Abstract:Mycoplasma pneumoniae pneumonia (MPP) poses significant diagnostic challenges in pediatric healthcare, especially in regions like China where it's prevalent. We introduce PneumoniaAPP, a mobile application leveraging deep learning techniques for rapid MPP detection. Our approach capitalizes on convolutional neural networks (CNNs) trained on a comprehensive dataset comprising 3345 chest X-ray (CXR) images, which includes 833 CXR images revealing MPP and additionally augmented with samples from a public dataset. The CNN model achieved an accuracy of 88.20% and an AUROC of 0.9218 across all classes, with a specific accuracy of 97.64% for the mycoplasma class, as demonstrated on the testing dataset. Furthermore, we integrated explainability techniques into PneumoniaAPP to aid respiratory physicians in lung opacity localization. Our contribution extends beyond existing research by targeting pediatric MPP, emphasizing the age group of 0-12 years, and prioritizing deployment on mobile devices. This work signifies a significant advancement in pediatric pneumonia diagnosis, offering a reliable and accessible tool to alleviate diagnostic burdens in healthcare settings.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper aims to address the diagnostic challenges of Mycoplasma pneumoniae pneumonia (MPP) in pediatric healthcare, especially in high-incidence areas like China. Specifically, the paper proposes a mobile application named PneumoniaAPP, which utilizes deep learning technology, particularly convolutional neural networks (CNN), to achieve rapid and efficient MPP detection. ### Background and Challenges 1. **Clinical Diagnosis Difficulties**: - MPP is a community-acquired infection that primarily affects children and young people, with clinical manifestations including fever and cough. - Traditional diagnostic methods rely on laboratory microbiology, serological tests, and imaging examinations, but these methods have drawbacks such as low sensitivity and specificity, and long detection times. 2. **Imaging Interpretation Challenges**: - Chest X-ray (CXR) images are crucial for assessing clinical severity and predicting prognosis, but inexperienced doctors may misinterpret these images, leading to misdiagnosis. - Interpreting CXR images is time-consuming and labor-intensive. 3. **Limitations of Existing Research**: - Existing AI-assisted diagnostic methods mostly focus on adult pneumonia or different types of pathogens, lacking research specifically targeting pediatric MPP. - Most studies use complex deep learning models with a large number of parameters, making efficient deployment on mobile devices difficult. ### Solution 1. **Dataset Construction**: - Collected CXR images of children aged 0-12 from multiple medical institutions, including 833 MPP images and 169 normal images, and combined with a public pediatric pneumonia dataset, resulting in a dataset of 3345 images. 2. **Model Training**: - Trained models using lightweight CNN architectures (such as ResNet-18, EfficientNet-B0, RegNetX-400mf, SwinV2-Tiny, and ConvNeXt-Tiny) to reduce the number of parameters and computational burden. - Utilized transfer learning techniques, initializing models with weights pre-trained on ImageNet-1K to improve model generalization and training efficiency. 3. **Interpretability Techniques**: - Integrated interpretability techniques (such as Class Activation Map, CAM) to help pulmonologists accurately locate lung lesion areas, enhancing the model's interpretability and credibility. 4. **Mobile Application Development**: - Developed an Android application named PneumoniaAPP, allowing users to obtain rapid and accurate MPP detection results by uploading CXR images. - The application has a user-friendly interface and is easy to operate, suitable for use in medical environments. ### Main Contributions 1. **Improved Accuracy and Efficiency of MPP Diagnosis**: - The model achieved an accuracy of 88.20% and an AUROC of 0.9218 on the test set, with an accuracy of 97.64% specifically for the MPP category. 2. **Enhanced Model Interpretability**: - Provided visualizations of key feature areas in images through CAM technology, helping doctors better understand the model's decision-making process. 3. **Efficient Deployment on Mobile Devices**: - Selected lightweight CNN architectures to ensure the model can run efficiently on mobile devices, facilitating practical application in medical scenarios. ### Conclusion PneumoniaAPP significantly improves the diagnostic efficiency and accuracy of pediatric MPP by integrating deep learning technology and interpretability methods, providing a reliable and easy-to-use tool for pediatric healthcare.

Pneumonia App: a mobile application for efficient pediatric pneumonia diagnosis using explainable convolutional neural networks (CNN)

An Improved Algorithm for Detecting Pneumonia Based on YOLOv3

Based on improved deep convolutional neural network model pneumonia image classification

Ensembles of Convolutional Neural Networks models for pediatric pneumonia diagnosis

The Use of Chest Radiographs and Machine Learning Model for the Rapid Detection of Pneumonitis in Pediatric

An Explainable Contrastive-based Dilated Convolutional Network with Transformer for Pediatric Pneumonia Detection

A transfer learning method with deep residual network for pediatric pneumonia diagnosis

An Efficient Deep Learning Approach to Pneumonia Classification in Healthcare

Identification and classification of pneumonia disease using a deep learning-based intelligent computational framework

Artificial Intelligence Enhanced Rapid and Efficient Diagnosis of Mycoplasma Pneumoniae Pneumonia in Children Patients

Detection of Pneumonia Infection by Using Deep Learning on a Mobile Platform

Detection of pneumonia in chest X‐ray image based on separable convolutional neural network

An Improved Weighted Cross-Entropy-Based Convolutional Neural Network for Auxiliary Diagnosis of Pneumonia

Convolutional Neural Networks (CNNs) for Pneumonia Classification on Pediatric Chest Radiographs

Identification of pneumonia on chest x-ray images through machine learning

Ultrasound Image Intelligent Diagnosis in Community-Acquired Pneumonia of Children Using Convolutional Neural Network-Based Transfer Learning

Omni-dimensional dynamic convolution feature coordinate attention network for pneumonia classification

A machine learning approach on chest X-rays for pediatric pneumonia detection

A new hybrid approach for pneumonia detection using chest X-rays based on ACNN-LSTM and attention mechanism

A Medical Multimodal Large Language Model for Pediatric Pneumonia

Pneumonia Detection Proposing a Hybrid Deep Convolutional Neural Network Based on Two Parallel Visual Geometry Group Architectures and Machine Learning Classifiers