Consistency of MR Cytometry Mappings across Different Acquisition Protocols in Prostate Imaging.

Rationale

AND

Objectives

To evaluate the consistency of MR cytometry mappings across protocols, and quantify the variations in prostate imaging between oscillating gradient spin-echo (OGSE) frequencies of 17/33Hz and 25/50Hz.

Materials And Methods

This prospective study enrolled 40 patients requiring prostate MRI examinations, who underwent 3T-MRI scans including pulsed gradient spin-echo (PGSE) and OGSE sequences. Two acquisition protocols with different OGSE sequences were employed: 17/33Hz and 25/50Hz. Regions of interest (ROIs) encompassing entire prostates were manually delineated and further subdivided into peripheral (PZ) and transitional zones (TZ) by a radiologist. Microstructural parameters were fitted using the IMPULSED model, and time-dependent apparent diffusion coefficients (ADCs) were calculated. Common first-order and radiomic features were extracted from their mappings. Wilcoxon signed-rank test was performed to evaluate paired differences (17/33Hz vs. 25/50Hz); relative variation coefficients were calculated for each patient; Pearson correlation analysis quantified correlations between two sets of results. All statistical analyses were performed separately for entire prostate, PZ, and TZ.

Results

The results for the entire prostate, PZ, and TZ were similar: Although several first-order features (such as mean and median) of imaging parameters showed strong correlations (r>0.7), most features exhibited inconsistencies between the two protocols, including significant paired differences, large relative variations and low correlations, particularly for cell diameter mappings. Over 30% of radiomic features exhibited correlations of r<0.5 across different protocols.

Conclusion

MR cytometry imaging, including IMPULSED model fitting and time-dependent ADC measurements, depends on the specific acquisition protocol; in particular, cell diameter mapping is highly dependent on OGSE frequencies.