Comparison of the Diagnostic Performance of Synthetic Versus Acquired High b-Value (1500 s/mm2 ) Diffusion-Weighted MRI in Women With Breast Cancers

Abstract

Background: Acquired high b-value (>1000 s/mm² ) diffusion-weighted imaging (DWI) has its strength in lesion detection. However, it is not easily used, due to a lower signal-to-noise ratio, eddy current distortions, and prolonged acquisition times. Synthetic DWI does not have these disadvantages because it is based on indirect acquisition, calculated in a voxel-wise manner.

Purpose: To compare the diagnostic performance of synthetic and acquired high b-value (1500 s/mm² ) DWI in women with breast cancer.

Study type: Retrospective.

Population: A total of 108 patients (median age 49 years [range 32-77]) with 133 breast cancers.

Field strength/sequence: 3T, echo-planar imaging.

Assessment: Three radiologists independently reviewed image sets of both synthetic (S-b1500) and acquired (A-b1500) high b-value DWI. Malignancy confidence of the lesion was scored using a 6-point Likert-type scale.

Statistical test: Jack-knife alternative free-response receiver-operating characteristic 1 (JAFROC1) analysis was used. Sensitivity and positive predictive value (PPV) were compared using generalized estimating equations. An independent t-test was used to compare the confidence. An intraclass correlation coefficient was calculated to compare interobserver agreement.

Results: The JAFROC1 figures of merit values were 0.816 and 0.808 in S-b1500 and A-b1500, respectively, with no statistically significant difference (P = 0.637). Sensitivity was higher in synthetic than in A-b1500 for readers 2 (P = 0.015) and 3 (P = 0.037). Although sensitivity was higher in S-b1500 than in A-b1500 for reader 1, the difference was not significant (P = 0.487). The PPV of S-b1500 was not significantly different from that of A-b1500 (P = 0.397). The malignancy confidence of true-positive tumors was higher in S-b1500 than in A-b1500 (P = 0.013). Interobserver agreement was good for both sequences.

Data conclusions: The synthetic high b-value DWI may improve the diagnostic sensitivity for breast cancer detection without affecting PPV compared with acquired high b-value DWI.

Level of evidence: 3 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:857-863.

Keywords: breast neoplasms; computer-assisted; diffusion magnetic resonance imaging; image interpretation; magnetic resonance imaging; predictive value of tests.