Login / Signup

Automatic Adaptive Gain for Magnetic Resonance Sensitivity Enhancement.

Mazin JoudaErwin FuhrerPedro SilvaJan G KorvinkNeil MacKinnon
Published in: Analytical chemistry (2019)
The decaying nature of magnetic resonance (MR) signals results in a decreasing signal-to-quantization noise ratio (SQNR) over the acquisition time. Here we describe a method to enhance the SQNR, and thus the overall signal-to-noise ratio (SNR), by dynamically adapting the gain of the receiver before analog-to-digital conversion (ADC). This is in contrast to a standard experiment in which the gain is fixed for a single data acquisition and is thus adjusted only for the first points of the signal. The gain adjustment in our method is done automatically in a closed loop fashion by using the envelope of the MR signal as the control signal. Moreover, the method incorporates a robust mechanism that runs along with signal acquisition to monitor the gain modulation, enabling precise recovery of the signals. The automatic adaptive gain (AGAIN) method requires minimal additional hardware and is thus general and can be implemented in the signal path of any commercial spectrometer system. We demonstrate an SNR enhancement factor of 2.64 when applied to a custom spectrometer, while a factor of 1.4 was observed when applied to a commercial spectrometer.
Keyphrases
  • magnetic resonance
  • contrast enhanced
  • high resolution
  • machine learning
  • magnetic resonance imaging
  • deep learning
  • diffusion weighted
  • electronic health record
  • artificial intelligence