Results
Below three \MRI\ images of the same tumour present in a rat are shown.
The images show the tumour before and at different times after being treated with
30mg/kg combretastatin-A4-phosphate (CA4P).
The complete details of the data acquisition of these three images is described in the
paper "Evaluation of the anti-vascular effects of combretastatin in rodent tumours by
dynamic contrast enhanced MRI".
The first image shows the rat tumour before treatment, the second shows the rat tumour one hour
after an injection of CA4P.
The final image shows the rat tumour six hours after
treatment. In each case, the images represent the sum of dynamic images obtained for the
first 90 seconds after injection of the contrast agent, Gd-DTPA.
|
Histogram Measures
The generated histogram measures of these three images are shown below.
|
The histogram of the first image clearly shows that a greater intensity range is present, and the original image has a greater contrast, compared to the last two images. The similarity between the last to images can be seen by looking at histograms. Few other inferences about the original data can be made from looking solely at the histograms.
Cooccurrence Measures
The co-occurrence measures clearly show how the range of intensities decrease
as the time after treatment increases, just as is shown by the histogram measures. This
is seen from looking at how the diagonal decreases for the images.
The two dissimilarity measures provided by the dissimilarity and contrast equations
also show how the dissimilarly or heterogeneity decreases as the time after treatment progresses.
The image before treatment shows significantly greater dissimilarly with a contrast
value of 1642 compared to 478 and 261 for the second and third images respectively.
|
Variogram Measures
Below is a plot of the three variograms of the rodent tumour at different times after treatment.
The Blue curve represents the variogram of the rat tumour just before the treatment and
correspond to to the first image, the violet curve shows the variogram data for image
an hour after treatment. The yellow curve represents the final image taken six hours
after treatment.
The maximum lag distance used for the images was 40, and is plotted along the x axis.
The dissimilarly is plotted along the y axis.
The curves show clearly how the dissimilarly is far greatest within the first image and the
rate of increase with the lag distance is the greatest. It also shows that
the dissimilarity of the first image in greater at all lag distances.
The final image shows the least dissimilarity overall, along with the slowest rate.
|
Conclusion
All the measures show how the heterogeneity was greatest for the initial image, that was taken before treatment. They also show that the heterogeneity continued to reduce for the image taken one hour after treatment and the image taken six hours after treatment. This is surprising as from looking solely at the original images, one would expect that the last shows more heterogeneity than the second, and so serves a good example of how the alternative measures, implemented by Hermes, can be useful in classifying images based on the heterogeneity. Although Hermes shows a clear change in the heterogeneity of the original image over time, a larger number of image sets need to be analysed before conclusions about the actual biological effects of the treatment can be inferred. In particular there tends to be a considerable amount of biological variation in studies of this sort.