Research Progress of Noise Removal Techniques for Oximeters Based on Signal Processing Algorithms

Blood oxygen saturation (SpO) is a crucial physiological index that can be affected by baseline drift, motion artifacts, and high-frequency noise, affecting clinical diagnosis accuracy. This paper focuses on noise reduction techniques for oximeters using signal processing algorithms, analyzing traditional methods like adaptive filtering and wavelet transform and modern deep learning models like CNN, LSTM, and GAN. The study reveals challenges such as insufficient generalization ability in complex noise environments, contradiction between real-time performance and computational efficiency, significant influence of individual differences, and scarcity of high-quality labeled data. Traditional methods offer advantages in specific noise environments but have limitations when dealing with complex and changeable scenes. Deep learning technology has made breakthroughs in noise reduction performance but is limited by model complexity and data requirements. Future interdisciplinary innovation and integration, combining biomedical engineering, artificial intelligence, and signal processing technologies, are expected to provide new solutions for oximeter noise removal and promote non-invasive oxygen monitoring technology.