Abnormal Chest X-ray Research Articles

The human role to guarantee an ethical AI in healthcare: a five-facts approach

AbstractWith the emergence of AI systems to assist clinical decision-making, several ethical dilemmas are brought to the general attention. AI systems are claimed to be the solution for many high-skilled medical tasks where machines can potentially surpass human ability as for example in identifying normal and abnormal chest X-rays. However, there are also warns that AI tools could be the basis for a human replacement that can risk dehumanisation in medicine. In recent years, important proposals in the domain of AI ethics in healthcare have identified main ethical issues, as for example fairness, autonomy, transparency, and responsibility. The human warranty, which implies human evaluation of the AI procedures, has been described to lower the ethical risks. However, as relevant these works have been, translating principles into action has proved challenging as existing codes were mostly a description of principles. There is a great need to produce how-to proposals that are specific enough to be action-guiding. We present five human-focussed facts designed into a framework of human action for an ethical AI in healthcare. Through the factors, we examine the role of medical practitioners, patients, and developers in designing, implementing, and using AI in a responsible manner that preserves human dignity. The facts encompass a range of ethical concerns that were commonly found in relevant literature. Given that it is crucial to bring as many perspectives as possible to the field, this work contributes to translate principles into human action to guarantee an ethical AI in health.

Effects of electric welding on hearing loss and respiratory damage

Objective: To investigate the occupational health status of electric welding workers and explore the effects of electric welding on hearing loss and respiratory damage. Methods: From August to December 2021, the cluster sampling method was used to select workers from an automobile manufacturer in Guangzhou City as research subjects: 636 welding workers in the welding workshop as the welding group, 757 assembly workers in the engine workshop and the final assembly workshop exposed to pure noise as the assembly group. Occupational disease hazard factors were detected for welding positions and assembly positions, and occupational health examination was carried out for research subjects. The occupational health status, the trends of hearing loss and respiratory abnormalities with working age were compared and analyzed between the two groups. Binary logistic regression was used to analyze the association between hearing loss and respiratory abnormalities in welding workers. Results: The excess rates of welding fumes, manganese and its compounds in the welding position were both 9.68% (3/31). Its noise exposure intensity [ (85.36±2.68) dB (A) ] and excess rate [48.39% (15/31) ] were not significantly different from those in the assembly position [ (84.86±3.28) dB (A) and 43.24% (16/37) ] (P>0.05). The results of the occupational health examination showed that the detection rates of hearing loss, digital radiography (DR) chest X-ray abnormality, alanine aminotransferase abnormality, deazelaic aminotransferase abnormality and white blood cell count abnormality of workers in the welding group were higher than those in the assembly group (P<0.05). The detection rates of hearing loss, DR chest X-ray abnormality, pulmonary ventilation abnormality in the welding group and the detection rate of hearing loss in the assembly group increased with the working age of the workers (P<0.05). The hearing loss detection rate and DR chest X-ray abnormality detection rate of the workers with ≥9 years working age in the welding group were both higher than those in the assembly group workers with same working age (P<0.05). The binary logistic regression analysis showed that abnormal pulmonary ventilation and abnormal DR chest X-ray were the risk factors for hearing loss in welding workers (OR=10.83, 95%CI: 7.31-16.06; OR=16.59, 95%CI: 5.72-48.10; P<0.05) . Conclusion: Hearing loss and respiratory damage are prominent problems among welding workers, and the detection rates of abnormality increase with the working age of the workers. Hearing loss in welding workers is associated with abnormal pulmonary ventilation and abnormal DR chest X-ray.

A retrospective review of the utility of chest X-rays in diving and submarine medical examinations.

Performance of routine Chest X-rays (CXRs) in asymptomatic individuals to assess hyperbaric exposure risk is controversial. The radiation risk may overshadow the low yield in many settings. However, the yield may be higher in certain settings, such as tuberculosis-endemic countries. We evaluated the utility of routine CXR in diving and submarine medical examinations in South Africa. Records of 2,777 CXRs during 3,568 fitness examinations of 894 divers and submariners spanning 31 years were reviewed to determine the incidence of CXR abnormality. Associated factors were evaluated using odds ratios and a binomial logistic regression model, with a Kaplan-Meier plot to describe the duration of service until first abnormal CXR. An abnormal CXR was reported in 1.1% per person year of service, yielding a cumulative incidence of 6.5% (58/894) of the study participants. Only four individuals had a clinical indication for the CXR in their medical history. A range of potential pathologies were seen, of which 15.5% were declared disqualifying and the rest (84.5%) were treated, or further investigation showed that the person could be declared fit. In South Africa, a routine CXR has a role to play in detecting abnormalities that are incompatible with pressure exposures. The highest number of abnormalities were found during the initial examinations and in individuals with long service records. Only four individuals had a clinical indication for their CXR during the 31-year span of our study. Similar studies should be performed to make recommendations in other countries and settings.

Multi-Label Classification of Chest X-ray Abnormalities Using Transfer Learning Techniques.

In recent years, deep neural networks have enabled countless innovations in the field of image classification. Encouraged by success in this field, researchers worldwide have demonstrated how to use Convolutional Neural Network techniques in medical imaging problems. In this article, the results were obtained through the use of the EfficientNet in the task of classifying 14 different diseases based on chest X-ray images coming from the NIH (National Institutes of Health) ChestX-ray14 dataset. The approach addresses dataset imbalances by introducing a custom split to ensure fair representation. Binary cross entropy loss is utilized to handle the multi-label difficulty. The model architecture comprises an EfficientNet backbone for feature extraction, succeeded by sequential layers including GlobalAveragePooling, Dense, and BatchNormalization. The main contribution of this paper is a proposed solution that outperforms previous state-of-the-art deep learning models average area under the receiver operating characteristic curve-AUC-ROC (score: 84.28%). The usage of the transfer-learning technique and traditional deep learning engineering techniques was shown to enable us to obtain such results on consumer-class GPUs (graphics processing units).

Using Computer Vision Techniques to Automatically Detect Abnormalities in Chest X-rays.

Our research focused on creating an advanced machine-learning algorithm that accurately detects anomalies in chest X-ray images to provide healthcare professionals with a reliable tool for diagnosing various lung conditions. To achieve this, we analysed a vast collection of X-ray images and utilised sophisticated visual analysis techniques; such as deep learning (DL) algorithms, object recognition, and categorisation models. To create our model, we used a large training dataset of chest X-rays, which provided valuable information for visualising and categorising abnormalities. We also utilised various data augmentation methods; such as scaling, rotation, and imitation; to increase the diversity of images used for training. We adopted the widely used You Only Look Once (YOLO) v8 algorithm, an object recognition paradigm that has demonstrated positive outcomes in computer vision applications, and modified it to classify X-ray images into distinct categories; such as respiratory infections, tuberculosis (TB), and lung nodules. It was particularly effective in identifying unique and crucial outcomes that may, otherwise, be difficult to detect using traditional diagnostic methods. Our findings demonstrate that healthcare practitioners can reliably use machine learning (ML) algorithms to diagnose various lung disorders with greater accuracy and efficiency.

How Many Echocardiograms Should be Performed to Detect One Cardiac Abnormality in Patients &lt;18 Years? Results from the “Performing Echocardiography to Detect Heart Disease (PED-HEART) Study”

Introduction: Echocardiography is an important cardiac investigating tool, however, its appropriate use in patients &lt;18 years has to be ascertained. Aim: The aim of this study was to assess the clinical indication of the referral for echocardiography in patients with abnormal and normal echocardiographic examination. Methods: Single-center, retrospective observational analysis was done in patients &lt;18 years who underwent echocardiography. This included patients’ clinical profiles and echocardiography examination findings. Results: Out of 7500 patients, normal echocardiography examinations (Group 1) were found in 75.33% (n = 5650) of the patients. Out of these (Group 1), 32.8% were referred for electrocardiographic changes, while 25% and 31% were for palpitation and chest pain, respectively. Patients with trivial findings on echocardiography were 350 (Group 2). Patients with a significant abnormality on echocardiography requiring medical management (Group 3) or cardiac intervention (Group 4) were 1250 and 250, respectively. The comparison of Group A (combined Groups 1 and 2 patients) with Group B (combined Groups 3 and 4 patients) showed that the referrals made for murmur, suspected heart failure, chest X-ray abnormalities, abnormal blood pressure/shock, and multiple parameters were found to be an independent predictor of abnormal findings on echocardiography. Conclusion: Abnormal findings on echocardiography requiring cardiac intervention in patients &lt;18 years is one in every five echocardiograms performed in this study. The implementation of appropriate use criteria for echocardiography may help in improving this ratio.

Multi-Label Local to Global Learning: A Novel Learning Paradigm for Chest X-Ray Abnormality Classification.

Deep neural network (DNN) approaches have shown remarkable progress in automatic Chest X-rays classification. However, existing methods use a training scheme that simultaneously trains all abnormalities without considering their learning priority. Inspired by the clinical practice of radiologists progressively recognizing more abnormalities and the observation that existing curriculum learning (CL) methods based on image difficulty may not be suitable for disease diagnosis, we propose a novel CL paradigm, named multi-label local to global (ML-LGL). This approach iteratively trains DNN models on gradually increasing abnormalities within the dataset, i,e, from fewer abnormalities (local) to more ones (global). At each iteration, we first build the local category by adding high-priority abnormalities for training, and the abnormality's priority is determined by our three proposed clinical knowledge-leveraged selection functions. Then, images containing abnormalities in the local category are gathered to form a new training set. The model is lastly trained on this set using a dynamic loss. Additionally, we demonstrate the superiority of ML-LGL from the perspective of the model's initial stability during training. Experimental results on three open-source datasets, PLCO, ChestX-ray14 and CheXpert show that our proposed learning paradigm outperforms baselines and achieves comparable results to state-of-the-art methods. The improved performance promises potential applications in multi-label Chest X-ray classification.

Possible Factors Related to Mortality of Pregnant Women with COVID-19: A Machine Learning Approach

Background: The COVID-19 pandemic has affected people of all ages and backgrounds, but pregnant women are a particularly vulnerable group. They are at increased risk of severe illness and death from COVID-19, yet there is limited and conflicting data about the factors that contribute to mortality in this population. In this study, we sought to use machine learning (ML) to predict the possible factors associated with mortality in pregnant women with COVID-19. Methods: We collected data from a large cohort of pregnant women who had been diagnosed with COVID-19 at Hung Vuong Hospital. We included a range of demographic, clinical, and laboratory variables in our analysis, such as age, gestational age, comorbidities, symptoms, and laboratory test results. Our goal was to identify the critical factors that could be used to predict mortality in this vulnerable population. We applied several ML models and analyzed the results to determine the most useful predictors. Results: We analyzed a cohort of 550 pregnant women diagnosed with COVID-19, comprising 525 survivors and 25 deaths. Using ML algorithms, we identified key clinical and patient factors that correlated with mortality risk, such as infection severity, pulse rate, breathing patterns, SpO2 levels, rapid diagnostic tests, internal pathology, breathing difficulties, and changes in consciousness. Additionally, several medical test factors were found to be significant, including white blood cells, neutrophils, platelets, C-reactive protein (CRP), procalcitonin, and chest X-ray abnormalities. Our analysis revealed that the Gaussian Naive Bayes (NB) model was the most accurate, with an AUC score of 0.79, demonstrating its potential as a valuable tool for identifying high-risk patients. Conclusions: Our results suggest that the available data can capture the relationship between COVID-19 symptoms and mortality in pregnant women. More studies are needed to improve the accuracy and generalizability of models for predicting mortality risk in this population.

Low CRB-65 Scores Effectively Rule out Adverse Clinical Outcomes in COVID-19 Irrespective of Chest Radiographic Abnormalities.

Background: CRB-65 (Confusion; Respiratory rate ≥ 30/min; Blood pressure ≤ 90/60 mmHg; age ≥ 65 years) is a risk score for prognosticating patients with COVID-19 pneumonia. However, a significant proportion of COVID-19 patients have normal chest X-rays (CXRs). The influence of CXR abnormalities on the prognostic value of CRB-65 is unknown, limiting its wider applicability. Methods: We assessed the influence of CXR abnormalities on the prognostic value of CRB-65 in COVID-19. Results: In 589 study patients (71 years (IQR: 57-83); 57% males), 186 (32%) had normal CXRs. On ROC analysis, CRB-65 performed similarly in patients with normal vs. abnormal CXRs for predicting inpatient mortality (AUC 0.67 ± 0.05 vs. 0.69 ± 0.03). In patients with normal CXRs, a CRB-65 of 0 ruled out mortality, NIV requirement and critical illness (intubation and/or ICU admission) with negative predictive values (NPVs) of 94%, 98% and 99%, respectively. In patients with abnormal CXRs, a CRB-65 of 0 ruled out the same endpoints with NPVs of 91%, 83% and 86%, respectively. Patients with low CRB-65 scores had better inpatient survival than patients with high CRB-65 scores, irrespective of CXR abnormalities (all p < 0.05). Conclusions: CRB-65, CXR and CRP are independent predictors of mortality in COVID-19. Adding CXR findings (dichotomised to either normal or abnormal) to CRB-65 does not improve its prognostic accuracy. A low CRB-65 score of 0 may be a good rule-out test for adverse clinical outcomes in COVID-19 patients with normal or abnormal CXRs, which deserves prospective validation.

Respiratory problems in full term newborns, which parameters are related to the length of in-patient stay?

To analyze respiratory problems of full term newborn babies during their first hours of life, compare parameters related to the length of neonatal intensive care unit (NICU) in-patient stay, and discuss perinatal care practices, which could prevent consequences, and/or reduce associated costs of NICU hospitalization. Retrospective chart review between July 31st 2017 and March 2nd 2018 in a tertiary maternity hospital. Chi-square testing analyzed qualitative data. Mann-Whitney test was used for quantitative variables. Multiple linear regression models determined odds ratios for associations were found to be significant in univariate analyses. 86 babies with respiratory problems were studied. Delivery mode was vaginal in 11.6% and caesarean section in 88.4% of newborns. About 52.3% of admitted newborns had transient tachypnea and improved relatively quickly. Newborns resuscitated in (P = 0.004) or admitted to the NICU from the delivery room (P = 0.000) displayed greater need for O2 administration. Twenty neonates were intubated; 16 of them were administered surfactant. Infection (P &lt; 0.05), abnormal chest x-ray (P = 0.022), and the severity of respiratory system morbidity (P &lt; 0.05) prolonged neonatal in-patient stay. Respiratory problems (P = 0.003) and intubation (P = 0.032) incurred greater hospitalization costs. Breastfeeding initiation before the 3rd day of life (P = 0.031), and O2 administration for more than 72 h (P = 0.036) were significantly associated with the length of in-patient stay in the multivariate regression analysis. Mandating the presence of a pediatrician in the delivery room would optimize justified direct NICU admissions and reduce the possibility of delayed diagnosis of respiratory distress during rooming-in. NICU in-patient stay in full term neonates should be based on combining diagnostic markers of sepsis, such as procalcitonin or interleukin-6. A specific protocol regarding surfactant administration in intubated full term infants is also necessary. The severity of respiratory system morbidity is not the key determinant of infant feeding. Maternal information regarding infant health and breastfeeding abilities may improve breastfeeding rates.

Incidental radiological findings during clinical tuberculosis screening in Lesotho and South Africa: a case series

BackgroundChest X-ray offers high sensitivity and acceptable specificity as a tuberculosis screening tool, but in areas with a high burden of tuberculosis, there is often a lack of radiological expertise to interpret chest X-ray. Computer-aided detection systems based on artificial intelligence are therefore increasingly used to screen for tuberculosis-related abnormalities on digital chest radiographies. The CAD4TB software has previously been shown to demonstrate high sensitivity for chest X-ray tuberculosis-related abnormalities, but it is not yet calibrated for the detection of non-tuberculosis abnormalities. When screening for tuberculosis, users of computer-aided detection need to be aware that other chest pathologies are likely to be as prevalent as, or more prevalent than, active tuberculosis. However, non­-tuberculosis chest X-ray abnormalities detected during chest X-ray screening for tuberculosis remain poorly characterized in the sub-Saharan African setting, with only minimal literature.Case presentationIn this case series, we report on four cases with non-tuberculosis abnormalities detected on CXR in TB TRIAGE + ACCURACY (ClinicalTrials.gov Identifier: NCT04666311), a study in adult presumptive tuberculosis cases at health facilities in Lesotho and South Africa to determine the diagnostic accuracy of two potential tuberculosis triage tests: computer-aided detection (CAD4TB v7, Delft, the Netherlands) and C-reactive protein (Alere Afinion, USA). The four Black African participants presented with the following chest X-ray abnormalities: a 59-year-old woman with pulmonary arteriovenous malformation, a 28-year-old man with pneumothorax, a 20-year-old man with massive bronchiectasis, and a 47-year-old woman with aspergilloma.ConclusionsSolely using chest X-ray computer-aided detection systems based on artificial intelligence as a tuberculosis screening strategy in sub-Saharan Africa comes with benefits, but also risks. Due to the limitation of CAD4TB for non-tuberculosis-abnormality identification, the computer-aided detection software may miss significant chest X-ray abnormalities that require treatment, as exemplified in our four cases. Increased data collection, characterization of non-tuberculosis anomalies and research on the implications of these diseases for individuals and health systems in sub-Saharan Africa is needed to help improve existing artificial intelligence software programs and their use in countries with high tuberculosis burden.

Role of Artificial Intelligence in CXR Interpretation in Pediatric Intensive Care Unit Ain Shams University

Background Artificial intelligence (AI) is a branch of computer science in which statistical techniques are used to classify, simulate, or optimize data based on previous observations. An AI system can interpret and prioritize abnormal chest X-rays with critical findings, potentially reducing the backlog of exams and bringing urgently needed care to patients more quickly. Aim of the Work Our study aims to test the diagnostic performance of the AI system in PICU CXR interpretation regarding its sensitivity, specificity, as well as its accuracy. we also investigate the effect of AI aid on residents performance regarding diagnostic accuracy, sensitivity time consumption. Patients and Methods This was a Cross sectional study which conducted at Pediatric Ain Shams University Hospitals from January to June 2021. The study was performed on convenient sample consists of 30 CXRs done at the Ain Shams Pediatric hospital intensive care unit and meeting the study inclusion criteria. Results Overall in this study we found that the resident diagnostic accuracy, sensitivity and time consumption improved across a large number of clinical CXRs findings as well as strengthening of the residents agreement with the expert radiologist when assisted by the deep-learning model. Effective implementation of the model has the potential to improve clinical practice. Research is underway to assess the generalizability of results to various clinical environments and health systems. Conclusion We raised the possibility of the potential use of AI systems in future radiology workflows for preliminary interpretations that target the most prevalent findings, leaving the final reads to be done by a more experienced radiologist in a suitable practice atmosphere to catch any potential misses from the less-prevalent fine-grained findings.

Accuracy of the NADAS Criteria in Helping Establish the Diagnosis in Children With Suspected to Have Congenital Heart Disease

Background : Prevalence of congenital heart disease is approximately 8/1000 live births and majority of them are diagnosed during evaluation of a murmur detected either during a routine follow-up or while assessment during an intercurrent illness. NADAS criteria can be used as screening test with hight a sensitivity and specificity regarding ability to differentiate pathologic from innocent murmurs.&#x0D; Aim : The aim of the study was to assess the accuracy of the NADAS criteria in helping to diagnose disease in children suspected of having congenital heart disease.&#x0D; Methods : This was a diagnostic study with a cross-sectional approach and Inclusion criteria were children aged 1 month - 12 years from history and physical examination found a history of recurrent ARI, poor nutritional status, failure to thrive, found cardiac murmurs, without/any cyanosis who attended the outpatient cardiology pediatric at Haji Adam Malik General Hospital between Januari - july 2021. Data Analysis was created using the program IBM SPSS Statistics version 25&#x0D; Results : 70 children were screened. 60 children were detected to have CHD. 63,3% of the study population in the age group &lt;5 year had a heart disease. All CHD patients had poor nutritional status, abnormal ECG and 88.33% had a history of recurrent ARI, 80% had abnormal chest x-ray, and 55% had abnormal S2. A grade 3 or higher systolic murmur was found in 78.33% of patients had CHD.&#x0D; Conclusion : NADAS criteria can be used as a screening test for helping to diagnose disease in children suspected of having CHD with accuracy 87.14%, sensitivity of 88.33%, specificity was 80% when score at 3 (1 major + 2 minor) and AUC value is 0.842.

Persisting symptoms common but inability to work rare: a one-year follow-up study of Finnish hospitalised COVID-19 patients

Background Difficulties in recovery persisting for months have been reported in patients with severe COVID-19. Our aim was to investigate respiratory and overall recovery one year after hospital discharge. Methods Finnish patients hospitalised due to COVID-19 during the first wave of the pandemic were recruited to a survey of symptoms, quality of life (RAND-36), work status, and health care use one year after hospital discharge. Patients with lung function test and chest x-ray results available from 3-6 months after hospital discharge underwent spirometry and a chest x-ray at one year. Results Ninety-six patients responded to the one-year survey, 32 underwent spirometry and 32 a chest x-ray. Of those working full-time before COVID-19, median duration of sick leave was 40 days and 10% had not returned to work at one year. Health-care service use related to COVID-19 after discharge was reported by 79%, 50% using primary care, 34% occupational health care and 32% specialist care, respectively. Tiredness, fatigue, and physical difficulties increased in follow-up (p = 0.022–0.033). Quality of life did not change. Chest x-ray abnormalities decreased in follow-up, with an abnormal chest x-ray in 58% at 3–6 months and 25% at one year. A restrictive spirometry pattern was more common at one year (16 vs. 34%, p = 0.014). Conclusions Prolonged symptoms are common, some patients have decreased lung function, and a small minority of patients still have not returned to work one year after severe COVID-19. This calls for further research into the underlying causes and risk factors for prolonged recovery.

Deep learning models for tuberculosis detection and infected region visualization in chest X-ray images

ObjectiveTuberculosis (TB) is among the most frequent causes of infectious-disease-related mortality. Despite being treatable by antibiotics, tuberculosis often goes misdiagnosed and untreated, especially in rural and low-resource areas. Chest X-rays are frequently used to aid diagnosis; however, this presents additional challenges because of the possibility of abnormal radiological appearance and a lack of radiologists in areas where the infection is most prevalent. Implementing deep-learning-based imaging techniques for computer-aided diagnosis has the potential to enable accurate diagnoses and lessen the burden on medical specialists. In the present work, we aimed to develop deep-learning-based segmentation and classification models for accurate and precise detection of tuberculosis in chest X-ray images, with visualization of infection using gradient-weighted class activation mapping (Grad-CAM) heatmaps. MethodsFirst, we trained the UNet segmentation model using 704 chest X-ray radiographs taken from the Montgomery County and Shenzhen Hospital datasets. Next, we implemented the trained UNet model on 1,400 tuberculosis and control chest X-ray scans to segment the lung region. The images were taken from the National Institute of Allergy and Infectious Diseases (NIAID) TB portal program dataset. Then, we applied the deep learning Xception model to classify the segmented lung region into tuberculosis and normal classes. We further investigated the visualization capabilities of the model using Grad-CAM to view tuberculosis abnormalities in chest X-rays and discuss them from radiological perspectives. ResultsFor segmentation by the UNet model, we achieved accuracy, Jaccard index, Dice coefficient, and area under the curve (AUC) values of 96.35%, 90.38%, 94.88%, and 0.99, respectively. For classification by the Xception model, we achieved classification accuracy, precision, recall, F1-score, and AUC values of 99.29%, 99.30%, 99.29%, 99.29%, and 0.999, respectively. The Grad-CAM heatmap images from the tuberculosis class showed similar heatmap patterns, where lesions were primarily present in the upper part of the lungs. ConclusionThe findings may verify our system's efficacy and superiority to clinician precision in tuberculosis diagnosis using chest X-rays and raise the possibility of a valuable setup, particularly in environments with a scarcity of radiological expertise.

The role of chest X-ray in coronavirus disease 2019 (COVID-19) infection: findings and correlation with clinical outcome.

It is possible to diagnose coronavirus disease 2019 (COVID-19) faster and more accurately with chest X-ray (CXR) and chest computed tomography (CT) than with reverse transcriptase PCR (RT-PCR) tests. The aim of this study was to verify the possibility of reducing the use of CT in diagnosis and follow-up of COVID-19 infection by using CXR. A total of 326 COVID-19 patients who were hospitalized in Ankara City Hospital were included in this retrospective study. A total of 326 patients were RT-PCR positive for COVID-19 infection; 178 were male (54.6%) and 148 were female (45.4%), with a median age of 45. Considering the results, the baseline CXR sensitivity in our experience was approximately 72%. The CXRs of 113 patients with abnormal CT were divided into 2 groups, the CXR normal and abnormal groups, and were then compared. In the 1st group with abnormal CXR, the mean age, the number of patients over 65 years old, and the comorbidity rate were higher. Additionally, it was determined that the number of patients requiring respiratory support and intensive care unit (ICU) admission in this 1st group was higher than in the 2nd group (with normal CXR). Most of the patients who died (91%, 10/11) were in Group 1. In the group with normal CXR, no patients in the critically ill category needed invasive or non-invasive mechanical ventilators. CXR can help in detecting clinically moderate and severe cases of COVID-19. CXR can assist clinicians in patient management and treatment planning regarding the clinical course, respiratory support, ICU need, and mortality and can help them prepare for potential negative outcomes.

Management of COVID-19 Pneumonia: Experience in a Peripheral Tertiary Level Hospital in Bangladesh

Background: Since the detection of the first case of the novel SARS-CoV-2 virus in Wuhan, China, on December 31, 2019, the world stepped into an era of the pandemic as declared by WHO on March 11, 2020. With no prior knowledge regarding this new pandemic, the evidence-based practice of medical management was at stake. Experience sharing from different corners of the world played an essential role in building knowledge to save the world. Methods: The researchers collected data for this retrospective study, from hospital records of the COVID-19 dedicated unit of Sylhet Women’s Medical College, Sylhet, Bangladesh, that was formed temporarily in July 2020, in response to the pandemic situation in the country. This article includes data on 3408 admitted COVID-19 patients with moderate to severe symptoms, managed in the isolation unit including its intensive care unit (ICU) up to November 2021. Results: The mean age of the patients was 57.21±12.58 years and male to female ratio was 1:1.8. Fever (44.4%), cough (38.9), and fatigability (33.3%) were the common symptoms. Diabetes mellitus (67.7%), systemic hypertension (63.7%), asthma (8.65%), chronic obstructive pulmonary disease (21.94%), and, ischemic heart disease (31.2%) were the top co-morbid conditions. Of the study subjects, 68.2% were RT-PCR positive for COVID-19, 67.3% had an abnormal baseline chest X-ray, and 91.1% had a ground glass shadow on a high-resolution CT scan of chest. Average C-reactive protein, D-dimer, and serum ferritin values were 71.68±60.38, 1.16±2.11, and 839±748.57 respectively. Hypoxemia, which was a common problem among them, was managed with oxygen therapy with an appropriate delivery device, with or without supplementation by dexamethasone (78.3%), or methylprednisolone (18.8%). Eighteen (18) patients refused to receive any antiviral therapy, 61 received oral Favipiravir and the rest of them received injectable Remdesevir. Standard care was augmented with Baricitinib in 60 patients. Thirty-six (36) patients in the ICU, with poor response to standard care, were treated with Tocilizumab. Death rates were 9.46% in the non-ICU block, but they were 6-10 times higher in the ICU. Conclusion: Every medical professional learned from the initial two years of the COVID-19 pandemic. Lessons learnt from experience helped building up the knowledge pool necessary for generating effective protocols for COVID-19 pneumonia management.

Joint Self-Supervised Image-Volume Representation Learning with Intra-inter Contrastive Clustering

Collecting large-scale medical datasets with fully annotated samples for training of deep networks is prohibitively expensive, especially for 3D volume data. Recent breakthroughs in self-supervised learning (SSL) offer the ability to overcome the lack of labeled training samples by learning feature representations from unlabeled data. However, most current SSL techniques in the medical field have been designed for either 2D images or 3D volumes. In practice, this restricts the capability to fully leverage unlabeled data from numerous sources, which may include both 2D and 3D data. Additionally, the use of these pre-trained networks is constrained to downstream tasks with compatible data dimensions. In this paper, we propose a novel framework for unsupervised joint learning on 2D and 3D data modalities. Given a set of 2D images or 2D slices extracted from 3D volumes, we construct an SSL task based on a 2D contrastive clustering problem for distinct classes. The 3D volumes are exploited by computing vectored embedding at each slice and then assembling a holistic feature through deformable self-attention mechanisms in Transformer, allowing incorporating long-range dependencies between slices inside 3D volumes. These holistic features are further utilized to define a novel 3D clustering agreement-based SSL task and masking embedding prediction inspired by pre-trained language models. Experiments on downstream tasks, such as 3D brain segmentation, lung nodule detection, 3D heart structures segmentation, and abnormal chest X-ray detection, demonstrate the effectiveness of our joint 2D and 3D SSL approach. We improve plain 2D Deep-ClusterV2 and SwAV by a significant margin and also surpass various modern 2D and 3D SSL approaches.

Indications, Efficacy, and Complications of Pediatric Bronchoscopy: A Retrospective Study at a Tertiary Center.

Background Bronchoscopy is an essential procedure for evaluating, diagnosing, and treating pediatric respiratory diseases. In this study, we demonstrate the indications and contraindications of bronchoscopy done in a tertiary referral hospital, Al Qassimi Woman's and Children's Hospital (AQWCH) in Sharjah, United Arab Emirates (UAE), in order to achieve better service. This study aims to evaluate patients' characteristics, diagnostic and therapeutic indications, and complications of bronchoscopy. Material and method This retrospective chart review included children aged between one day and 13 years, admitted to AQWCH, who underwent bronchoscopy (rigid or flexible) procedures between January 2018 and December 2019. All patients were identified by using a computerized search of hospital discharge diagnosis, which was codified as "pediatric bronchoscopy, flexible, rigid, bronchoalveolar lavage". The main study outcome measure was to evaluate patients' characteristics, diagnostic or therapeutic indications, bronchoalveolar lavage (BAL) analysis, as well as complications of bronchoscopy at AQWCH. Results There were 72 pediatric bronchoscopies (rigid and flexible) performed in patients aged less than 13 years old; the reason for bronchoscopy procedure was diagnostic in 51% and both diagnostic and therapeutic in 49%. Cough was the most common symptom (n=53; 74%), and chest recession was the most common clinical finding (n=46; 64%). Foreign body aspiration was the main indication (n=23; 32%), followed by stridor (26%). Consolidation was the most common radiological finding. Foreign body was the common finding, seen in 25% of bronchoscopies, followed by tracheomalacia in 17%. The suspected diagnosis was confirmed in 89%, and management change was needed in 54% of patients. The main complication during the procedure was desaturation (26%), and cough was the main post-bronchoscopy complication (14%). BAL was done for 28 (39%) patients, in which BAL culture was positive in 75%.Rigid bronchoscopy was done when foreign body aspiration was suspected based on positive history in 70%, abnormal physical examination in 60%, and chest X-ray abnormalities in 39% of patients. Sensitivity and specificity for patient history, physical examination, and chest X-ray were 80% and 83%, 66% and 60%, and 40% and 66 %, respectively. Conclusion Bronchoscopy is an important tool for evaluating, diagnosing, and treating pediatric respiratory diseases. While it is a safe procedure, it still needs a careful selection of patients as it is invasive.

Survey on deep learning techniques for thorax disease classification task

In recent years, medical image analysis has been widely studied, largely because it provides useful tools to support daily clinical routines. Thorax disease classification (TDC) is an important one among these tools since Chest X-ray (CXR) has become prevalent in daily clinical examinations. It aims to distinguish between normal and abnormal CXR images according to different diseases. To this aim, many thorax disease classification systems have been proposed to increase performance. Most of them adopt deep learning techniques to train a deep neural network (DNN) by using labeled CXR datasets. The TDC performance has been continually refined. Aiming to provide future researchers with the work being done on TDC to date, we review the DNN models for this task. A brief review of each method along with their evaluations on a set of benchmark datasets is included. Moreover, we give a detailed comparison of these methods as well as a conclusion.