Analysis of CT Tube Degradation Based on Internet of Medical Things (IoMT) Data

This study investigates the degradation patterns of CT X-ray tubes using real-time operational data from three CT equipment at West China Hospital, collected via an Internet of Medical Things (IoMT) edge computing framework. The research aims to advance predictive maintenance (PdM) strategies by analyzing multidimensional sensor data, including filament current, oil temperature, and radiation dose. A dataset of 1.2 petabytes (April 2019-June 2024) was preprocessed using interquartile range (IQR) outlier removal, linear interpolation, and Savitzky-Golay (SG) smoothing to address noise and nonuniform sampling. Exploratory analysis identified filament current as the primary degradation indicator, revealing three distinct failure trajectories: linear decline (CT011), accelerated latestage degradation (CT008), and stochastic fluctuations (CT012). Dominant operational modes, particularly the SEQ protocol (120 kV/140 Hz), were linked to faster degradation compared to lowdose modes. SG smoothing with quadratic fitting outperformed traditional filters, preserving transient anomalies (e.g., arcing spikes) with amplitude error while reducing noise. Standardized degradation trends enabled machine-specific PdM models, reducing prediction errors versus global approaches. These findings underscore the value of IoMT-driven, usage-aware maintenance strategies to optimize tube lifespan and healthcare resource allocation.