showed the best discrimination, with an area under the curve of 0.855 (95% CI, 0.750–0.960). BAL fluid total WBC count ≥510/µL had a sensitivity of 83.3%, specificity of 78.3%, positive likelihood ratio (PLR) of 3.83, and negative likelihood ratio (NLR) of 0.21. When analyzed in combination with serum procalcitonin or C-reactive protein, sensitivity was 95.8%, specificity was 95.7%, PLR was 8.63, and NLR was 0.07. BAL fluid total WBC count ≥510/µL was an independent predictor of bacterial pneumonia with an adjusted odds ratio of 13.5 in multiple logistic regression analysis.ConclusionsCellular analysis of BAL fluid can aid early differential diagnosis of bacterial pneumonia from viral pneumonia in critically ill patients.

Lung ultrasound (LUS) has emerged as a rapid, immediately available, non-invasive bedside tool for detection of several pulmonary and/or pleural diseases [1,2]. However, no data are available for its impact on patient management in the emergency department. We studied 50 adult patients (25 women, median age 80.5 years, interquartile range 12.3 years) presenting with acute undifferentiated dyspnea to the emergency department of ‘Edoardo Agnelli’ Hospital (Pinerolo, Turin), and evaluated the LUS diagnostic impact by comparing the main diagnosis (cardiac, respiratory, or combined dyspnea), the most likely pathophysiologic dysfunction, and the etiological diagnosis, as indicated by the emergency physician caring for the patient, before and after LUS. Reference diagnosis was established by two expert emergency physicians, blinded to LUS results, who independently reviewed the entire medical record. Moreover, we asked the emergency physician how LUS findings changed patient management. We used a previously described eight-region scanning protocol [3] and assessed five major syndromes (see Figure 1) [1]. Diffuse interstitial syndrome (IS) was the most common finding (58%), followed by pleural effusion (52%), focal IS (18%), alveolar consolidation (14%), and pneumothorax (8%). Twenty-seven patients had two or more LUS pathological findings (for example, 17 patients showed pleural effusion and IS), while in six patients LUS detected no alterations. Figure 1 Lung ultrasound major syndromes. (A) Diffuse B-lines representing interstitial syndrome. (B) Alveolar consolidation. (C) M-mode findings of a pneumothorax (stratosphere sign or bar code sign; left), and a normal lung (seashore sign). (D) Pleural effusion. ... The agreement between clinical assessment and reference diagnoses was fair (Cohen’s kappa coefficient = 0.25, 0.32, and 0.26 for main, pathophysiologic, and etiological diagnosis, respectively; P <0.01), while agreement was excellent between LUS-implemented and reference diagnoses (kappa coefficient = 0.94, 0.84, and 0.81, respectively; P <0.01). Overall, LUS changed the main clinical diagnosis in 44% of cases. The agreement for the different main diagnosis is reported in Figure 2. Figure 2 Agreement for the main diagnosis between clinical assessment or lung ultrasound (LUS)-implemented assessment, and reference diagnosis. Therapeutic management was changed, because of LUS findings, in 58% of patients. LUS led to prescribing a new drug in 19 cases, and to holding a drug previously considered for treatment in 10 cases; moreover, LUS results led to a new procedure (for example, thoracentesis) being performed in six cases, and to a change of the disposition plan in five cases. Our study is the first to address the issue of how LUS affects patient management in real-world practice. We found a high diagnostic and therapeutic impact, mainly related to the sequential approach used integrating clinical assessment and LUS, that paired the detection of IS, a very sensitive but poorly specific ultrasound sign [1], with the pretest probability of clinical assessment. These data need to be confirmed in larger cohort and multicenter studies. Abbreviations IS: Interstitial syndrome; LUS: Lung ultrasound. Competing interests The authors declare that they have no competing interests.

Lung ultrasound (LUS) has emerged as a rapid, immediately available, non-invasive bedside tool for detection of several pulmonary and/or pleural diseases [1,2]. However, no data are available for its impact on patient management in the emergency department. We studied 50 adult patients (25 women, median age 80.5 years, interquartile range 12.3 years) presenting with acute undifferentiated dyspnea to the emergency department of ‘Edoardo Agnelli’ Hospital (Pinerolo, Turin), and evaluated the LUS diagnostic impact by comparing the main diagnosis (cardiac, respiratory, or combined dyspnea), the most likely pathophysiologic dysfunction, and the etiological diagnosis, as indicated by the emergency physician caring for the patient, before and after LUS. Reference diagnosis was established by two expert emergency physicians, blinded to LUS results, who independently reviewed the entire medical record. Moreover, we asked the emergency physician how LUS findings changed patient management. We used a previously described eight-region scanning protocol [3] and assessed five major syndromes (see Figure 1) [1]. Diffuse interstitial syndrome (IS) was the most common finding (58%), followed by pleural effusion (52%), focal IS (18%), alveolar consolidation (14%), and pneumothorax (8%). Twenty-seven patients had two or more LUS pathological findings (for example, 17 patients showed pleural effusion and IS), while in six patients LUS detected no alterations. Figure 1 Lung ultrasound major syndromes. (A) Diffuse B-lines representing interstitial syndrome. (B) Alveolar consolidation. (C) M-mode findings of a pneumothorax (stratosphere sign or bar code sign; left), and a normal lung (seashore sign). (D) Pleural effusion. ... The agreement between clinical assessment and reference diagnoses was fair (Cohen’s kappa coefficient = 0.25, 0.32, and 0.26 for main, pathophysiologic, and etiological diagnosis, respectively; P <0.01), while agreement was excellent between LUS-implemented and reference diagnoses (kappa coefficient = 0.94, 0.84, and 0.81, respectively; P <0.01). Overall, LUS changed the main clinical diagnosis in 44% of cases. The agreement for the different main diagnosis is reported in Figure 2. Figure 2 Agreement for the main diagnosis between clinical assessment or lung ultrasound (LUS)-implemented assessment, and reference diagnosis. Therapeutic management was changed, because of LUS findings, in 58% of patients. LUS led to prescribing a new drug in 19 cases, and to holding a drug previously considered for treatment in 10 cases; moreover, LUS results led to a new procedure (for example, thoracentesis) being performed in six cases, and to a change of the disposition plan in five cases. Our study is the first to address the issue of how LUS affects patient management in real-world practice. We found a high diagnostic and therapeutic impact, mainly related to the sequential approach used integrating clinical assessment and LUS, that paired the detection of IS, a very sensitive but poorly specific ultrasound sign [1], with the pretest probability of clinical assessment. These data need to be confirmed in larger cohort and multicenter studies. Abbreviations IS: Interstitial syndrome; LUS: Lung ultrasound. Competing interests The authors declare that they have no competing interests.

Background: We hypothesized that children with viral pneumonia have higher severity of illness and higher mechanical ventilation (MV) requirement than children with bacterial pneumonia. We aim to compare respiratory support requirements of severe viral and bacterial pneumonia cases admitted to the pediatric intensive care unit (PICU). Methods: This is a retrospective review of patients with microbiologically proven severe viral and bacterial pneumonia admitted to PICU in KK Women’s and Children’s Hospital, Singapore, from 2010 to 2014. Demographic, clinical and ventilatory data up to 14 days of PICU admission were extracted and analyzed. Results: Forty-nine and 68 patients had sole viral and bacterial pathogens, respectively. Patients with viral pneumonia were more likely to be vs . 27.9%, P=0.011), to have underlying comorbidities (59.2% vs . 35.3%, P=0.010) and had higher pediatric index of mortality 2 (PIM2) score [3.0 (1.1, 8.0) vs . 1.6 (0.8, 3.0), P vs . 24.5%, P=0.008) and required longer duration of MV [7.0 (4.0, 10.0) vs . 4.0 (1.0, 10.8) days, P=0.031]. Oxygenation index (OI) of children with viral pneumonia was higher on day 1 [OI: 11.7 (6.6, 19.3) vs . 5.7 (3.7, 10.8), P=0.006] and 3 [OI: 8.0 (6.0, 20.0) vs . 5.0 (3.0, 8.0), P vs . 0%, P Conclusions: Children with viral pneumonia were more likely to require alternative modes of MV and longer duration of MV.

While it is too late to confine the COVID-19 coronovirus outbreak to China, a wealth of data spurs epidemiological and vaccine research.

Introduction and Objective: Endocan has been used as a biomarker in the differential diagnosis of pulmonary diseases in adults. However, there are only a limited number of studies on its use in children. In this context, the objective of this study is to evaluate the relationship between serum endocan levels in children with bacterial and viral pneumonia. Materials and Methods: The population of this prospective case-control study consisted of hospitalized children aged 1 month to 15 years diagnosed with pneumonia between August 2020 and July 2021, whereas the control group consisted of randomly selected healthy children. The demographic and clinical characteristics of all participants were recorded. Participants' endocan levels, white blood cell (WBC) and neutrophil counts, and C-reactive protein (CRP) and procalcitonin (PCT) levels were measured within the scope of the laboratory tests. Results: The study sample consisted of 41 children, of whom 21 had bacterial pneumonia and 20 had viral pneumonia, whereas the control group consisted of 47 healthy children. Serum endocan levels, WBC and neutrophil counts, and PCT and CRP levels were significantly higher in children with bacterial pneumonia than in children with viral pneumonia and healthy children (P < 0.05). Additionally, serum endocan levels were significantly higher in children with viral pneumonia than in healthy children (P < 0.001). The endocan levels in children with bacterial pneumonia were significantly associated with the need for intensive care (P = 0.004) and correlated with the length of hospital stay (LoS) (r = 0.592, P = 0.005). Conclusion: The findings of this study indicated that serum endocan levels can be used in the differential diagnosis of bacterial and viral pneumonias. Additionally, it was found that the need for intensive care and LoS were significantly correlated with endocan levels in children with bacterial pneumonia.

To determine the efficacy of extracorporeal membrane oxygenation (ECMO) for nonneonatal acute respiratory failure. Single-institution, retrospective medical record review from February 1990 to March 2008. Tertiary care hospital. Eighty-one nonneonatal patients (mean age, 23 years; age range, 2 months to 61 years) with acute respiratory failure who had failed maximal ventilator support received ECMO therapy between 1990 and 2008. Patients were grouped into 6 categories based on diagnosis: sepsis (n = 8), bacterial or fungal pneumonia (n = 15), viral pneumonia (n = 9), trauma or burn (n = 10), immunocompromise (n = 15), and other (n = 24). Main Outcome Measure Survival to hospital discharge. Overall survival was 53%. Survival was highest in patients with viral pneumonia (78%), followed by bacterial pneumonia (53%), sepsis syndrome (44%), and immunocompromise (40%). Patients treated following trauma or burns had the lowest survival (33%). The average age was 19 years for survivors as compared with 27 years for nonsurvivors. Survival was lower in patients with multiple organ failure as compared with those with single organ failure (33% vs 60%, respectively), in patients who experienced mechanical ventilation for longer than 10 days prior to the initiation of ECMO as compared with those who received ventilatory support for less than 10 days prior to the initiation of ECMO (31% vs 57%, respectively), and in patients requiring more than 400 hours of ECMO support as compared with those requiring less than 400 hours of ECMO support (42% vs 55%, respectively). Therapy with ECMO may provide a survival benefit in carefully selected patients with nonneonatal acute respiratory failure who have failed maximal ventilator support. Nonneonatal survival with ECMO therapy is strongly dependent on diagnosis, with the highest survival seen in those with viral or bacterial pneumonia. Older age, multiple organ failure, prolonged ventilation prior to ECMO initiation, and long ECMO runs are associated with decreased survival.

Background: Viral pneumonia is a common complication of influenza like illness including COVID-19. Virus is the most common cause of pneumonia in children. While there are important cues in history, the laboratory examination that can help differentiate viral and bacterial causes. There are limited clinical manifestations to differentiate viral and co-infection bacterial and viral pneumonia. Objective: To determine basic clinical manifestations and laboratory performed at bedside as predictors to differentiate viral and co-infection bacterial and viral pneumonia. Materials and Methods: A retrospective study was conducted in pediatric patients with radiographic evidence of severe pneumonia and admitted in Nakornping Hospital, Chiang Mai, Thailand between October 2017 and April 2020. Multiplex real time polymerase chain reactions (PCRs) (RP-24-26) were used to identify the cause of the pneumonia. Demographic data and basic clinical predictors such as age, comorbidity, symptoms, and physical findings, and basic laboratory such as complete blood count were collected. Patients were divided into three groups, no infection detected, viral infection, and bacterial and viral co-infection. Polytomous logistic regression was performed to investigate predictors for type of infection. Area under the receiver operating characteristic curve (AuROC) and 95% confidence interval (CI) were further determined for a final model to differentiate type of infection. Results: Two hundred eight cases participated in this study and included 122 males (58.7%). The etiology of pneumonia identified by RP-24-26, pathogen was detected among 166 cases (79.8%). A virus was detected in 141 cases (67.8%), co-infection was detected in 25 cases (12%), and no infection was detected in 42 cases (20.2%). The statistically significant predictor for viral pneumonia was cough, with an adjusted relative risk ratio (RRR) of 4.42 (1.41 to 13.82). The statistically significant predictors for co-infection were age 13 to 24 months and lymphocytes at 40.0% or greater with adj. RRR 14.28 (1.47 to 138.52), and 4.60 (1.24 to 17.04), respectively. Pneumonia patients with cough were 4.42 times more to have viral cause. Those with lymphocytes of 40.0% or greater were 4.60 times more to be co-infected with both virus and bacteria, especially in the age group 13 to 24 months compared with 1 to 12 months. The age group13 to 24 months, cough, and lymphocytes at 40.0% or greater were 74% (95% CI 0.66 to 0.82) correctly predicted to viral pneumonia. Conclusion: Coughing is a predominant symptom for childhood pneumonia caused by virus. In addition, pneumonia patients age 13 to 24 months who have lymphocyte of more than 40% should start and continue antibiotics until complete course. Keywords: Childhood pneumonia, Respiratory panel 24, Respiratory panel 26, Co-infection bacterial pneumonia, Predictor

With 92,000 deaths and 18 percent of the total child mortality every year, pneumonia is the leading cause of child mortality in children under 5 in Pakistan. Pakistan is one of the top 5 countries for childhood pneumonia deaths around the world. Bacteria and viruses are most common infectious agents of pneumonia. The diagnostic test for pneumonia detection is chest x-ray. Basic diagnostic tests facilities are available even at rural health centers. In proposed study, a pre-trained convolutional neural network; VGG19 model is fine-tuned on dataset of 5863 chest x-ray images of healthy, viral, and bacterial pneumonia. The VGG19, model 1 is trained on viral and bacterial pneumonia images, and model 2 is trained on multi-class data. The model 1 with viral and bacterial pneumonia images showed training accuracy of 0.83 and validation accuracy of 0.84. The model 2 with normal, viral, and bacterial pneumonia, showed training accuracy of 0.84 and validation accuracy of 0.85. The results show that the VGG19 model has powerful prediction capacity to identify correct features of types of pneumonia with reasonable accuracy even with smaller and unbalanced dataset. The results predict that already developed and trained algorithms can be used as ready to use clinical diagnostic tool, if fine-tuned with larger balanced dataset with few targeted changes. These tools can be used as second reader tool by the physicians, can process thousands of images in limited time with high accuracy, relieving the burden of patients on limited capacity of the healthcare facilities.

Lung ultrasound (LUS) is a crucial diagnostic tool for identifying pneumonia in the pediatric age group. However, it plays a limited role in the early distinction between viral and bacterial pneumonia in children. The objectives of our study were to determine if LUS and the neutrophil-lymphocyte ratio (NLR) were useful in identifying and distinguishing between viral and bacterial pneumonia in Egyptian children under the age of two. Within the first 12 h of being admitted to our department, 52 children with clinical symptoms and signs suggestive of community-acquired pneumonia (CAP) underwent LUS and the NLR. LUS and the NLR strongly differentiated children with viral from those with bacterial pneumonia. For the early diagnosis and differentiation between viral and bacterial pneumonia in young Egyptian children, LUS was proven to be a noninvasive and reliable method. Combining the NLR with LUS increased the diagnostic accuracy when evaluating children suspected of having pneumonia.