This study enrolled 34 women with a mean age±standard deviation of 57±12 years (age range of 30 to 83 years).
37 breast lesions were considered since some women had multiple lesions.
From these 37 lesions: 4 were benign lesions,
namely Fibroadenomas (FA),
and 33 were malignant lesions,
including 27 Invasive Ductal Carcinomas (IDC) and 6 Ductal Carcinoma In Situ (DCIS) (Fig. 1,
Fig. 2 and Fig. 3).
Regarding the 27 IDC lesions: 5 IDC lesions were classified as on G1 grade,
14 IDC lesions on G2 and the other 8 IDC lesions on G3,
where G1 corresponds to more differentiated lesions and G3 to less differentiated lesions.
Informed consent was obtained for all patients.
Inclusion/exclusion criteria
- Lesions with edema or hemorrhage were excluded;
- MRI examination was done before breast biopsy or at least 7 days after biopsy,
to avoid edema and hemorrhage;
- Women who had undergone chemotherapy and/or radiotherapy treatments and previous breast surgery were excluded,
as treatments can change signal intensity and tissue organization.
Data was acquired on a 1.5T MRI scanner with a bilateral 4-channel breast coil.
Each patient was submitted to normal breast MRI examination protocol (T2-weighted sequence,
DWI sequence with 2 b-values (0 and 1000 s/mm2) with Apparent Diffusion Coefficient (ADC) map calculation,
and dynamic contrast-enhanced T1-weighted sequence (axial plane and post-processing image subtraction and reconstructions in axial and sagital planes).
An additional diffusion-weighted image acquisition was done before contrast administration,
which consisted of a single-shot echo-planar imaging sequence (SS-EPI) with 6 b-values (0,
50,
250,
500,
750,
1000 s/mm2) in 3 diffusion-sensitizing directions.
The technical parameters were as follows: TR/TE=12931/85 ms; FOV=340x340 mm2; Matrix=228x226; number of slices=50; thickness=3 mm; gap between slices=0.6 mm; bandwidth=1686.5 Hz; NEX=1.
Scan time was approximately 4 minutes.
- Image analysis and data processing
Lesions were identified in 2 different slices,
where they were best visualized,
and regions-of-interest (ROIs) were placed on each b-value image.
Lesion's signal intensity values were read for each b.
The apparent diffusion coefficient (ADC),
which combines both diffusion and perfusion effects,
was calculated by fitting the full data (b=0 to 1000 s/mm2) to the mono-exponential model [1-3]:
Sb = S0 exp(-bADC)
where Sb is the lesion's signal intensity for a particular b-value,
and S0 is the signal at b=0 s/mm2,
derived from the fit.
IVIM model: true diffusion (D),
pseudo-diffusion (D*) and perfusion fraction (PF) were calculated according to the method presented by Patel et al.
[2].
Initially,
data for high b-values (b=250 to 1000 s/mm2) was fitted to a mono-exponential model:
Sb=Sint exp(-bD),
with Sb given as above and Sint,
the signal at b=0 s/mm2 resulting from the fit [2].
D was also derived from the fit.
PF was calculated from PF=(S0-Sint)/S0 with S0 being the signal at b=0 s/mm2,
as given above.
Then D and PF values were used in the bi-exponential model with the full data (b=0 to 1000 s/mm2) :
Sb/S0 = (1-PF) exp(-bD) + PF exp (-bD*)
to derive the D* coefficient [1-4].
Mean values of these parameters were calculated and compared between benign and malignant lesions,
and also between FA,
IDC and DCIS lesion groups.
Parameters were also compared among different grades of IDC lesions.
In order to evaluate if there were significant differences between these groups of lesions,
non-parametric tests were applied (significance α=0.05).