ADHD datasets for 3 dimensional neural network predicting model

Submitted by Yasu on Thu, 05/31/2018 - 11:29
Hi
 
My name is Yasuyuki Okubo and I'm building 3 dimensional convolutional neural network predicting model inspired by the paper below.
 
https://ieeexplore.ieee.org/document/8067637/all-figures
 
From this site, I reached R-fMAPs project for the datasets and succussfully downloaded the ADHD datasets for ADHD200 (4. Attention Deficit Hyperactivity Disorder 200 Sample (ADHD-200) ).
 
http://mrirc.psych.ac.cn/DownloadRfMRIMaps
 
I'm, however, confused the datasets which I should use as there are many datasets though I read explanation.
 
I choosed both fMRI - fALFF and sMRI - GM below for my model but don't know whether I picked right ones.
 
fMRI-fALFF: ADHD200_RfMRI_1/results/fALFF_FunImgARCW/
 
sMRI-GM: ADHD200/VBM_1/c1/
 
I sincerelly appreciate if you address me for the datasets.
 
If this question should go to another groups or forums, then I sincerelly appreciate if you address me for the groups or forums.
 

 

Thank you very much.

 

 

 

Forums:

YAN Chao-Gan

Fri, 06/08/2018 - 06:08

For GM, you should go to wc1, that's gray matter density in MNI space.

fALFF is correct.

Yasu

Sun, 06/10/2018 - 04:15

Thank you very much for your reply and advise.

I really appreciate.

I have one more question regarding datasets

Do I have to preprocess the data using tools such as DPARSF?

My 3D Neural Network model does not seem to categorize properly.

Even I changed my predicting model such as tuning layers, parameters, and data number, the model gives always same accuracy.

Regarding fALFF, I reshape 61/73/61 to fit 3D model.

Thank you very much in advance for your help.

YAN Chao-Gan

Thu, 06/28/2018 - 07:35

No need for reprocessing.

Please check your model and your code for CNN.