I am conducting a study on different image reconstruction algorithms for computed tomography (CT). My aim is to investigate which of the image reconstruction algorithms that have the highest signal-to-noise ratio (SNR). My dataset is 100 patients who underwent a CT brain scan. Each CT scan will be reconstructed with 4 different image reconstruction algorithms. Then I will measure the SNR for each algorithm and patient and compare the different reconstruction algorithms with regard to the SNR. So, we scan the patient's head and then the computer at once provides four different image reconstruction algorithms of the scan (4 different image sets) and then we measure the SNR of each image set and compare it to the other reconstruction algorithms (other image sets). Or, if it makes it easier for you to think. Consider the following corresponding scenario: You take a blood sample (corresponding to "CT scan") from the patient and measure the glucose level with 4 different apparatus (corresponding to the 4 different image reconstruction algorithms). And then you compare the blood glucose level (corresponding to "the SNR") between the four different apparatus. So, my question is if I should use a repeated measures ANOVA or just an ordinary one-way ANOVA.
3
-
$\begingroup$ It will be easier for people to understand the question/problem if you could describe the dataset. Also, what is your research question ? $\endgroup$– Robert LongJan 8, 2021 at 20:06
-
$\begingroup$ Hello Robert, thank you for your reply. The dataset is the following: 100 patients who underwent a CT brain scan. The CT brain scans will be reconstructed with 4 different image reconstruction algorithms. I will measure the signal-to-noise (SNR) ratio of a ROI placed on a certain brain level at a certain area of the brain. The SNRs will be compared between the different image reconstructions. Should I perform a repeated measures ANOVA since it is the same patients reconstructed with 4 different image reconstruction algorithms? Or should i choose an ordinary one-way ANOVA? Thank you! $\endgroup$– MichaelJan 8, 2021 at 21:20
-
$\begingroup$ I have updated my question and provided more information about my study as well as a corresponding scenario. $\endgroup$– MichaelFeb 27, 2021 at 11:53
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
15
Since you have repeated measures within subjects it is necessary to control for the non-independence of observations within subjects. That is, observations within one subject are likely to be more similar than observations in other subjects.
So in this case you will need to use a repeated measures ANOVA
answered Jan 9, 2021 at 18:56
-
-
$\begingroup$ @Michael does this answer your question ? If so then please consider upvoting, and marking it as the accepted answer. If not please let us know why ? $\endgroup$ Jan 11, 2021 at 11:03
-
$\begingroup$ Hello Robert! Thank you again for your reply. Do you mean "observations within one subject are likely to be more similar THAN observations in other subjects", not "observations within one subject are likely to be more similar TO observations in other subjects". Apart from that I totally agree with you. $\endgroup$– MichaelFeb 18, 2021 at 22:14
-
$\begingroup$ @Michael YES ! thank you for spotting that typo - I've edited my answer accordingly :) $\endgroup$ Feb 18, 2021 at 22:21
-
1$\begingroup$ I still think some more details can be helpful. How many patients? How do you measure SNR? ... And pleas, add all new info incomments (both here and on the Q) to the Q as an edit. Not everybody reads comments! $\endgroup$ Feb 27, 2021 at 15:25