The peaks at small T2 appear less intense than the peaks at higher T2 even if the peaks area is correctly evaluated. This suggests a relaxation time dependent distortion of the reproduced peak line-width.
If that helps, I recommend reading:DOI: 10.1002/jmri.24870 , in which the dependency of biexponential T2's on the choice of echo times investigated. There might be estimation errors. If after reading you found i relevant and you had further questions do not hesitate to contact me.
This is a numerical simulation. So the start data was well sampled and there are no unwanted baselines. But indeed, comparing to other components, the faster signal (shorter T2) should be more less sufficiently sampled. Are there any strategies to solve this problem?
Cannot read the .mat file, unfortunately - I'm not a Matlab user.
However, the color code goes as a function of the spin-lattice time (and consequently as a function of the scan delay time), right? While the .docx document reports the ILT of one CPMG plot at fixed T1 as a function of the inversion routine, right ? Which T1?
In order to assess whether CPMG data are a good input to the ILT, they should be put in a semi-logarithmic scale - this allows you to appreciate fine details of the plot. I've attached an experimental example of CPMG data that are not a good input: two T2 times here (one around 90 us and one around 10 ms) with no significant baseline, but very poor sampling of the fastest decay due to limitations of the exp apparatus.