A Method for Detecting Gradient Coil Cooling Faults for Magnetic Resonance Imaging

This paper proposes a method for monitoring and diagnosing cooling system faults in the gradient coils of magnetic resonance imaging (MRI). In fact, cooling system fault can potentially lead to significant imaging deviations. Hence, a Hash Cerebellar Model Articulation Controller (LCMAC) is devised to learn the normal characteristics of gradient coils. When cooling failure affects the gradient coils' performance, the LCMAC response deviates from the norm. Additionally, two characteristic variables, degradation index, and waveform index, are introduced for residual error analysis. Notably, this approach offers the advantage of not requiring the dismantling of the MRI equipment's outer casing or the addition of extra sensors. Moreover, an alternative strategy is formulated to identify degradation types and estimate their severity. Testing a typical gradient coil under five fault conditions in the air conditioning unit validates the effectiveness and robustness of this strategy.