Clemens Reisinger MD, Abstract Co-Author: Nothing to Disclose

Fides Schwartz, Abstract Co-Author: Nothing to Disclose

Markus Klarhofer, Abstract Co-Author: Employee, Siemens AG

David Grodzki, Abstract Co-Author: Employee, Siemens AG

Anna Hirschmann MD, Presenter: Nothing to Disclose

Prospectively assess acoustic noise levels and image quality while using a noise reduction algorithm on proton-density fat-supressed images in knee MR imaging.  

Institutional review board approval and informed consent were obtained. MRI examinations of the knee were obtained of 20 patients using a 3 Tesla whole body system (MAGNETOM Prisma, Siemens Healthcare, Germany) with a standard knee protocol, including a coronal proton-density fat-supressed sequence (A). Adapted versions of this sequence were added, once using a prototype noise reduction algorithm that smooths gradient pulses wherever possible while keeping imaging parameters constant (B) and a second time with minimal adjustments to acquisition bandwidth (increase of 10%) and echo spacing of the turbo spin echo train (increase of echo spacing by 10%) in addition to the noise reduction algorithm (C). Acoustic noise was assessed quantitatively with a sound level meter and qualitatively on a 10-point scale (0=silence, 10= painful noise). A questionnaire was completed by every patient, evaluating noise levels as well as patient comfort. Image quality was evaluated quantitatively by SNR and qualitatively by image impression. Significant differences between acoustic noise levels and SNR were sought using Wilcoxon signed-rank test. A p-value <0.05 was considered statistically significant. 

Acquisition time was similar for sequences A and B (2.53 min) and slightly longer for C (3.03 min). Noise levels for the correlating sequences were 87.8 dB (A), 82.2 dB (B) and 74.6 dB (C), respectively. Noise levels as percieved by the patients were significantly lower when comparing sequence B (mean 4.5) and A (mean 6)  and also between C (mean 3.3) and B. SNR showed no significant differences when comparing sequences A (mean: 73) and B (mean: 71). There was significant reduction of SNR though, when comparing sequence C (mean: 66) to sequence A or B.    

Significant acoustic noise reduction is possible using adjusted proton-density sequences on MR imaging of the knee joint without reduction of SNR.

Algorithm based adjustments of TSE sequences are a valid method to reduce acoustic noise while preserving image quality without need for hardware modification, thus increasing patient comfort.

Reisinger, C, Schwartz, F, Klarhofer, M, Grodzki, D, Hirschmann, A, Optimizing Sequence Design for Acoustic Noise Reduction in MRI of the Knee in 20 Patients.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14045544.html