Clustering using CLI

[Sunghyoung Hong]

Hi Frank, I'm an undergraduate from SKKU in south Korea.
I applaud about your amazing DSI studio.

However, I had some problem about 'clustering' using DSI studio by command line interface. Clustering has been performed well with DSI studio by GUI, and downloading each cluster into .tt file was available.
However, it didn't work in CLI environment. It only produced the number of tracts of each cluster rather than .tt file of each cluster.

I attached the code below which I had written and the pictures that showed our problems. 


shkim@hercules:/local_raid2/03_user/sohyeong/01_project/01_dti/04_toolbox$ singularity exec ./dsistudio_latest.sif dsi_studio --action=trk --source=./HCP842.1mm.fib.gz --fiber_count=100000 --cluster=1,50,0,jenny.txt
DSI Studio Aug 3 2021, Fang-Cheng Yeh
source=./HCP842.1mm.fib.gz
action=trk
loading ./HCP842.1mm.fib.gz...
otsu_threshold=0.6
fa_threshold=0
dt_threshold=0.2
turning_angle=0
step_size=0
smoothing=0
min_length=30
max_length=300
method=0
initial_dir=0
interpolation=0
seed_plan=0
random_seed=0
check_ending=0
tip_iteration=0
fiber_count=100000
seed_count=0
thread_count=72
start tracking.
thread_count=72
finished tracking.
100000 tracts are generated using 138907 seeds.
0 tracts are removed by pruning.
The final analysis results in 100000 tracts.
compressing trajectories
saving to HCP842.1mm.fib.gz.tt.gz
cluster=1,50,0,jenny.txt
cluster method: 1
cluster count: 50
cluster resolution (if method is 0) : 0 mm
run clustering.
cluster label saved to jenny.txt


I'd appreciate if you check the code and let me know how to get the clusters in .tt or .trk file with CLI.

Thanks!
Joseph

[Frank Yeh]

The command line output is the cluster label in text format, and it can be loaded using [Step T3 Fiber Tracking][Tracts Misc][Open Cluster Labels] to get the same results.

Would that work for you?

Frank

--
You received this message because you are subscribed to the Google Groups "DSI Studio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/cf9624fa-206b-4126-8d58-3da3729156can%40googlegroups.com.

[Sunghyoung Hong]

Thanks, Frank.

It did work partially, but there are some problems still remained.

I expected the .fib file was supposed to be clustered bundle by bundle like the attached image_4 like what I did in the GUI,

however, the clusterings were not separated into bundles rather whole brain itself with its sparse density like the attached image_5.

How can I get clustered bundles like what I did in the GUI?

And I have an extra question.

When I cluster bundles using CLI, is there a way to obtain output as .tt file immediately without going through the GUI?

Thanks for your rapid response!

Joseph

2021년 9월 9일 목요일 오전 8시 0분 28초 UTC+9에 Frank Yeh님이 작성:

[Frank Yeh]

I expected the .fib file was supposed to be clustered bundle by bundle like the attached image_4 like what I did in the GUI,

however, the clusterings were not separated into bundles rather whole brain itself with its sparse density like the attached image_5.

How can I get clustered bundles like what I did in the GUI?

I am sorry it is unlikely to be implemented in CLI. 

And I have an extra question.

When I cluster bundles using CLI, is there a way to obtain output as .tt file immediately without going through the GUI?

A possible approach is to write a MATLAB code to separate the output into multiple files based on the cluster label.

Best,

Frank 

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/6dbe3078-102f-44d5-9c55-86c70b29241dn%40googlegroups.com.

[Sunghyoung Hong]

Thanks and sorry, Frank.

I changed parameters in CLI and it worked well what I did in the GUI.

Best regards.

Joseph 

2021년 9월 10일 금요일 오후 12시 37분 46초 UTC+9에 Frank Yeh님이 작성: