Ensemble of Deep Capsule Neural Networks: An Application to Pneumonia Prediction

Pneumonia is the primary cause of death in children under the age of 5 years. Faster and more accurate laboratory testing aids in the prescription of appropriate treatment for children suspected of having pneumonia, lowering mortality. In this work, we implement a deep neural network model to efficiently evaluate pediatric pneumonia from chest radio graph images. Our network uses a combination of convolutional and capsule layers to capture abstract details as well as low level hidden features from the the radio graphic images, allowing the model to generate more generic predictions. Furthermore, we combine several capsule networks by stacking them together and connected them with dense layers. The joint model is trained as a single model using joint loss and the weights of the capsule layers are updated using the dynamic routing algorithm. The proposed model is evaluated using benchmark pneumonia dataset\cite{kermany2018identifying}, and the outcomes of our experimental studies indicate that the capsules employed in the network enhance the learning of disease level features that are essential in diagnosing pneumonia. According to our comparison studies, the proposed model with Convolution base from InceptionV3 attached with Capsule layers at the end surpasses several existing models by achieving an accuracy of 94.84\%. The proposed model is superior in terms of various performance measures such as accuracy and recall, and is well suited to real-time pediatric pneumonia diagnosis, substituting manual chest radiography examination.

PneumoniaNet: Automated Detection and Classification of Pediatric Pneumonia Using Chest X-ray Images and CNN Approach

Pneumonia is an inflammation of the lung parenchyma that is caused by a variety of infectious microorganisms and non-infective agents. All age groups can be affected; however, in most cases, fragile groups are more susceptible than others. Radiological images such as Chest X-ray (CXR) images provide early detection and prompt action, where typical CXR for such a disease is characterized by radiopaque appearance or seemingly solid segment at the affected parts of the lung due to inflammatory exudate formation replacing the air in the alveoli. The early and accurate detection of pneumonia is crucial to avoid fatal ramifications, particularly in children and seniors. In this paper, we propose a novel 50 layers Convolutional Neural Network (CNN)-based architecture that outperforms the state-of-the-art models. The suggested framework is trained using 5852 CXR images and statistically tested using five-fold cross-validation. The model can distinguish between three classes: viz viral, bacterial, and normal; with 99.7% ± 0.2 accuracy, 99.74% ± 0.1 sensitivity, and 0.9812 Area Under the Curve (AUC). The results are promising, and the new architecture can be used to recognize pneumonia early with cost-effectiveness and high accuracy, especially in remote areas that lack proper access to expert radiologists, and therefore, reduces pneumonia-caused mortality rates.

732. Sensitivity and Specificity of Point of Care Lung Ultrasound vs. Chest X-Ray for the Diagnosis of Pediatric Pneumonia in Limited resource settings: The Zambia Experience

Background Pediatric pneumonia is the leading cause of child mortality in low-income countries. Pneumonia diagnosis is a challenge. Chest x-ray (CXR) is considered the gold standard, but it exposes children to ionizing radiation, and access to CXR is limited to hospital settings. Lung Point of Care Ultrasound (POCUS) is a portable and non-radiating alternative to CXR. Methods We enrolled 200 children aged 1-59 months from the University Teaching Hospital (UTH) Emergency Department (ED) in Lusaka, Zambia who met the WHO (World Health Organization) case definition for severe pneumonia. From each child, we collected demographic and clinical data, a CXR, and a set of ultrasound images using a Butterfly ultrasound probe. Images were independently interpreted by two radiologists blinded to the results of the other imaging modality. Using CXR as the gold standard, we determined the sensitivity and specificity, positive and negative predictive values, and likelihood ratios for pneumonia using lung POCUS. Results This preliminary analysis included 50 children seen between May-October 2020. Median age (9 months) (Range 4-15). 58% were male, (29/50). Median temperature was 37.3⁰C (range 36.5-38.0); median respiratory and pulse rates were 41 breaths/min (range 31-50) and 139 beats/min (range 124-160) respectively; median SpO2 on RA was 91% (range 89-95). 50% of cases had difficulty breathing (82%, 41/50); chest retractions (70%, 35/50) and grunting (62%, 31/50). Ultrasound images for 49/50 (98%) cases and CXRs for 50/50 (100%) of cases we analyzed. Sensitivity of lung POCUS in the detection of CAP was 61% (95% Cl: 0.52-0.84). The specificity was 77% (95% Cl: 0.56-0.91). Positive predictive value (PPV) 70% (95% CI: 0.62-0.94) and negative predictive value (NPV) 69% (95% CI: 0.56-0.79). Conclusion Preliminary findings of this study demonstrated the lower diagnostic accuracy of lung POCUS versus CXR in the detection of pneumonia in children 1- 59 months. The high specificity of the test will aid in ruling out severe pneumonia in children. Due to its availability, ease of interpretation, and absence of radiation exposure, lung POCUS should still be considered as an important initial imaging tool for the diagnosis of CAP in children in limited-resource settings. Disclosures All Authors: No reported disclosures

Drug utilization study on pediatric pneumonia patients in a tertiary care teaching hospital

Background: Pneumonia remains the prime infectious disease that increases the mortality rate among children under five claimed the lives of nearly 1.5 million children in 2015. Mortality due to childhood pneumonia is linked to the prevalence and relapse increases and usage of antibiotics is more. In such circumstances, the main. Aim: The Aim of the study was to assess the utilization of drugs and to spot the factors that contribute to the pediatric pneumonia patients at tertiary care teaching hospitals. Material and methods: A prospective, observational study was conducted from May 2021 to October 2021Overall 310 prescriptions were collected, 204 patients were included in the study based on inclusion criteria and data was collected from a proforma and by using WHO prescribing indicators study is analyzed. Results: Age group with pneumonia of<5years was 171 patients, followed by 33 patients between 6-8years.The male children constituted the major portion i.e. 120(58.82%) followed by female children were 84(41.18%). Distribution of drugs per prescription, majority of prescriptions with<5 drugs i.e., 110(56%) followed by 82 prescriptions with 5-10 drugs constituted (43%), this indicates the polypharmacy and 1 prescription with >10 drugs i.e., (1%). The distribution of antibiotics where majorly preferred drug in penicillin’s were amoxiclav- 164(59.20%), least preferred was piperacillin/tazobactam-9(3.28%), followed by cefotaxime-38(13.71%), ceftriaxone -17(6.13%), amikacin- 31(11.19%), and azithromycin- 18(6.49%). Conclusion: In our study, we observed an irrational use of antibiotics and polypharmacy of drugs in the treatment of pediatric pneumonia.