depending if your data are normal or not (shapiro.test), you can use an anova (parametric analysis of variance)  or its non-parametric equivalent, the Kurskal-Wallis test (if your data are not normal).

Not really an answer but a caution. Count data is rarely normal especially if you have mostly low numbers - (the values are bounded at zero).

Several points:

1. What is the research question? Are you trying to understand whether platelets from different individuals react differently to the treatment by obtaining different variances of some parameter?

2. If so consider using a regression model and look at the distribution of the residuals. For this by the way, I think,  you need a good sample size of at least 20 samples in each group.

3. Some statistical programs have tests which check are SDs same or not, sort of homogeneity analysis which is also presented in ANOVA tables (but I would not go there...).

4. I think you need to have some very serious reading done :)

ANOVA followed by Dunnett’s t-test and Bonferuoni will be help full

Thanks for your answers.. Hanna I think I should go for regression line . Yes I am looking for the variance in platelets response from different individuals.

You have to test the normality of your data using SPSS program or any other statistical software. Then depending on the decision you obtain from the tests, you can use parametric or non-parametric tests. The type of tests you use depend on the number of groups you have and also dont forget to do levene test for testing the homogeneity of your data.