Purpose: Diffusion-weighted imaging (DWI) may improve the diagnostic performance of conventional breast magnetic resonance imaging (MRI). Diffusion tensor imaging (DTI) is an extension of DWI. If DTI-derived measurements are to be clinically useful, particularly for predicting and/or monitoring therapeutic effects, they must be robust and reliable. The purpose of this study was to assess intra- and interobserver reproducibility of DTI-derived fractional anisotropy (FA) and apparent diffusion coefficient (ADC) at 3.0 T.
Materials and methods: This prospective study was approved by the Institutional Review Board, and participants provided written informed consent. Sixty normal contralateral breasts of 60 patients (28-85 years, median 57) were analysed with a DWI sequence following a standard MRI protocol. Four authors performed all postprocessing and analyses independently and in different sessions. The same authors, blinded to the initial results, repeated the image postprocessing and analysis 4 weeks after the initial session.
Results: Mean ADC and FA for DTI sequences were, respectively, 1.92±0.30 and 0.32±0.09. Intra- and interobserver agreement of the four radiologists for ADC and FA were good (acceptable). Kappa values for ADC were intra-R1=0.82; intra-R2=0.84; intra-R3=0.89; intra-R4=0.88; inter-R1-R2=0.73; inter-R1-R3=0.74; inter-R1-R4=0.81; inter-R2-R3=0.76; inter-R2-R4=0.77; inter-R3-R4=0.83. Kappa values for FA were intra-R1=0.60; intra-R2=0.72; intra-R3=0.84; intra-R4 = 0.66; inter-R1-R2=0.64; inter-R1-R3=0.69; inter-R1-R4=0.72; inter-R2-R3=0.80; inter-R2-R4=0.71; inter-R3-R4=0.73. Within-subject coefficient of variation was 15% for ADC and 30% for FA. Repeatability with α=0.05 was 0.37×10-3 mm(2)/s for ADC and 0.112 for FA.
Conclusions: ADC and FA measurements obtained with DTI are reproducible and may be valid, reliable and sensitive to change. ADC values obtained with DTI are more reproducible than FA.