# vim-1: EstimatedResponses.mat have some unexpected NaN values.

### vim-1: EstimatedResponses.mat have some unexpected NaN values.

Posted by Huawei Xu at November 01. 2020Hi, there!

I extracted roi data from EstimatedResponses.mat. I found there are some unexpected Nan values.

for subject1's training data, V1 has 1331 voxel; V2 has 2208 voxel; V4 has 1550 voxel; LO has 928 voxel; the total is **6017**.

for subject2's training data, V1 has 1513 voxel; V2 has 1982 voxel; V4 has 1022 voxel; LO has 358 voxel; the total is **4875**.

It is in accordance with a study reported — "If we consider only the main afferent pathway of the ventral stream (V1, V2, V4, and LO) then 1786 of **6017** and 768 of **4875** voxels remained for S1 and S2, respectively." (Guclu, U. , & Van Gerven, M. A. J. . (2015). Deep neural networks reveal a gradient in the complexity of neural representations across the ventral stream. The Journal of Neuroence, 35(27).)

But there are some NaN values in these data. After voxels which contain NaN values were droped:

for subject1's training data, V1 has 1294 voxel; V2 has 2083 voxel; V4 has 1535 voxel; LO has 928 voxel; the total is **5840**.

for subject2's training data, V1 has 1419 voxel; V2 has 1890 voxel; V4 has 1022 voxel; LO has 358 voxel; the total is **4689**.

There were many studies used the data. But I didn't remeber any study had reported NaN value existed.

So may be there was something wrong with the data?

Thank you for your help!

### Re: vim-1: EstimatedResponses.mat have some unexpected NaN values.

Posted by Jeff Teeters at March 04. 2021I asked the data contributor about this. He replied that:

"The data should still be correct. NaNs occur when insufficient coverage of the brain occurs in a given run or scan session due to head motion. (The study used a partial brain coverage protocol.) Thus, the NaNs essentially simply mean that data are unavailable for those cases."