Optimized Convolutional Neural Network Recognition for Athletes' Pneumonia Image Based on Attention Mechanism

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Sports Medicine

Background:

• High-intensity exercise compromises athletes' immune function, elevating pneumonia risk.
• Pulmonary infections can severely impact athlete health and career longevity.
• Current pneumonia diagnosis methods are inefficient due to reliance on expert interpretation and staff shortages.

Purpose of the Study:

• To develop an efficient and accurate method for early pneumonia diagnosis in athletes.
• To leverage artificial intelligence and image enhancement for improved diagnostic capabilities.
• To overcome the limitations of traditional diagnostic approaches in sports medicine.

Main Methods:

• Image enhancement techniques including contrast boost and edge enhancement.
• Application of inverse curvelet transformation for improved lung image quality.
• Utilizing an optimized convolutional neural network with an attention mechanism for image recognition.

Main Results:

• The proposed method achieved higher recognition accuracy for athlete lung images compared to DecisionTree and RandomForest.
• Image enhancement techniques effectively highlighted crucial edge information in lung images.
• The convolutional neural network with attention mechanism demonstrated superior performance in identifying pneumonia.

Conclusions:

• The developed AI-based method offers a promising solution for early and accurate pneumonia detection in athletes.
• Image enhancement combined with attention-based CNNs can significantly improve diagnostic efficiency.
• This approach can aid in timely recovery and potentially prevent early retirement for affected athletes.