How to adjust the parameters of the coronary boundary?

[yan qing]

Hi everyone :
I have finished the coronary demo project, and I want simulation my own coronary project. I'm confuse how to adjust the parameters of the coronary boundary? It is impossible to adjust the prameters after one simulation, It will cost to much time.
I have noticed the ROM simulation, it simplefy the simulation. should I adjust the parameters to make the simulation pressure more close to the clinical pressure? Is there any tricks to help me to get the perfect parameters to finish my project?

Thanks,
Yan

[Weiguang Yang]

Yes, using 3D simulations to tune your coronary BCs would be expensive and slow unless you know exactly which parameter should be tuned. There are numerous papers on BC tuning. An easy but effective way is to use a lumped parameter model. For example, you can use several parallel resistors to approximate the 3D model and connect them to the same coronary lumped parameter models used in your 3D simulation. In the beginning, you need to calibrate your LPM to make sure the resistance values for 3D branches represent 3D resistance well. Your lumped parameter model should be able to read input parameters and output resulting pressures and flow, for example systolic, diastolic and mean pressures and coronary flow/systemic flow ratio. Then use any optimizer you like to reduce the differences from clinical targets. When you are happy with the result, apply the coronary parameters to your 3D simulation and see if it is good enough.

[yan qing]

Hello,
Thank you for your reply.
I have confused about "In the beginning, you need to calibrate your LPM to make sure the resistance values for 3D branches represent 3D resistance well."
Is that mean I set the initial 3D model resistors arbitrary or I get the initial resistors using the method like the 1D Mesh? The 3D resistors and the coronary lumped model are all unkown, how can I calibrate the resistance values for 3D branches.
I list my pipeline blow, can you tell me is there any question I need to correct?
1.Generate the 1D mesh and generate the simplefy 1D parameters about the 3D model
2.Set the initial coronary lumped model parameters acoding to the diameter of the coronary
3.Optimize the coronary lumped model to reduce the differences from clinical targets using oneDsovler
4.Run the 3D model using the optimized coronary lumped parameters.

And I have another question, Is there any way I can use the heart lumped model as the input condition in SV?

Thank you
Yan

[Weiguang Yang]

When you use lumped parameter models (LPMs) to represent your 3D simulations, you have two parts: 1 the LPM for the 3D geometry 2. the LPM for the distal coronary arteries which is kept the same as the outflow BCs used in the 3D simulations. You want to identify the parameters for the 2 part to have 3D simulations match the clinical targets. Because the LPM is used to tune these parameters, you have extra unknown LPM parameters for the 3D geometry. So you can try the following:
1. fix the parameters for the outlet LPMs and adjust the parameters for the 3D geometry such that the LPM output is close to the 3D simulation output though the result doesn't match the clinical target well. 2. Then fix the parameters for the 3D geometry and tune the parameters for the outlet LPMs. i hope these clarify my previous response.

Yes, you can use 1D model as a surrogate to tune BCs for your 3D simulations.
How does the 1D output compare with the 3D output for the same set of BCs? If they agree well, the procedure you list looks fine to me. If the 1D results consistently overestimate or underestimate the pressure/flow compared to 3D results, you may need to account for the difference during the tuning/optimization.

If you want to incorporate a heart lumped parameter model into your 3D simulations, you can use GenBC. GenBC allows user defined LPMS coupled to 3D inlets and outlets. Here is some information https://simvascular.github.io/docsGenBC.html

[yan qing]

Hello,
Thank you for your patient reply, now I undstand the procedure of the optimization of lumped model.If I finish the simulation, I will tell you.

Thank you
Yan

[Wei Wu]

When you use lumped parameter models (LPMs) to represent your 3D simulations, you have two parts: 1 the LPM for the 3D geometry 2. the LPM for the distal coronary arteries which is kept the same as the outflow BCs used in the 3D simulations. You want to identify the parameters for the 2 part to have 3D simulations match the clinical targets. Because the LPM is used to tune these parameters, you have extra unknown LPM parameters for the 3D geometry. So you can try the following:
1. fix the parameters for the outlet LPMs and adjust the parameters for the 3D geometry such that the LPM output is close to the 3D simulation output though the result doesn't match the clinical target well. 2. Then fix the parameters for the 3D geometry and tune the parameters for the outlet LPMs. i hope these clarify my previous response.

Yes, you can use 1D model as a surrogate to tune BCs for your 3D simulations.
How does the 1D output compare with the 3D output for the same set of BCs? If they agree well, the procedure you list looks fine to me. If the 1D results consistently overestimate or underestimate the pressure/flow compared to 3D results, you may need to account for the difference during the tuning/optimization.

If you want to incorporate a heart lumped parameter model into your 3D simulations, you can use GenBC. GenBC allows user defined LPMS coupled to 3D inlets and outlets. Here is some information https://simvascular.github.io/docsGenBC.html

Thank you for your detailed explanation on tuning. However, there are some points I need your futher help:
1, you mentioned "parameters for the 3D geometry", are they RCR parameters for aortic branches?
2, for "such that the LPM output is close to the 3D simulation output though the result doesn't match the clinical target well", I don't understand which 3D simulation your refer?

Best

[Weiguang Yang]

Hi,
1. The parameters for the 3D geometry can be a resistor or an rcr or several LPMs. For example, you want to create an LPM surrogate for a 3D bifurcation model, you can choose to use 3 resistors to represent the three segments or use 2 resistors or 2 RCR models in parallel to represent the 3D model. You can start with resistors and gradually increase the number of LPM parameters for your surrogate if the simplest one doesn't work.
2. I mean the 3D simulation with your default or initial BCs. I suppose you have some 3D simulation results. They may not match your clinical target very well. But you can use these results to calibrate your surrogate making these two models (3D and surrogate) agree with each other for the same BCs. Then you fix the surrogate model and tune BCs using the surrogate model output.
I hope these clarify my previous posts.
Thanks

[Wei Wu]

Hi,
1. The parameters for the 3D geometry can be a resistor or an rcr or several LPMs. For example, you want to create an LPM surrogate for a 3D bifurcation model, you can choose to use 3 resistors to represent the three segments or use 2 resistors or 2 RCR models in parallel to represent the 3D model. You can start with resistors and gradually increase the number of LPM parameters for your surrogate if the simplest one doesn't work.
2. I mean the 3D simulation with your default or initial BCs. I suppose you have some 3D simulation results. They may not match your clinical target very well. But you can use these results to calibrate your surrogate making these two models (3D and surrogate) agree with each other for the same BCs. Then you fix the surrogate model and tune BCs using the surrogate model output.
I hope these clarify my previous posts.
Thanks

Your explanation is very clear.
Thank you so much, Weiguang!!

[Mandar Kulkarni]

Hello,

Thanks for all the explanation in these posts.

In the Coronary Normal example results, there is a mention of the makecortfile.py script for tuning the scaling factors for appropriate coronary flows. Is it available on github, possible with an example?

Thanks,
Mandar

[David Parker]

Hi Mandar,

I have found the file and put it here https://github.com/ktbolt/cardiovascula ... rt_file.py for now. Later I will add it to the Coronary Project download.

Cheers,
Dave