Related Experiment Videos
Predictability of SNR and reader preference in clinical MR imaging
1Hospital of the University of Pennsylvania, Philadelphia 19104.
Magnetic Resonance Imaging
|January 1, 1990
Summary
Researchers evaluated the agreement between predicted and measured signal-to-noise ratios (SNR) in MRI scans. An image quality index (IQI) incorporating SNR and voxel volume better predicted radiologist preference than SNR alone.
Area of Science:
- Magnetic Resonance Imaging (MRI)
- Medical Imaging Analysis
- Radiology
Background:
- Accurate prediction of signal-to-noise ratio (SNR) is crucial for optimizing MRI scan protocols.
- Understanding factors influencing radiologist preference is essential for improving image quality.
Purpose of the Study:
- To ascertain the agreement between predicted and measured SNR in MRI scans of the brain and knee.
- To determine quantitative criteria for reader preference in MRI images.
Main Methods:
- Fifty-four independent MRI scans were performed on two volunteers, varying parameters like number of excitations (NEX), field of view (FOV), section thickness (dz), and phase-encoding steps (Ny).
- A blinded study involved radiologists rating image desirability.
- Correlation analyses and plots of mean rating versus SNR, voxel volume, and image quality index (IQI) were conducted.
Main Results:
- Correlation coefficients between measured and predicted SNR were 0.82 and 0.86 for the brain and knee, respectively.
- Improved correlations were observed when NEX was held constant.
- The image quality index (IQI = SNR/voxel volume) was a better predictor of radiologist preference than SNR or spatial resolution alone.
Conclusions:
- T1-weighted MRI protocols with an SNR of approximately 20 are preferable.
- Excess SNR can be traded for reduced voxel size or shorter scan times to enhance image quality and reader preference.