search

thierry3000 / tumorcode

Simulation of Vascularized Tumors including Angiogenesis, Oxygen, Drugs and Interstitial Fluid
http://www.uni-saarland.de/fak7/rieger/homepage/research/tumor/tumor.html
10 stars 6 forks source link

BulkTissueTumor gneration problems #12

Closed ceserthen closed 6 years ago

ceserthen commented 6 years ago

I am attempting to reproduce the study from your 2013 PLOS paper. But am having trouble reproducing the results. right now I am attempting to generate the tumors with the following commands ( I updated the deafult config settings to run for ~1000 hours (tend))

submitVesselgeneration -p default -t 0 -w 8000 -i 2 -e 1
submitBulkTissue defaultconfig_bulktissue vessels-default-typeA-61x77L130-sample00.h5

When I ran the simulation I got a weird result where the tumor didn't grow. i-2-8mmtumor I then ran a new simulation with the following:

submitVesselgeneration -p default -t 0 -w 8000 -i 1 -e 1
submitBulkTissue defaultconfig_bulktissue vessels-default-typeA-61x77L130-sample00.h5

This seemed to work but the tumors were still smaller than what was reported in the paper. tumbulk-default-typea-sample00-defaultconfig_bulktissue-out0040 png_slice

tumbulk-default-typea-sample06-defaultconfig_bulktissue-out0020 png_slice It's also getting to hot to run my PC for so long right now so I wanted to explore the 6 mm cube instead of the 8 mm cube. So I ran the following commands

submitVesselgeneration -p default -t 0 -w 6000 -i 1 -e 3
submitBulkTissue defaultconfig_bulktissue vessels-default-typeA-47x59L130-sample0$i.h5

After running this I got very similar results to what we found for the 8 mm vessel set with the 2 iterations during vascular generation.

Am I using the wrong settings/parameters/commands? Maybe I need to adjust the lattice scale or something? If so, what do you recommend I use to replicate the study from 2013? Let me know what other information you might need.

thierry3000 commented 6 years ago

I understand your desire to reproduce that study and I agree that the repository should in principle offer that possibility! I contacted the Phd student maintaining the project before I joined 2014. One of the main differences, he explained, is the fact that the 2013 study was conducted on rectangular vessel lattices. In principle I kept the possibility to do rectangular vessel networks, but in practice I never used it in the past 4 years.

Unfortunately the project was not yet versioned in 2013. However I still have some sources and an all the hdf files from the 2013 study at hand. If you could debug the vessel generator and the interface to work with rectangular vessel, I could dig in the hdf files for all the parameters.

thierry3000 commented 6 years ago

After consulting with Dr. Welter, he allowed me to share his sources from 2013. As mentioned, I do not have time to integrated that into the repository right now. But I made it available for your convenience at: www.uni-saarland.de/fak7/rieger/presentation/tumor/ifp_paper_code.tar.gz

I hope that helps.