asd_diagnosis_fmri

Presentation: https://connorparish.com/extra_files/parish_capstone_pres.pdf

Running CPAC on AWS EC2 instance

• Amazon linux c.5.2xlarge
• 150GB EBS used

1. sudo yum update -y
2. sudo yum install git -y
3. git clone https://github.com/Connorpar/asd_diagnosis_fmri.git
4. rsync -av -e "ssh -i ~/.aws_creds/cparish_aws.pem" ~/.aws_creds/rootkey.csv [email protected]:~/asd_diagnosis_fmri/
5. https://docs.aws.amazon.com/AmazonECS/latest/developerguide/docker-basics.html
6. Pull the latest cpac image using sudo docker pull fcpindi/c-pac:dev-1.7.2
7. sudo docker run --rm -u $UID:$UID -v ~/asd_diagnosis_fmri:/wkdir -v /tmp:/scratch fcpindi/c-pac:dev-1.7.2 s3://fcp-indi/data/Projects/ABIDE/RawDataBIDS s3://cpacpreprocessed/abideI_output/ participant --pipeline_file /wkdir/cpac_configs/settings/abide_cpac_pipeline_cloud.yml

Running ASD Class training

• notebooks/Subject_Generator.ipynb : Create Subject objects
• notebooks/Data_generator.ipynb : Generate Datasets for training using Subjects
• notebooks/colab/ASD_class_pure.ipynb : LSTM and Transformer training without pre-training
• notebooks/colab/rs_fMRI_transfomer_pretraining_total.ipynb : Transfomer multi-task and pre-training

Data Sources

• s3://fcp-indi/data/Projects/ADHD200/Outputs/cpac/raw_outputs/pipeline_adhd200-benchmark__freq-filter/
• s3://fcp-indi/data/Projects/ACPI/Outputs/ants_s1_g0/

Errors

• NYU_2: FATAL ERROR:�[0m tpattern file /tmp/resting_preproc_sub-29155_ses-1/_scan_rest_run-1/scan_params_0/tpattern.txt has 33 values but have 34 slices