WE-G-217A-08: Routine ACR SNR Measurement Failed to Detect 32-Channel Head Coil Receiver Malfunction

To study if malfunction of a receiver can be detected robustly using the simple ACR SNR measurement approach on a 32-channel head coil.

Standard ACR T1W images (11slice) were acquired with a commercial 32 channel head coil on a 3T Philips Achieva MR scanner following the ACR recommended setup. Raw data were saved and were used to reconstruct 32 image datasets, each with one coil channel turned off and signal were excluded from reconstruction. Routine simple SNR evaluation method was used to measure SNR for each dataset. Specifically, region of interest (ROI) analysis was performed on slice #7 for each dataset. Signal was the mean value of the pixel intensity measured using an ROI with area of 200 cm² positioned at the center of phantom. Noise was the standard deviation derived from an ROI positioned in the background in a corner of the image. SNR was then calculated from signal divided by noise. For comparison purposes, we empirically chose 5% SNR drop compared to the full 32 channel dataset SNR as a significant SNR drop that is correlated a potential coil channel defect.

Among the 32 image datasets reconstructed each with one receiver turned off, only 4 showed SNR drop of more than 5% or more compared to the reference SNR obtained from the original dataset. Four other datasets had SNR drop between 0.1-5%. The rest (24 image sets) did not show any SNR drop. Therefore, SNR monitoring based on the large ROI approach as the routine ACR QC procedure failed detect receiver malfunction in this coil.

More advanced and thorough coil evaluation methods, instead of the routine simple ACR SNR measurement method, have to be applied to evaluate the performance of the phased-array head coil with 32 or more channels.