Jul 2026· International Journal of Health Engineering and Technology· Vol 5· 0 citations· 9 references
TL;DR
The results showed that residual learning on ResNet50 effectively addressed the vanishing gradient problem in deep networks and achieved clinically relevant classification accuracy, supporting its potential integration into computer-aided diagnosis (CAD) systems.
Abstract
Pneumonia is one of the leading causes of death from infectious diseases worldwide, making rapid and accurate radiological diagnosis crucial for successful medical treatment. This study implemented and compared three deep learning architectures—a custom Convolutional Neural Network (CNN), ResNet50, and VGG16—for binary classification of chest X-ray images into Normal and Pneumonia categories. The Chest X-ray Pneumonia dataset from Kaggle (5,863 images) was used with an 80/10/10 (train/validation/test) data split and data augmentation to address class imbalance. ResNet50 with transfer learning from ImageNet weights achieved the best performance: 95.1% accuracy, 92.3% precision, 96.7% recall, 94.4% F1-score, and 97.5% AUC-ROC, outperforming the custom CNN (89.4% accuracy, 95.2% AUC) and VGG16 (93.7% accuracy, 96.1% AUC). Statistical analysis confirmed that the performance difference between ResNet50 and the custom CNN was statistically significant (p < 0.05). The results showed that residual learning on ResNet50 effectively addressed the vanishing gradient problem in deep networks and achieved clinically relevant classification accuracy, supporting its potential integration into computer-aided diagnosis (CAD) systems.
The rising incidence of pneumonia and tuberculosis presents major public health challenges, particularly in resource-limited regions such as Ethiopia, where shortages of skilled radiologists hinder timely diagnosis. Although deep learning has shown promise in medical image analysis, robust multiclass classification of these diseases from chest X-ray images remains technically challenging. This study proposes a deep learning–based multiclass classification framework for automated detection of pneumonia and tuberculosis from chest X-rays. A total of 2,100 chest X-ray images were collected from major hospitals in the Amhara region. Images were preprocessed and augmented to improve robustness. Multiple convolutional neural network architectures including Sequential CNN, VGG19, MobileNetV3-Large, EfficientNetV2-B0/B1, and InceptionResNet were trained and evaluated using accuracy, precision, recall, F1-score, and AUC metrics. A two-stage transfer learning strategy with systematic hyperparameter optimization was applied. Among the evaluated models, the customized VGG19 achieved the best performance, with 99% accuracy, precision, recall, and F1-score, and an AUC of 99.93%. Other models also demonstrated strong performance, with accuracies ranging from 95 to 98%. These results highlight the importance of model selection and optimization in medical image classification. An independent external validation using 3,568 chest X-ray images demonstrated robust generalization, with an accuracy of 88.83% and macro F1-score of 83.92%. Future work will focus on prospective multicenter validation, patient-level evaluation, and expert radiologist verification of the explainability framework.
Addisu Baye Flatie, Zegeye Regasa Wordofa, Abraham Keffale Mengistu et al.· Discover Artificial Intellig...· 0 citations
Background: Deep learning models, particularly convolutional neural networks (CNNs), have shown promising performance for pneumonia detection using chest X-ray images. However, the impact of preprocessing, architecture selection, data augmentation, and ensemble strategies has not been systematically evaluated. This study investigated how these factors affect model robustness and diagnostic performance. Methods: A public pediatric chest X-ray dataset was used to systematically evaluate pixel normalization methods, six CNN architectures, progressive data augmentation strategies for class imbalance, and both feature-level and decision-level ensemble approaches. Model performance was assessed by considering not only overall classification accuracy but also clinically relevant risk metrics, particularly false-negative rates. Results: Pixel normalization to the 0–1 range improved model convergence, while Xception and InceptionV3 achieved the best overall performance. Model-specific augmentation strategies were more effective than a fixed 1:1 class ratio for reducing false negatives. Feature-level ensembles tended to overfit, whereas decision-level ensembles provided more stable but only modest performance improvements. Conclusions: These findings demonstrate that reliable medical AI systems require systematic optimization of preprocessing techniques, model architecture, data augmentation strategies, and clinically meaningful evaluation metrics rather than maximizing a single performance indicator. The proposed framework provides practical guidelines for developing robust deep learning models for pneumonia diagnosis in clinical settings.
YongJun Kim, Ji-Yeoun Lee· BioMedInformatics· 0 citations
The proposed deep learning models provide an efficient and accurate tool for multiclass pneumonia detection from CXR images and have the potential to support healthcare professionals in making more accurate diagnoses.
Timothy Karani, Stephen Waithaka· Journal of the Kenya Nationa...· 0 citations
Pneumonia remains a leading cause of morbidity and mortality worldwide, and timely radiological screening is essential for effective treatment, particularly in settings where specialist radiologists are scarce. Convolutional neural networks (CNNs) classify chest X-ray images as normal or pneumonic with high accuracy, but their computational cost limits deployment on portable, low-power clinical devices. This paper presents an end-to-end pipeline that trains a ResNet-50 network, using transfer learning, to classify chest radiographs and deploys the trained model on a Xilinx ZCU104 Zynq UltraScale+ FPGA board through the Vitis AI toolchain. Images from a public pediatric pneumonia dataset (5216 training, 624 test, 16 validation images) are resized, normalized, and evaluated under several classical filtering operations before being passed to the network, which combines residual convolutional blocks, ReLU activations, and a sigmoid output layer. The model is trained for 30 epochs with the Adam optimizer and binary cross-entropy loss, reaching a validation accuracy in the 90–92% range. After quantization and compilation for the on-board Deep-Learning Processing Unit, the FPGA implementation achieves an inference throughput of 1250 frames per second at 4.1 W, compared with 44 frames per second at 19.9 W on a general-purpose CPU baseline — approximately a 28-fold gain in throughput and a 138-fold gain in energy efficiency. These results indicate that SoC-FPGA platforms are a practical route to real-time, energy-efficient pneumonia screening at the point of care.
V. M. Vinaya, R. P. Vishwanath, N. P. Nainika et al.· Indian Journal of Electronic...· 0 citations
Accurate lung disease diagnosis plays a critical role in medical treatment and improving patient outcomes. Conventional diagnostic methods often require experienced radiologists and are time-consuming. This study develops a lung disease diagnosis system based on Magnetic Resonance Imaging (MRI) data using the Convolutional Neural Network (CNN) method with Transfer Learning, specifically utilizing MobileNetV2 and DenseNet121 architectures. The dataset comprises 7,141 MRI images collected from Kaggle and RS Islam Aysha Bogor, classified into nine categories: Bacterial Pneumonia, Covid-19, Normal, Tuberculosis, Pneumothorax, Viral Infection, Asthma, Bronchitis, and Bronchopneumonia. Images were preprocessed to 224×224 pixels with pixel normalization to [0,1]. Four experimental scenarios were evaluated, varying optimizer, learning rate, batch size, and number of epochs. Results showed that MobileNetV2 achieved the best accuracy of 92.14%using RMSprop optimizer, learning rate 0.001, batch size 32, and 40 epochs. DenseNet121 achieved 87.82% accuracy with Adam optimizer under the same configuration. Validation using the confusion matrix yielded an overall accuracy of 91%, precision of 91.37%, recall of 92.25%, and F1-score of 91.79%. The best model was deployed as a web-based application built with Python Flask, enabling automatic image normalization and real-time classification without manual preprocessing. This research demonstrates that CNN-based Transfer Learning is effective for automated lung disease diagnosis with limited datasets
Universitas Negeri, D. Ramadhan, Eneng Tita Tosida et al.· Jurnal ilmu komputer dan apl...· 0 citations
Pneumonia screening from chest X-ray (CXR) images is clinically important but often suffers from limited data and class imbalance in public benchmarks. This work develops a transfer-learning-based binary classifier using ResNet50 pretrained on ImageNet. Images are resized to 224×224 and normalized to [0,1]. Online augmentation (rotation, zoom, and shifts) is applied during training. To mitigate imbalance, class weights computed from the training labels are used in the loss. We fine-tune the network by freezing the early layers and updating higher-level layers with a small learning rate (Adam, 1e- 4) and early stopping. On the held-out test set (624 images), the model achieves about 0.88 accuracy. For the pneumonia class, precision, recall, and F1-score reach approximately 0.93, 0.88, and 0.90, respectively. The results indicate that combining class-weighted training with fine-tuned ResNet50 provides a practical baseline for automated pneumonia screening on CXR images.
Danning Dong· International Conference on...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
