HW 2

Due by 11:59pm the evening of 9/24.

Textbook Reading

Read chapter 8.

Regular Expressions

Write a regular expression to extract (i.e. store in a regex group) the italicized portions of the following strings. The same regex should work for each set of strings.

Your expressions will be evaluated using the python mode of regex101.com

#####A BIDS-style file name (set 1)

• sub-1000_task-nback_acq-singleband_run-03_bold
• sub-EXP23_task-rest_acq-multiband_run-01_bold

#####A 10 digit US phone number (set 2): You should capture 3 groups, containing the area code, exchange and number, but not any separators or country code.

• +1 123-456-7891
• (123) 555-5555
• 860.865.9805
• 1-800-865-9805

Submit your two regular expressions in a Markdown file regex/README.md

Discussion Paper Reading

Read at least one of the papers posted in #discussion on Slack and be prepared to discuss in class. Feel free to discuss in the #discussion channel before class!

Start Learning Python

Although shell scripting is a useful tool that we will continue to rely on, some things are better done in a more powerful language such as Python. Begin the free modules of the Python Code Academy course. You should try to get 1/3 to 1/2 through or so.

Overview

In this homework, you will practice downloading your data from NiDB and converting it into a BIDS structure.

• Input: a directory of DICOM files downloaded from NiDB, possibly containing subdirectories
• Output: a directory of NIFTI files, organized in BIDS format
• Method: You will accomplish the conversion by writing a script (located at scripts/convert.sh with respect to the hw2 repository) that can be run in a container as

docker run [-v ...] rhancock/burc-lite <PATH>/convert.sh PARTICIPANT

where

• [-v ...] is replaced by appropriate options to attach parts of your local filesystem as volumes to the container
• <PATH>/convert.sh is the appropriate path to the script
• PARTICIPANT is the identifier for the participant to convert.

Download the data

Download the IBRAIN002 (S0318ANN) data from NiDB. More documentation about NiDB is available on the Wiki

The appropriate export options are below

After downloading the data, unzip the file, which will create a directory named IBRAIN002

Directory structure

1. Fork and clone the bircibrain/hw2 repository to your computer
2. Create the following directories in your hw2 folder:
   • bids/, for your BIDS-compliant output directory
   • raw/, for the DICOMs you downloaded from NiDB
   • The existing scripts folder is where you will put your code
3. Place the IBRAIN002 folder (the folder, not the contents) in your raw directory.

Run the container

Run the rhancock/burc-lite Docker container interactively with the appropriate -v flags to attach your hw2 directory inside the container. You can attach to /bind inside the container. You should also include -v /tmp:/tmp.

Generate BIDS sidecars

Use the dcm2bids_helper command inside the container to extract metadata from the DICOM files. This takes two arguments:

• -d DICOM_DIR, where DICOM_DIR is the path to the IBRAIN002
• -o OUTPUT_DIR, where OUTPUT_DIR is the path where the BIDS files will go

This will convert the DICOMs and create some json files containing information about the scans.

Edit scripts/dcm2bids.json

scripts/dcm2bids.json contains the dictionary that matches converted DICOM files to their corresponding BIDS structure.

It contains several sections of the form:

{
  "dataType": "func",
  "modalityLabel": "sbref",
  "customLabels": "task-words",
  "criteria": {
      "ProtocolName": "INSERT Correct ProtocolName",
      "ImageComments": "Single-band_reference",
      "PhaseEncodingDirection": "j-"
  }
},

for each of these, you will need to replace the ProtocolName value in scripts/dcm2bids.json with the appropriate ProtocolName from the json file corresponding to the scan. Do this for for the for the following "modalityLabel": "sbref"

007_IBRAIN002_fMRI_words_run1_20171114132143
013_IBRAIN002_fMRI_motor_localizer_20171114132143.nii.gz

and the following "modalityLabel": "bold" scan series:

008_IBRAIN002_fMRI_words_run1_20171114132143.nii.gz
014_IBRAIN002_fMRI_motor_localizer_20171114132143.nii.gz

Create the conversion script

Use the dcm2bids command inside the container to convert the files. You should modify scripts/convert.sh to add the appropriate commands at line 4.

In the interactive container, you can run

dcm2bids -h

to view the documentation.

You will want to set appropriate values for -d, -p (IBRAIN002), -c (the dcm2bids.json file), and -o.

Hints

• Consider which paths need to be referenced relative to your own computer and which need to referenced relative to the container
• Shell scripts can take arguments. These can be referred to using the special variable references $1, $2, ..., where $1 provides the value of the first argument after the script name. It would be useful to use $1 to control which participant is processed.
• Documentation for dcm2bids

Submit

Commit your work and create a pull request. The data files will not be committed to conserve space.