I'm planning a scientific study where I would investigate the correlation between a non-invasive imaging test with information taken from a CT scan.

My idea is to develop a study protocot, then to subject people who undergo CT scan to this non-invasive test (which is already vastly in use but not for this purpouse). Then the plan is to compare values from both imaging studies and check for Pearson's correlation coefficient.

However this seems to me as a weak plan (I'm new to scientific research, and surrounded by people who has poor interest in that). My questions are:

  • Would this be enough to prove that the non invasive method is at least comparable to the CT scan?
  • I plan to enroll people randomly from the daily schedule. Do I need to plan a randomization of the population taken into exam?
  • 1
    $\begingroup$ Got first -1 in less than a second after I posted, can I ask why? $\endgroup$ – KingBOB Apr 26 '18 at 14:20
  • $\begingroup$ A CT scan yields a set of images. What sets of numbers would you correlate? $\endgroup$ – rolando2 Apr 26 '18 at 14:25
  • $\begingroup$ It's unclear how your question can be answered. $\endgroup$ – SmallChess Apr 26 '18 at 14:25
  • 1
    $\begingroup$ Not sure people on this site will be able to answer, but you will want to look into validity and validation, including concurrent validity, representativeness. and so on. I'm guessing your method of validation will work better for some sub-populations than others. $\endgroup$ – rolando2 Apr 26 '18 at 17:25
  • 1
    $\begingroup$ Besides correlations, you will want to check comparability of means, and look for similar shapes of the distributions. E.g., is the new test producing more skew or more outliers than the old. This can be done by sub-pop. too. $\endgroup$ – rolando2 Apr 27 '18 at 11:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.