Between January and February 2014,
conventional US and elastography were performed in 22 women who had been scheduled to undergo a ultrasound guided core biopsy or FNAC.
A total of 24 lesions were evaluated with both elastographic and B-mode US,
using a 5–13 MHz linear-array transducer.
All the examinations were performed by two radiologists with different experience in breast US and elastography (first: operator with low experience; second: skilled operator).
Two radiologists performed sequentially US and elastography acquisition; each operator was blinded to results of elastographic data acquisition and images interpretation obtained by the other one.
US lesions features were first evaluated (i.e.
echo pattern,
boundary,
margin,
shape,
micro-calcifications,
size,
posterior acoustic features) and then elastographic information were sequentially obtained in a real time approach.
Operators made five-eight compressions,
keeping a constant pressure of probe with an orthogonal angle respecting to skin surface (free-hand elastography).
In order to get a correct elastogram,
we included fat tissue,
used as reference.
Elastographic images were displayed with color map from red (soft) to green (intermediate) to blue (hard).
According to five-point scale classification [4],
a color-map pattern was assigned: score 1 if lesion was evenly shaded in green; score 2 when lesion with a mosaic pattern of green and blue is visualized; score 3 if lesion with strain at the periphery (green) and sparing of the core (blue); score 4 when strain at the periphery (blue) and no strain in the lesion; score 5 if no strain in the lesion or in the surrounding area (both in blue).
Figure 1
To obtain strain ratio,
two circular Regions Of Interest (ROIs) were manually superimposed on the elastogram.
The first ROI (target) was placed to be bounded by the inner margin of nodule.
The second ROI (reference) was placed in the fat tissue.
Considering the same series of compression,
target and reference ROIs were placed semi-automatically using ASR with a short acquisition time.
Target ROI was automatically placed into the lesion and reference ROI in the fat tissue encoded red/green because it represented soft/intermediate tissue stiffness in the chosen area.
According to our experience and to literature,
in the current study strain ratio cut-off used was 3.2.
[12] Assist Strain Ratio data were then compared with those obtained placing ROIs manually.
Figure 2
Acquisition time of Assist and manual strain ratio were both evaluated.
In the final data analysis histological results from core biopsy or FNAC were used as reference standard.
Statistical analysis
Continuous data are presented as mean ± standard deviation and as median with minimal and maximal range.
Categorical data are reported as counts and percentages.
Sensitivity,
specificity,
positive predictive value (PPV),
negative predictive value (NPV) and their confidence intervals were calculated using Wilson statistical test.
The predictive role of different elastographic generated variables was studied through logistic regression analysis.
Statistically significant differences were indicated by p < 0.05.
All computations were carried out with STATA software V.
12.