ADC mapping with 12 b values: an improved technique for image quality in diffusion prostate MRI

Purpose To compare diffusion images and coefficients obtained with 4 b-value versus 12 b-value apparent diffusion coefficient (ADC) mapping for characterization of prostate lesions and how these coefficients relate and compare to the PI-RADS™ classification and Gleason grading system. Methods Patients with indications for prostate cancer testing (n=158) underwent multiparametric 3T magnetic resonance imaging (MRI). Two diffusion sequences were acquired, one with 4 b values and one with 12 b values. ADC maps were calculated for each (ADC4 and ADC12) and the respective coefficients were tested for correlation with PI-RADS™ classification and Gleason score. Results The ADC12 sequence produced images of superior quality and sharpness than ADC4. Normal-area means (ADC4, 1793.3×10−6mm2/s; ADC12, 1100×10−6mm2/s) were significantly lower than those of lesion areas (ADC4, 1105.9×10−6mm2/s; ADC12, 689.4×10−6mm2/s) (p<0.001). Both techniques behaved similarly and correlated well with PI-RADS™ classification, distinguishing scores 3, 4, and 5 and with means tending to decline with increasing Gleason grade. ADC12 mapping yielded higher specificity than ADC4 (82.6% vs. 72.3%). Conclusions Diffusion with 12 values is a viable technique for examination of the prostate. It produced higher-quality images than current techniques and correlates well with PI-RADS™ classification and Gleason score.

4 89 and restriction of a given molecule, is expressed in square millimeters per unit time in 90 seconds. It is a reproducible measure of diffusion that can be obtained at any dedicated 91 workstation by measuring a delimited region of interest (ROI).

92
With the advent of the second version of PI-RADS™, an imaging classification 93 that aims to stratify prostate imaging findings according to severity and risk [10,11] for 94 tumors of the peripheral zone, DWI has become the key sequence for approximating 95 severity of a possible focal lesion.

96
In addition, the correlation between diffusion, as expressed as a numerical ADC 97 value, and the aggressiveness of prostate tumors has also been widely studied in the 98 literature. Significant diffusion restriction is associated with high histological 99 aggressiveness, as measured by Gleason scores. Therefore, ADC correlates closely 100 with prognosis and treatment planning for these patients [12][13][14].

156
The MRI sequences are summarized in Table 1      Likewise, the scores assigned to image quality were tabulated, analyzed, and 212 compared between the two sequences also between the two observers.

215
According to the inclusion criteria, 51 patients presented with suspicious lesions 216 that were measurable by the methodology used in the study design. Normal areas were 217 measured both in patients with lesions and in patients without lesions, for a total of 158.

218
Means, medians, standard deviations and ranges are summarized in Table 2.    (Fig 2). A similar relationship was also observed for values obtained from 232 ADC 12 , with significantly higher means in normal versus lesion areas, thereby 233 demonstrating similar behavior in the two techniques (Fig 3).

241
Given this similarity in behavior between the two techniques, measurements were 242 analyzed through a regression model between normal areas and lesion areas (Fig 4),

267
In ADC 12 measurements, similar correlations were observed between groups 3 and 4 268 and between groups 3 and 5.

269
Of the 158 patients included, 52 underwent prostate biopsy with the following

273
The correlation between Gleason score and ADC 4 and ADC 12 values was 274 calculated and presented in Table 3. Gleason scores were pooled to facilitate analysis: 275 score 7 included both the results 3 + 4 and 4 + 3, while score 9 included both the results 276 4 + 5 and 5 + 4.

296
The analysis of interobserver agreement related to the classification of image 297 quality and sharpness were made by the kappa coefficient and Spearman correlations, 298 that revealed low but significant agreement across all parameters, except for correlation 299 of the ADC 12 anatomy classification, which did not demonstrate agreement that was 300 significantly different from zero. In short, the two observers tended to make similar 301 classifications.

302
On comparative analysis between ADC 4 and ADC 12 in relation to anatomy and 303 lesion identification, we obtained significantly higher mean classification values for both 304 observers with ADC 12 than with ADC 4 (p < 0.001) (Figure 7), demonstrating that the new 305 technique provides a higher degree of sharpness than the standard sequence, both to 306 study the anatomy of the prostate and to identify suspicious lesions (Figure 8).

334
In this study, the main objective was to improve the technical parameters and 335 image quality of diffusion sequencing while maintaining its high diagnostic value.

336
As the primary result, the conspicuity and sharpness of images obtained by

353
On analysis of predictive value, both sequences proved to be significant 354 predictors of cancer, with ADC 12 having a higher specificity than ADC 4 , demonstrating 355 that it is the technique best able to rule out the possibility of cancer.

356
Although a statistically significant correlation between ADC 4 or ADC 12 values and

357
Gleason score obtained through biopsy was not found, probably due to the small number