Chest X-ray Research Articles

Efficient federated learning for pediatric pneumonia on chest X-ray classification

According to the World Health Organization (WHO), pneumonia kills about 2 million children under the age of 5 every year. Traditional machine learning methods can be used to diagnose chest X-rays of pneumonia in children, but there is a privacy and security issue in centralizing the data for training. Federated learning prevents data privacy leakage by sharing only the model and not the data, and it has a wide range of application in the medical field. We use federated learning method for classification, which effectively protects data security. And for the data heterogeneity phenomenon existing in the actual scenario, which will seriously affect the classification effect, we propose a method based on two-end control variables. Specifically, based on the classical federated learning FedAvg algorithm, we modify the loss function on the client side by adding a regular term or a penalty term, and add momentum after the average aggregation on the server side. The federated learning approach prevents the data privacy leakage problem compared to the traditional machine learning approach. In order to solve the problem of low classification accuracy due to data heterogeneity, our proposed method based on two-end control variables achieves an average improvement of 2% and an accuracy of 98% on average, and 99% individually, compared to the previous federated learning algorithms and the latest diffusion model-based method. The classification results and methodology of this study can be utilized by clinicians worldwide to improve the overall detection of pediatric pneumonia.

Diagnosis and Treatment of Acute Heart Failure: A Retrospective Observational Study and Medical Audit.

Background: Acute Heart Failure (AHF) is a leading cause of hospitalizations and remains a significant socioeconomic burden. Despite advances in medical care, mortality and rehospitalization rates remain high. Previous AHF audits have revealed regional differences and a poor adherence to guidelines. This study aimed to assess guideline adherence in a public teaching hospital to identify areas for improvement. Methods: This retrospective observational study examined clinical routine data of patients hospitalized for AHF at a Swiss public teaching hospital between 2018 and 2019. AHF management was evaluated against the relevant guidelines of the European Society of Cardiology. Results: The study included 760 AHF cases of 726 patients (median age 84 years, range 45-101, 50% female). NT-pro-BNP levels were measured in 92% of the cases. Electrocardiography was performed in 95% and chest X-rays in 90% of cases. Echocardiography was conducted in 54% of all cases and in 63% of newly diagnosed AHF cases. Intravenous furosemide was initiated in 76%. In the subgroup of cases with reduced ejection fraction (HFrEF), 86% were discharged with beta-blockers and 69% with angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers. Among cases with left ventricular ejection fraction ≤ 35%, mineralocorticoid receptor antagonists were prescribed in 55%. Conclusions: We observed generally good adherence to guideline recommendations. However, several improvements are needed in initial assessment and documentation, diagnostic procedures such as echocardiography, discharge medication, and lifestyle recommendations. Compared to other studies, our diagnostic workup was more aligned with guidelines, the use of intravenous diuretics was similar, and the duration of hospital stay and mortality rates were comparable.

Machine learning in the prevention of heart failure.

Heart failure (HF) is a global pandemic with a growing prevalence and is a growing burden on the healthcare system. Machine learning (ML) has the potential to revolutionize medicine and can be applied in many different forms to aid in the prevention of symptomatic HF (stage C). HF prevention currently has several challenges, specifically in the detection of pre-HF (stage B). HF events are missed in contemporary models, limited therapeutic options are proven to prevent HF, and the prevention of HF with preserved ejection is particularly lacking. ML has the potential to overcome these challenges through existing and future models. ML has limitations, but the many benefits of ML outweigh these limitations and risks in most scenarios. ML can be applied in HF prevention through various strategies such as refinement of incident HF risk prediction models, capturing diagnostic signs from available tests such as electrocardiograms, chest x-rays, or echocardiograms to identify structural/functional cardiac abnormalities suggestive of pre-HF (stage B HF), and interpretation of biomarkers and epigenetic data. Altogether, ML is able to expand the screening of individuals at risk for HF (stage A HF), identify populations with pre-HF (stage B HF), predict the risk of incident stage C HF events, and offer the ability to intervene early to prevent progression to or decline in stage C HF. In this narrative review, we discuss the methods by which ML is utilized in HF prevention, the benefits and pitfalls of ML in HF risk prediction, and the future directions.

UNSEEN TUBERCULOSIS: STUDY AMONG INPATIENTS OF A TERTIARY CARE HOSPITAL OF NORTH KARNATAKA

Objective: Tuberculosis (TB) remains a significant global health challenge, particularly in high-burden countries like India. However, high proportion of people never diagnosed or treated. Delayed diagnosis poses a significant threat to hospital inpatients, health-care workers, and medical trainees. Hence, systematic screening of high-risk groups, including hospital inpatients, has been recommended as a strategy to identify TB. Using signs and symptoms as an independent screening tool remains a global challenge for its relatively low accuracy. However, it could still be of value when it was used in a specific target population like inpatients of hospital. Hence, this study is aimed to determine the proportion of presumptive and active TB cases among inpatients. Methods: A hospital-based cross-sectional study was conducted among inpatients above 18 years of age admitted in non-emergency wards admitted during March to April 2023 at ESIC Hospital in Kalaburagi. A pre-existing TB diagnosis and patients admitted to critical care units were excluded. The convenient sampling method was used to recruit participants. Data were collected through using a pre-designed, semi-structured questionnaire. The patients who presented with one of the four symptoms for screening had undergone investigations such as cartridge-based nucleic acid amplification test, chest X-ray, and sputum microscopy to confirm the diagnosis, and the findings were recorded as proportion. Results: Out of 1363 inpatients, 35 (2.5%) patients had classified as presumptive TB case. Out of 35, 3 (8.57%) were found to be actual TB cases. All patients who had positive results for TB were young patients and belonged to low socioeconomic class. Two of three were rifampicin-sensitive TB cases. Conclusion: Timely screening of the patient’s is of utmost importance to avoid further spread of TB infection among inpatients. The higher prevalence of actual TB cases in younger adults needs further exploration.

Deep learning on chest X-ray and computed tomography scans for detection of COVID-19 as a part of a network-centric digital health stack for future pandemics

Developing a reliable rapid screening protocol for highly infectious diseases like COVID-19 is of paramount interest since it facilitates the isolation of infected patients from the rest of the population. Reverse-transcription polymerase chain reaction (RT-PCR) test is presently the most widely accepted gold-standard test to detect COVID-19. In this method, the RNA of the virus is duplicated by a process called reverse transcription to form DNA for facilitating the copying process. Fluorescent dye is attached to the viral genetic material and copied billions of times through the process called polymerase chain reaction. Enhanced fluorescence is used to identify the presence of genetic material of the virus. These tests are time-consuming and have significant false negatives, i.e., a person with COVID-19 might be categorized as not having the virus. Large-scale RT-PCR testing has its own share of problems such as logistics, availability and affordability in underdeveloped nations, and reliability of the test results. Machine learning algorithms can act as a cheaper supplementary/alternative diagnostic tool for the testing process. In the current study, using publicly available chest X-ray image datasets, different convolutional neural network (CNN)-based models were developed for efficient identification of COVID-19 infected patients, and their efficacies were compared. Key innovations in training the CNNs are discussed. Our results indicate that EfficientNet, SeResNext, and ResNet are best at classifying normal, pneumonia and COVID-19 cases, respectively. The ResNet architecture with transfer learning performed best at detecting COVID-19 with an accuracy of 94%, a rate far superior to that in the RT-PCR test, which is typically in the range of 70 – 80%. This is particularly attractive as an additional noninvasive protocol since such technology-augmented detection is likely to help in reducing the psychological refractory period due to COVID-19 infections. Toward the healthy lung initiative in the post-COVID-19 era, we propose close coupling of the present diagnostic protocols with digital approaches to ensure more reliable personal care within the ambit of large-scale pandemic control mechanisms. Such integration with emerging technological tools can create a benchmark for the first line of defense against future global pandemics.

Nuts and bolts of lung ultrasound: utility, scanning techniques, protocols, and findings in common pathologies.

Point of Care ultrasound (POCUS) of the lungs, also known as lung ultrasound (LUS), has emerged as a technique that allows for the diagnosis of many respiratory pathologies with greater accuracy and speed compared to conventional techniques such as chest x-ray and auscultation. The goal of this narrative review is to provide a simple and practical approach to LUS for critical care, pulmonary, and anesthesia providers, as well as respiratory therapists and other health care providers to be able to implement this technique into their clinical practice. In this review, we will discuss the basic physics of LUS, provide a hands-on scanning technique, describe LUS findings seen in normal and pathological conditions (such as mainstem intubation, pneumothorax, atelectasis, pneumonia, aspiration, COPD exacerbation, cardiogenic pulmonary edema, ARDS, and pleural effusion) and also review the training necessary to achieve competence in LUS.

Assessing the Serum Triglyceride and Fasting plasma Glucose Ratio as a Predictor of Left Ventricular Diastolic Dysfunction in Type 2 Diabetes Mellitus, A Cross-sectional study.

Introduction: Cardiovascular complications, including left ventricular diastolic dysfunction (LVDD), are major concerns in Type 2 Diabetes Mellitus (T2DM) patients, often preceding heart failure and increasing mortality. Early detection of LVDD is essential to improving outcomes, yet current diagnostic methods are often costly and impractical for routine screening. This study aimed to evaluate the correlation between the serum triglyceride to fasting plasma glucose (TG/FPG) ratio and LVDD in T2DM patients, as a potential cost-effective biomarker for cardiovascular risk stratification. Methodology: This observational analytical study was conducted over a period of one year at a tertiary care hospital in northeast India to assess the correlation of TG/FPG ratio with LVDD in patients with T2DM. A total of 100 participants, aged 18 years or above and diagnosed with T2DM, were included. Exclusion criteria comprised Type 1 Diabetes Mellitus, pregnancy, pre-existing hypertension, cardiovascular diseases, and patients denying consent. Participants were recruited using convenient sampling, with every alternate eligible patient enrolled until the target sample size was reached. Data collection included measuring fasting plasma glucose (FPG) and postprandial blood glucose, HbA1c, fasting lipid profile, kidney function test, and urinalysis. The cardiovascular evaluation consisted of chest radiography, electrocardiogram, and echocardiography using standardized parameters to classify LVDD into grades I, II, and III, based on E/A and E/e’ ratios. Statistical analysis was performed using SPSS version 15, with Chi-square test and student’s t-tests applied for significance (p-value < 0.05), and regression analysis was used to identify associations. Ethical approval was obtained, and participant confidentiality was maintained throughout the study. Result: A cross-sectional study was conducted over one years at a tertiary hospital in northeast India, enrolling 100 participants with T2DM, excluding those with comorbidities or cardiovascular symptoms. The mean age of participants was 45.52 years, with a male-to-female ratio of 1.21:1. Diastolic dysfunction was evaluated using echocardiography, and the TG/FPG ratio was calculated for each patient. LVDD was found in 58.7% of the participants, classified into grade I (40.4%), grade II (13.5%), and grade III (4.8%). Statistical analysis revealed a significant positive correlation between the TG/FPG ratio and LVDD (p- value= 0.0001). Higher TG/FPG ratios were associated with worse diastolic function, suggesting that this ratio reflects combined metabolic disturbances contributing to myocardial dysfunction. Patients with LVDD had significantly higher lipid profiles and poorer glycemic control compared to those with normal LV function. The TG/FPG ratio, being easily obtainable from routine laboratory tests, offers a convenient tool for identifying individuals at higher risk for LVDD. Conclusion: The TG/FPG ratio may serve as a simple, accessible, and cost-effective marker for early detection of LVDD in T2DM patients. Incorporating this ratio in clinical practice could improve cardiovascular risk stratification and facilitate timely interventions, ultimately reducing the burden of cardiovascular disease in diabetic populations.

Deep Conformal Supervision: Leveraging Intermediate Features for Robust Uncertainty Quantification.

Trustworthiness is crucial for artificial intelligence (AI) models in clinical settings, and a fundamental aspect of trustworthy AI is uncertainty quantification (UQ). Conformal prediction as a robust uncertainty quantification (UQ) framework has been receiving increasing attention as a valuable tool in improving model trustworthiness. An area of active research is the method of non-conformity score calculation for conformal prediction. We propose deep conformal supervision (DCS), which leverages the intermediate outputs of deep supervision for non-conformity score calculation, via weighted averaging based on the inverse of mean calibration error for each stage. We benchmarked our method on two publicly available datasets focused on medical image classification: a pneumonia chest radiography dataset and a preprocessed version of the 2019 RSNA Intracranial Hemorrhage dataset. Our method achieved mean coverage errors of 16e-4 (CI: 1e-4, 41e-4) and 5e-4 (CI: 1e-4, 10e-4) compared to baseline mean coverage errors of 28e-4 (CI: 2e-4, 64e-4) and 21e-4 (CI: 8e-4, 3e-4) on the two datasets, respectively (p < 0.001 on both datasets). Based on our findings, the baseline results of conformal prediction already exhibit small coverage errors. However, our method shows a significant improvement on coverage error, particularly noticeable in scenarios involving smaller datasets or when considering smaller acceptable error levels, which are crucial in developing UQ frameworks for healthcare AI applications.

Children with sickle cell disease: are they protected from serious COVID-19?

COVID-19, the pandemic that hit the world in 2020, resulted in millions of deaths, with the elderly and adults succumbing to the disease more often than children. However, the presence of underlying morbidities increased the risk of death. Sickle cell disease (SCD) was previously classified as a major risk factor for severe COVID-19 disease. However, presently, there are only a limited number of studies that identify the clinical course of children with SCD and COVID-19. We conducted a retrospective observational study on children with SCD admitted due to COVID-19 at three different institutions in Saudi Arabia between March 2020 and March 2022. We studied the demographic and clinical characteristics of patients admitted to the hospital. Seventy-six patients with SCD had PCR-confirmed SARS-CoV-2 during the study period; 50.0% of our patient population were children (6-12 years old). Gender was evenly distributed, with 53.9% girls and 46.1% boys. Symptoms more commonly related to the COVID-19 infection included fever, cough, malaise, and vomiting. Chest x-ray findings revealed mild and non-specific symptoms only in approximately one-third (28) of the included children. The most common symptoms associated with SCD were vaso-occlusive crisis (47.4%) and abdominal pain (11.8%). The overall general appearance of most of the patients was reassuring. The median length of hospital stay was 4.2 ± 2.7 days. The mean white blood cell count was 11.4 ± 5.2 × 109/L, and the mean hemoglobin level was 8.3 ± 1.5 g/dl. Despite the fact that higher levels of mean D-dimer, lactate dehydrogenase, and ferritin were reported in these patients, the clinical outcome was not affected. All recruited patients received hydroxyurea as maintenance therapy. The outcome of our study was reassuring, with no significant morbidity or mortality observed among the recruited patients. Despite SCD being a chronic disease with known specific complications, there has been a claim that COVID-19 infection adds further risk. The results of this study suggest that the overall outcome of COVID-19 was favorable, with no reported mortality. Further research is needed to understand the factors that contributed to this favorable outcome. In children with SCD, it is still questionable whether hydroxyurea is one of the protective factors against severe COVID-19. Validation through large-scale research is recommended.

Synthetic chest X-ray images from text prompts.

Synthetic chest X-ray images from text prompts.

Comprehensive Investigation of Demographic, Clinical, and Paraclinical Findings in Hospitalized Children with COVID-19

Background: Given the differing characteristics of COVID-19 in children compared to adults, there is a need for comprehensive investigations of associated factors in hospitalized children. Objectives: Therefore, we aimed to assess demographic, clinical, and paraclinical findings in hospitalized children with COVID-19. Methods: This study divided hospitalized patients aged three months to 18 years with a positive polymerase chain reaction (PCR) test into two groups: Severe and mild cases. Risk factors, including age, sex, weight, place of residence, clinical symptoms, lung involvement in chest X-rays, and laboratory parameters, were compared between these two groups (mild and severe). Results: The study included 375 hospitalized patients (75 severe and 300 mild cases). The mean age was 7.50 ± 5.57 months. Among the patients, 215 males (58%) were enrolled. Severe cases exhibited a higher frequency of sleepiness (40%) and shortness of breath (41.3%). Clinical signs such as respiratory rate (35 breaths/min) and heart rate (120 bpm) were higher in severe cases, while blood oxygen levels (92%) were lower. Severe cases also showed higher lung involvement in radiographs (85.7%). Among laboratory factors, white blood cell count (8650), neutrophils (6762), and lymphocytes (2090) were higher in severe cases, while alkaline phosphatase (320.5) was lower. Conclusions: Our results highlighted key findings regarding demographic, clinical, and paraclinical features in hospitalized children with COVID-19, offering valuable insights into this novel phenomenon. The study recommends close monitoring of patients with symptoms such as shortness of breath and sleepiness to prevent disease progression to a severe state.

Diagnostic accuracy and added value of dynamic chest radiography in detecting pulmonary embolism: A retrospective study

PurposeThis study aimed to assess the diagnostic performance of dynamic chest radiography (DCR) and investigate its added value to chest radiography (CR) in detecting pulmonary embolism (PE). MethodsOf 775 patients who underwent CR and DCR in our hospital between June 2020 and August 2022, individuals who also underwent contrast-enhanced CT (CECT) of the chest within 72 h were included in this study. PE or non-PE diagnosis was confirmed by CECT and the subsequent clinical course. The enrolled patients were randomized into two groups. Six observers, including two thoracic radiologists, two cardiologists, and two radiology residents, interpreted each chest radiograph with and without DCR using a crossover design with a washout period. Diagnostic performance was compared between CR with and without DCR in the standing and supine positions. ResultsSixty patients (15 PE, 45 non-PE) were retrospectively enrolled. The addition of DCR to CR significantly improved the sensitivity, specificity, accuracy, and area under the curve (AUC) in the standing (35.6–70.0 % [P < 0.0001], 84.8–93.3 % [P = 0.0010], 72.5–87.5 % [P < 0.0001], and 0.66–0.85 [P < 0.0001], respectively) and supine (33.3–65.6 % [P < 0.0001], 78.5–92.2 % [P < 0.0001], 67.2–85.6 % [P < 0.0001], and 0.62–0.80 [P = 0.0002], respectively) positions for PE detection. No significant differences were found between the AUC values of DCR with CR in the standing and supine positions (P = 0.11) or among radiologists, cardiologists, and radiology residents (P = 0.14–0.68). ConclusionsIncorporating DCR with CR demonstrated moderate sensitivity, high specificity, and high accuracy in detecting PE, all of which were significantly higher than those achieved with CR alone, regardless of scan position, observer expertise, or experience.

5086 Approaching the Enigma: Unveiling a Pituitary Stalk Lesion with Imaging and Hormonal Patterns

Abstract Disclosure: N. Rodriguez-Morales: None. P. Guido: None. Background: Pituitary stalk lesions (PSL) pose a diagnostic dilemma for endocrinologists. Multiple studies have attempted to associate hormonal imbalances and imaging patterns to a specific etiology (1,2). However, the diagnosis remains a mystery on many occasions. We describe a case of secondary hypogonadism and central DI attributed by a PSL. Clinical Case: A 39-year-old male presented with symptoms of hypogonadism and polyuria. Initial workup was consistent with secondary hypogonadism; decreased free and total testosterone levels (1.8 pg/dL and 9 ng/dL, n&amp;gt;35 pg/dL, n&amp;gt;250 ng/dL), LH (0.2 mIU/mL, n&amp;gt;1.5 mIU/mL), and FSH (&amp;lt;0.7 mIU/mL, n&amp;gt;1.6 mIU/mL). The remaining anterior pituitary hormones were within normal limits. The patient was also polyuric (6.5L urine output in 24hr). HgbA1c was normal. Urine Na, urine Osm demonstrated water diuresis (13 mmol/L and 54 mOsm/kg, respectively). A water deprivation test was performed. Initial urine Osm after water restriction was 119 mOsm/kg, which more than tripled to 575 mOsm/kg after administration of desmopressin, diagnostic of central DI. Pituitary MRI showed a well-circumscribed thickening of the pituitary stalk (6x6x5mm), with uniform enhancement. Evaluation for etiology to rule out germ cell tumor, lymphoma, sarcoidosis, opportunistic infection, tuberculosis, Langerhans cell histiocytosis showed a normal serum hCG, HIV, ACE, LDH, AFP. A lumbar puncture demonstrated normal cytology and was negative for AFB. Whole body bone scintigraphy and chest X-ray were normal. Pituitary deficits were treated with hormone replacement (testosterone and desmopressin), and a repeat MRI showed a stable lesion. Conclusion: Noninvasive testing is key to the evaluation of a pituitary stalk lesion. Careful consideration of inflammatory, infectious, and neoplastic etiologies along with attention to pituitary deficits is very important. Our patient most likely has a craniopharyngioma based on data from other case series (1,2). This case serves to bring additional data to the body of literature on this complicated clinical scenario.

8472 A Little Too Much Of Tums: A Unique Presentation Of Milk Alkali Syndrome

Abstract Disclosure: Z. Al-Hiti: None. A. Misbah: None. Milk-Alkali Syndrome is characterized by a triad of hypercalcemia, metabolic alkalosis, and renal dysfunction. Historically associated with the excessive consumption of milk and alkali-containing antacids, it has re-emerged in recent years due to the widespread use of calcium and vitamin D supplements. It primarily results from an imbalance between calcium intake and renal excretion, leading to elevated calcium levels in the bloodstream. We present a case of milk-alkali syndrome with persistent renal dysfunction after normalizing calcium levels. We present the case of a 65-year-old male who presented to the emergency department with generalized weakness, poor oral intake, and confusion for 3 days. He was noted to have been more adherent with taking his prescribed medications over the last 2 weeks, which include calcitriol 0.5 mcg BID, calcium acetate 667 mg TID, and calcium carbonate 800 mg TID. On admission, he was afebrile and hemodynamically stable. . On examination, he had dry mucous membranes. He had no focal neurological deficits but he was confused and lethargic, and was only oriented to self. Other examination was noncontributory. Complete blood count and urinalysis were largely unremarkable. Chemistry demonstrated an elevated calcium level of 16.0 (8.3-10.6 mg/dL), ionized calcium of 7.3 (4.6-5.3 mg/dL), sodium of 134 (135-145 mmol/L), potassium of 5.4 (3.5-5.1 mmol/L), high phosphorus of 5.3 (2.4-5.1 mg/dL), and bicarbonate of 36 (20-31 mEq/L). Creatinine was elevated at 2.9 (0.6-1.1 mg/dL) from baseline of 1.3 to 1.4 mg/dL, and he had a BUN of 40(9-23 mg/dL). Venous blood gas showed a pH of 7.54( 7.32-7.42), pCO2 of 56 (40-52 mm Hg), and bicarbonate of 38 (23-28 mEq/L), indicating metabolic alkalosis. TSH was elevated at 36.15 (0.55- 4.78 uIU/mL) with a level of 140 two weeks back. Chest X-ray, renal ultrasound and CT head all showed no acute disease. Further work-up revealed a low PTH level of 6.3 (18.4-80 pg/mL), normal PTHrP level and 25-OH Vitamin D. There was no monoclonal gammopathy. Management included aggressive IV fluid resuscitation and IV calcitonin. However, no zoledronic acid was given due to a history of parathyroidectomy and risk of hypocalcemia. His serum calcium improved and eventually normalized. His confusion resolved. However, his creatinine and BUN increased to 3.4 mg/dL and 77, respectively, and did not improve to his baseline. This case highlights the importance of recognizing Milk-Alkali Syndrome as a potential consequence of over-the-counter calcium supplementation and highlights the potential for acute kidney injury which could be irreversible. Healthcare providers should be vigilant in identifying patients at risk, which includes older patients and patients with chronic kidney disease. Early recognition and appropriate management are crucial to prevent further renal deterioration and improve outcomes in individuals with Milk-Alkali Syndrome. Presentation: 6/2/2024

A full-scale lung image segmentation algorithm based on hybrid skip connection and attention mechanism

The segmentation accuracy of the lung images is affected by the occlusion of the front background objects. To address this problem, we propose a full-scale lung image segmentation algorithm based on hybrid skip connection and attention mechanism (HAFS). The algorithm uses yolov8 as the underlying network and enhancement of multi-layer feature fusion by incorporating dense and sparse skip connections into the network structure, and increased weighting of important features through attention gates. Finally the proposed algorithm was applied to the lung datasets Montgomery County chest X-ray and Shenzhen chest X-ray. The experimental results show that the proposed algorithm improves the precision, recall, pixel accuracy, Dice, mIoU, mAP and GFLOPs metrics compared to the comparison algorithms, which proves the advancement and effectiveness of the proposed algorithm.

Radiographically confirmed pneumonia in Malawian children and associated pneumococcal carriage after introduction of the 13-valent pneumococcal conjugate vaccine

BackgroundThe 13-valent pneumococcal conjugate vaccine (PCV-13) was introduced in Malawi in 2011 with an expected impact of reducing pneumococcal pneumonia in children. We aimed to describe clinical characteristics and nasopharyngeal (NP) carriage of pneumococcus by serotype in children hospitalized with primary end-point pneumonia (PEP) between 2013 and 19 after the introduction of PCV-13.MethodsWe conducted a secondary analysis of children aged under-5-years hospitalized with acute respiratory illness (ARI) in Malawi. Chest radiographs conducted at admission were read by two independent clinicians according to WHO criteria for PEP, and a third reviewer resolved discordant diagnoses. NP swab specimens were processed and Streptococcus pneumoniae growth was serotyped. Multivariable regression analysis was conducted to assess the association between clinical characteristics, NP serotypes, and PEP.ResultsWe had complete radiographic and NP serotype data for 500 children, of which 54 isolates were vaccine-type (VT) (10.8%), 165 were non-VT (NVT; 33.0%), and 281 had no pneumococcal growth (56.2%). Among these, 176 (35.2%) had PEP on chest x-ray. Among those with PEP, pneumococcal carriage was documented in 43.8% of cases, and VT serotypes accounted for 10.8%. For children with PEP, we found no association between clinical characteristics and carrying either VT, NVT, or no pneumococcus.ConclusionCarriage of S. pneumoniae remains high among children hospitalized with ARI in Malawi, but children with VT carriage were no more likely to have PEP than children carrying no pneumococcus or those with NVT carriage. There were no differences in clinical characteristics between those carrying VT, NVT, or no pneumococcus.

Significantly improving zero-shot X-ray pathology classification via fine-tuning pre-trained image-text encoders

Deep neural networks are increasingly used in medical imaging for tasks such as pathological classification, but they face challenges due to the scarcity of high-quality, expert-labeled training data. Recent efforts have utilized pre-trained contrastive image-text models like CLIP, adapting them for medical use by fine-tuning the model with chest X-ray images and corresponding reports for zero-shot pathology classification, thus eliminating the need for pathology-specific annotations. However, most studies continue to use the same contrastive learning objectives as in the general domain, overlooking the multi-labeled nature of medical image-report pairs. In this paper, we propose a new fine-tuning strategy that includes positive-pair loss relaxation and random sentence sampling. We aim to improve the performance of zero-shot pathology classification without relying on external knowledge. Our method can be applied to any pre-trained contrastive image-text encoder and easily transferred to out-of-domain datasets without further training, as it does not use external data. Our approach consistently improves overall zero-shot pathology classification across four chest X-ray datasets and three pre-trained models, with an average macro AUROC increase of 4.3%. Additionally, our method outperforms the state-of-the-art and marginally surpasses board-certified radiologists in zero-shot classification for the five competition pathologies in the CheXpert dataset.

9332 Breathless Struggle: Unraveling The Interplay Between DKA And Acute Respiratory Distress Syndrome

Abstract Disclosure: S. Zahra: None. R. Subramani: None. M. Abbas: None. K.Y. Hasan: None. F. Manas: None. C. Bajracharya: None. Introduction: Diabetic Ketoacidosis (DKA) is a life threatening emergency with mortality rate of 0.15%-0.30%. Acute Respiratory Distress Syndrome (ARDS) is a rare complication of DKA which presents as a sudden onset of dyspnea and progressive hypoxia. It is an inflammatory process that damages the alveolar-capillary membrane leading to the fluid leakage into the alveolar space. The incidence of ARDS in DKA remains unknown. It is more commonly seen in adolescents and young adults. The exact pathogenesis of ARDS in DKA is currently unknown. Some proposed mechanisms are that a severe metabolic acidosis in DKA along with increased cytokines production due to long standing hyperglycemia leads to increased capillary membrane permeability allowing fluid to accumulate in the alveolar space. TNF-a and IL-6 along with other pro-inflammatory cytokines are known to cause endothelial damage and increase pulmonary permeability leading to pulmonary edema. However most DKA patients with severe metabolic acidosis do not develop ARDS. Case Presentation:Case 1: A 19 year old male with uncontrolled type I diabetes mellitus presented with altered mental status. On examination, he was tachycardic and tachypneic with dry mucous membranes. Lab workup revealed leukocytosis, severe hyperglycemia, and severe anion gap metabolic acidosis with beta hydroxybutyrate level of 14.20 mmol/L (normal: 0.02-0.27 mmol/L). He was diagnosed with DKA and started on intravenous fluids, and insulin as per institutional DKA protocol. Approximately 12 hours post-admission, the patient experienced sudden-onset dyspnea with deteriorating hypoxia, necessitating emergent intubation and mechanical ventilation. Chest radiography displayed bilateral patchy ground-glass opacifications consistent with acute respiratory distress syndrome (ARDS). Despite the administration of 8 liters of intravenous fluids, the patient exhibited a negative net fluid balance of 2.3 liters, implicating fluid shift. Further investigations, including bronchoscopy and bronchoalveolar lavage, revealed an unobstructed airway, minimal secretions, and no evidence of hemorrhage or infection. Comprehensive infectious work-ups yielded negative results as well, reinforcing the diagnosis of DKA as the underlying cause of ARDS. Conclusion: Acute Respiratory Distress Syndrome is a rare but potentially fatal complication of Diabetic Ketoacidosis. Early recognition of this serious pulmonary complication of DKA and its prompt management is essential to prevent respiratory failure and mortality. Presentation: 6/1/2024

8238 Thyroidectomy for Thyroid Storm Complicated by Drug-Induced Hepatotoxicity and Thyrotoxic Cardiopulmonary Disease

Abstract Disclosure: T.A. John: None. A.C. Suarez: None. J. Genkil: None. S. Patil: None. C. Anastasopoulou: None. Case Presentation: A 59-year-old female with hypertension and type 2 diabetes presented with palpitations, and several months of hyperthyroid symptoms, lower extremity swelling, and dyspnea on exertion. On presentation, patient was diaphoretic, tachycardic, volume overloaded and had symmetric swelling of the thyroid gland. Lab results revealed FT4 3.28 ng/dL, total T3 384.59 ng/dL, TSH &amp;lt; 0.01 µIU/mland elevated TSI. Electrocardiogram showed sinus tachycardia. Chest radiography showed pulmonary edema. Echocardiography showed a reduced ejection fraction of 35%, dilation of right ventricle, severe mitral and tricuspid regurgitation, and severely elevated pulmonary artery systolic pressure (55-60 mm Hg). Thyroid ultrasound revealed enlarged, heterogeneous thyroid gland with normal vascularity. The diagnosis of thyroid storm due to Graves' disease was made with a score of 50 on the Burch-Wartofsky Point Scale. Patient was started on propylthiouracil (PTU), metoprolol tartrate and furosemide. Corticosteroid therapy was not utilized due to the risk of worsening heart failure from sodium retention. After two days of PTU, patient developed drug induced liver injury (DILI). Patient was then transitioned to methimazole which was also discontinued thereafter due to progression of DILI. She was started on potassium iodide to suppress thyroid hormone production, and N-Acetyl cysteine (NAC) for DILI. Due to persistent thyrotoxicosis, the patient underwent a total thyroidectomy on hospital day 13 and was started on levothyroxine for post-surgical hypothyroidism. Liver function returned to normal on day 16. A surveillance echocardiogram obtained 3 months later showed recovered biventricular function (ejection fraction 55-60%), mild mitral and tricuspid regurgitation and normal pulmonary pressures. Educational value: Thyroid storm is an uncommon, life-threatening emergency. Thioamides remain the cornerstone of treatment, but rarely, they can result in adverse events such as hepatotoxicity. When thioamides are contraindicated, other agents (e.g., iodine, lithium, potassium perchlorate, cholestyramine, glucocorticoids) can be utilized; however, refractory cases might require total thyroidectomy as definitive treatments. Our patient had marked recovery of cardiac function and resolution of pulmonary hypertension and DILI after timely surgical intervention. Multidisciplinary care is pivotal for the successful management of thyroid storm. Presentation: 6/3/2024

8278 Myxedema Coma With COVID-19 Infection

Abstract Disclosure: R. Kalas: None. V. Manohar: None. N.S. Bapat: None. Y. Kim: None. Myxedema coma is a rare medical emergency that occurs in severe hypothyroidism with a mortality rate reaching 60%. The hallmarks include depressed mental status and hypothermia, but hypotension, hyponatremia, bradycardia, and hypoventilation can also occur. This complication can occur in patients with a longstanding history of severe hypothyroidism especially when precipitated by an acute event such as infection, myocardial infarction, or surgery. Prompt recognition and treatment of the condition is important to prevent fatal outcomes. We present a case of myxedema coma in the setting of COVID-19 infection and nonadherence to medication. A 91-year-old woman with history of hypothyroidism, hypertension, coronary artery disease, cardiomyopathy, and chronic obstructive pulmonary disease presented with dyspnea. Her home medications included levothyroxine 75 mcg daily, which she reported not taking for at least 3 weeks. Vital signs revealed a normal heart rate and blood pressure of 84 beats/min and 137/78 mmHg, but the patient was hypothermic to 35.7 °C. On exam, she was somnolent. Labs were notable for markedly elevated TSH level of 111 (0.380 - 4.700 uIU/mL), low free T4 level of 0.57 (0.93 - 1.70 ng/dL), random cortisol level of 25.5 (6.2-18.4 ug/dL), and positive SARS-COV-2 PCR test. Chest x-ray revealed bilateral lower lung parenchymal consolidations and arterial blood gas revealed a pH of 7.29 and pCO2 of 52.5 (35-45 mm Hg), indicating hypercapnia. She was treated for COVID pneumonia and given concern for COVID induced myxedema coma she was given IV levothyroxine 100 mcg for 2 days, after which the patient’s hypothermia and somnolence improved, with an improved repeat free T4 level 0.87 ng/dL. She was subsequently transitioned to oral levothyroxine 112 mcg daily and discharged after improvement of her COVID symptoms. Myxedema coma is a rare complication of decompensated hypothyroidism, frequently triggered by an infection. Thyroid hormone plays an important role in regulating both the innate and adaptive immune systems. As a result, individuals with hypothyroidism are at increased risk of acquiring infections, which can further lead to myxedema coma. COVID-19 infection has been associated with thyrotoxicosis, thyroiditis and hypothyroidism. Given the associated high mortality rate, regardless of the underlying precipitant, prompt diagnosis and early administration of thyroid hormone replacement therapy is crucial to improve outcomes. Presentation: 6/3/2024