search

PyTomography / slicer_spect_recon

3d slicer extension for pytomography to enable easy reconstruction of SPECT images
MIT License
16 stars 3 forks source link

What is the Random Seed Represent? #15

Open akmal871026 opened 2 weeks ago

akmal871026 commented 2 weeks ago

Dear @ObedDzik @lukepolson @ahxmeds ,

May I know what the Random Seed Represents?

What is the number should i put?

image

ObedDzik commented 2 weeks ago

Any random whole number should be fine. It is just to ensure reproducibility of results.

lukepolson commented 2 weeks ago

SIMIND data gives you the mean number of counts per each detector bin. To generate clinically realistic data, you need to generate Poisson counts for each bin. The random seed is the seed used in the random number generator when getting this Poisson data. It allows for reproducibility since the same seed will give you the same random numbers.

akmal871026 commented 2 weeks ago

Ok, tq luke. Then what should number should I put? Is it enough I put 1? Or 10?

On Tue, 10 Sept 2024, 06:52 Luke Polson, @.***> wrote:

SIMIND data gives you the mean number of counts per each detector bin. To generate clinically realistic data, you need to generate Poisson counts for each bin. The random seed is the seed used in the random number generator when getting this Poisson data. It allows for reproducibility since the same seed will give you the same random numbers.

— Reply to this email directly, view it on GitHub https://github.com/PyTomography/slicer_spect_recon/issues/15#issuecomment-2339288672, or unsubscribe https://github.com/notifications/unsubscribe-auth/ASFIPQYLZ7ZHIK7ITXM5TILZVYRB3AVCNFSM6AAAAABN4DFFFCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGMZZGI4DQNRXGI . You are receiving this because you authored the thread.Message ID: @.***>

lukepolson commented 2 weeks ago

Any number is fine. It's more for scientific reproducibility: i.e. somebody with your exact SIMIND data should be able to generate the same noisy data as you. So it's a good idea to keep track of whatever number you use.

akmal871026 commented 2 weeks ago

ok Tq Dr. Luke..

May I get your email address?

___

MOHD AKMAL BIN MASUD

H/P: +60143321631

On Tue, Sep 10, 2024 at 8:31 AM Luke Polson @.***> wrote:

Any number is fine. It's more for scientific reproducibility: i.e. somebody with your exact SIMIND data should be able to generate the same noisy data as you. So it's a good idea to keep track of whatever number you use.

— Reply to this email directly, view it on GitHub https://github.com/PyTomography/slicer_spect_recon/issues/15#issuecomment-2339384348, or unsubscribe https://github.com/notifications/unsubscribe-auth/ASFIPQ5ISRFFZSDKO5PFEETZVY4W3AVCNFSM6AAAAABN4DFFFCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGMZZGM4DIMZUHA . You are receiving this because you authored the thread.Message ID: @.***>