Calibration of Phantom Elasticity
Fig.12 shows the SWE mode overview of background phantom with 4,
8,
12,
16,
20 and 24 g of gelatine added.
The colour scale on the top right depicts quantitative values of elasticity shown in the ROI while the qualitative values of elasticity measured in each ROIs were displayed on the right.
The values of the elasticity were shown in three ways: mean,
min and max.
The mean elasticity values were taken into account as it is the most relevant values.
Fig.13 shows the relationship between mass of gelatine added into the phantoms and the mean elasticity value the phantoms give.
It can be observed that the mean elasticity increases linearly with the amount of gelatine added.
The coefficient of determination,
R2 was 0.983,
which means that 98.3% of the total variation in mean elasticity can be explained by the linear relationship between mean elasticity and mass of gelatine added.
This showed a rather strong correlation between the mean elasticity of the phantom and the mass of gelatine added,
indicating that mass of gelatine added was useful as a predictor for the phantom’s mean elasticity from the calibration graph.
Comparison between SWE and the gold standard
Fig.14 shows the relationship between mass of gelatine added into spherical inclusions and the mean elasticity measured using SWE and Instron microtester.
It can be observed that the mean elasticity increases linearly with mass of gelatine added for both methods of measurements.
However,
the elasticity measurements by Instron microtester were always lower than the elasticity measurements by SWE.
The p-value was smaller than 0.05.
Therefore,
it can be concluded that there was a statistically significant difference between both method.
As the mean elasticity measured by SWEwas always greater than the measurements made by Instron microtester,
it can be said that the elasticity measurement made by SWEwas an overestimation of the actual elasticity by 7 to 39 kPa. This overestimation was probably contributed by the assumption made by the SWE system that all density of tissue in the body equals to 1 g cm-3.
In reality,
the density of human tissue differs slightly from one another,
depending on the types of tissue [Farvid,
2005; Ward,
2005].
Investigation of Factors Affecting the Elasticity Measurements
Fig.15 shows how different sizes of inclusions affect the SWE measurements.
It can be observed from the Fig.15 that the elasticity of different sizes inclusions did not vary much,
with standard deviations of 6.012 among the measured elasticity. However the error bars of each measurement appeared to be quite large but they all overlap with one another.
This indicated that there was a much lower likelihood that the elasticity values differ significantly from one another.
Fig.16 shows how inclusions at different depth affect the SWE measurements.
Only three inclusions closest to the surface of the phantom (2.7,
4.7 and 6.7 cm from the surface) could be measured probably because the transducer (15 MHz) used was designed for superficial measurement of up to 7 cm depth.
Due to lack of data,
the effect of depth on elasticity was inconclusive.
Therefore,
repetition of measurement is needed to investigate this effect.
Fig.17 shows that the elasticity increases with increasing mass of gelatine added to the inclusions.
Different amount of gelatine was added to the inclusions to produce inclusions with different stiffness.
The more gelatine was added into the inclusions,
the stiffer the inclusions would be.
Fig.18 shows the relationship between mass of gelatine added and mean elasticity measured using SWE for cylindrical and spherical inclusions.
The graph shows a linear relationship between the mass of gelatine and mean elasticity for both shapes of inclusions.
It can be observed that the mean elasticity of cylindrical inclusions was always lower than the mean elasticity of the spherical inclusions for the same mass of gelatine added.
However,
as the mass of gelatine added increases,
both graphs converge and finally meet at 125 kPa at 24 g of gelatine added.
The p-value was greater than 0.05.
Therefore,
it was concluded that there is no statistically significant difference between the mean elasticity of the cylindrical and spherical inclusions.
Further works need to be done to find the factors of the overestimation.