Scatter rate estimation for LM-TOF-MLEM

Dear Castor Developers and Users,

I am trying to incorporate the information about the scatter correction in LM-TOF-MLEM reconstruction. I have calculated the scatter fraction for each TOF bin. The values of scatter rate are in range 1-2. I have applied them for the reconstruction, however the reconstructed images seem to be not corrected for scatter. Please find the attached screenshot. On the top panel, the reconstructed image with true+scatter coincidences is presented and on the bottom the same coincidences were used but with scatter correction applied. On the right corresponding line profiles are presented (ots - no scatter correction; corrected1 - scatter correction applied)

FYI: The total number of coincidences used for the reconstruction is 1 order of magnitude lower than the number of possible LORs.

Do you have any idea what kind of values should be applied for the scatter rate? Did you have any similar case and found the solution how the scatter rate should be calculated?

All the best,
Jakub

Hi Jacub,

Have you divided the estimated TOF bin scatter rates with the spatial width of the TOF bin, directly in the list-mode datafile, as explained in the documentation (general and TOF-specific) ?

Marina

Hi Marina,

many thanks for your answer. could you please clarify what do you mean to "divide the estimated TOF bin scatter rates with the spatial width of the TOF bin, directly in the list-mode datafile". Once I estimate the scatter fraction I presume I should divide it by the TOF bin (e.g. 200ps)? However, I do not specify the TOF bin size in my header. Unless the bin size is then defined as a TOF resolution?

Example of the header which I am using:

Data filename: data_df.Cdf
Number of events: 298175691
Data mode: list-mode
Data type: PET
Start time (s): 0
Duration (s): 500
Scanner name: scanner1
Calibration factor: 1
Isotope: unknown
TOF information flag: 1
TOF resolution (ps): 200
List TOF measurement range (ps): 3000
Scatter correction flag: 1

What is the way to proceed in that case?

All the best,
Jakub

Hi Jakub,

You should divide your scatter rate with the size of the TOF bin converted to mm. That’s if you’re using the “continuous” LM-TOF recon, which is run if you don’t specify the size of the TOF bin in the header. If you specify the size of the TOF bin in the header, then you don’t have to change your scatter rates per bin, that’s easier, I should have mentioned this option first. The TOF resolution field is only for TOF resolution. The documentation is maybe not so clear about these options?
Let me know if this helps.

Cheers

Marina

Hi Marina,

many thanks for your comments. Unfortunately, the solution you provided didn't change the results much. The greater background in the middle of the imaged object, characteristic for the lack of scatter correction is still visible. We are suspecting that probably we need greater coincidence statistics to sort it out.

I will let you know if we found anything interesting.

All the best,
Jakub