huthlab / deep-fmri-dataset Goto Github PK

Hi,

Thanks for the great work!

As mentioned in #8 , you recommend using fmriprep to preprocess the raw data and it gives very similar results. I'm currently utilizing fmriprep as well and was curious about the specific parameters you're using with fmriprep. Additionally, I'd like to know which confounds you've selected for denoising the preprocessed fMRI data. Thank you for sharing your insights!

Best regards

Hey,
first of all, thanks for your work!

I was wondering how i can test the trained encoding model on new data? do you have any inference code?

Thank you for sharing this remarkable dataset with us! I noticed that you have open-sourced preprocessed fMRI data and visualization data using pycortex. While using this data for experiments, I encountered two issues.

Firstly, the selected voxels after preprocessing do not have ROI labels, making it difficult to select specific voxels (such as those in the AC region). I noticed that sub-UTS02 provides an rois.npz file, which provides ROI labels for 303,823 voxels. However, this doesn't match the dimensions of the 94,251-dimensional vectors extracted from the preprocessed fMRI data, which is a bit confusing.

Secondly, when using pycortex to plot volume data, I noticed that each voxel does not have corresponding i, j, k coordinates relative to the volume with a shape of (54, 84, 84). This seems to be a missing piece of crucial information for projecting the values onto surfaces.

If it is possible, I kindly request that you provide me with the necessary information and access to it. I understand that sharing data requires time and effort, and I genuinely appreciate your consideration of my request.

Dear HuthLab team,

I would like to start by expressing my sincere gratitude for sharing such high-quality code and dataset. Your project is exceptionally well-structured, and it has been immensely helpful in my research work. I have successfully replicated the experiment results using your code, which is a significant stride forward for me.

However, while going through your paper and trying to understand the code, I noticed a reference to a script named "significance_testing.py." As described, this script implements a parallelized block-wise permutation test, taking in vectors of predicted and true voxel responses, the number of permutations to test, and the size of each permutation block, returning the associated p-value for each voxel. Additionally, it contains a function to control the false discovery rate using the Benjamini-Hochberg procedure.

Regrettably, I was unable to find the "significance_testing.py" file in the current open-source code repository. I would be immensely grateful if you could provide this missing piece of code. I am looking forward to fully validating and understanding your research work.

Thank you once again for your remarkable contribution, and I am eagerly awaiting your response.

Warm regards

Hello, thank you for your wonderful work and making this data open-source!

I'm having an issue when trying to download the data, getting the following error:

[email protected]: Permission denied (publickey).
install(error): data/ds003020 (dataset)
[Failed to clone from all attempted sources: ['[email protected]:OpenNeuroDatasets/ds003020.git']

Any idea what the issue could be?

Thanks,
Chandan

Hello, thanks again for this wonderful dataset!

Apologies if this is a basic question but I have been struggling to visualize the results of the encoding model (specifically the test correlations in corrs.npz) as a flatmap using pycortex. I see that you've kindly provided the anatomical files for each subject, but when I try to load them in (e.g. following the docs here) the directory structure doesn't seem to align for visualization.

Is it possible to visualize the resulting corrs with the provided data or is some of the necessary anatomical information missing? Thanks a bunch!

Best,
Chandan

Not sure if the TextGrids are available on openneuro + need to download 'em too!

(Note: I assume that my installation of git annex is correct and I have installed all dependencies from requirements.txt)

(cs1430) marcs@marcos-mbp-6 encoding % git annex init "serre_fmri_encoding"
init serre_fmri_encoding (recording state in git...)

Remote origin not usable by git-annex; setting annex-ignore

https://github.com/HuthLab/deep-fMRI-dataset.git/config download failed: Not Found
(Auto enabling special remote s3-PUBLIC...)
ok
(recording state in git...)

(cs1430) marcs@marcos-mbp-6 encoding % python encoding.py --subject UTS03 --feature eng1000
Saving encoding model & results to: /Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/results/eng1000/UTS03
Pattern match failed. File format not recognized.
Traceback (most recent call last):
File "/Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/encoding/encoding.py", line 57, in
downsampled_feat = get_feature_space(feature, allstories)
File "/Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/encoding/feature_spaces.py", line 178, in get_feature_space
return _FEATURE_CONFIGfeature
File "/Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/encoding/feature_spaces.py", line 159, in get_eng1000_vectors
wordseqs = get_story_wordseqs(allstories)
File "/Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/encoding/feature_spaces.py", line 15, in get_story_wordseqs
grids = load_textgrids(stories, DATA_DIR)
File "/Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/encoding/ridge_utils/stimulus_utils.py", line 13, in load_textgrids
grids[story] = TextGrid(open(grid_path).read())
File "/Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/encoding/ridge_utils/textgrid.py", line 153, in init
self.text_type = self._check_type()
File "/Users/marcs/Downloads/serre_lab_downloads/combined/serre_lab_complete/deep-fMRI-dataset/encoding/ridge_utils/textgrid.py", line 208, in _check_type
raise ValueError("File format not recognized")
ValueError: File format not recognized

Your dataset is wonderful, thank you for your great contribution!

I notice that in the dataset paper, you said in the "fMRI preprocessing" session that "fMRI preprocessing was only done on the derivative data." I wonder if the derivative here is generated from the raw data and then further processed to generate the "preprocessed_data"? If so, could you share how you get the derivatives and the code for preprocessing?

Looking forward to your replies!