-
Notifications
You must be signed in to change notification settings - Fork 4
Projects
Welcome to the project page of Brainhack Western 2021!
The projects are ordered based on when participants added them. Here's a list if you're here browsing.
- BIDS - Creating the events.tsv file
- Automated skull-stripping fetal MRI data
- Building a Jupyter Book tutorial for intracellular electrophysiology analysis in python
- Quality Control Visualization App
- Placeholder
BIDS - Creating the events.tsv file
Suzanne Witt (@switt4)
While most of the steps to create a valid BIDS (Brain Imaging Data Structure) directory can be automated, the events.tsv file containing all of the experimental task information still needs to be largely dealt with manually. Making this process a bit more difficult is that for tasks that cannot be easily parameterized using just reaction time and correct responses, the current documentation is limited. For this project, I would like to gather anyone who is at the point of needing to make an events.tsv file for their own study and work on creating more user-friendly documentation for completing the events.tsv file with more real-world examples. And, if we decide it is needed, we will develop the framework for an app that will help researchers take their existing experimental output files and create a valid events.tsv file.
Automated skull-stripping fetal MRI data
Emily Nichols
My project has to do with skull-stripping fetal fMRI data - traditional tools don’t work on fetal fMRI because of the surrounding tissue and the lack of a clear skull, which means 1000s of hours are spent manually segmenting the brain. There are some tools available, but they don’t work well and still result in many hours spent fixing the automatic segmentations. I’m hoping to improve upon available tools so that automated segmentation works.
Building a Jupyter Book tutorial for intracellular electrophysiology analysis in python
Sam Mestern (@sammestern)
Intracellular electrophysiology experiments – specifically patch-clamp and its associated methodology – produce immense datasets consisting of highly structured time-series data. While many sub-fields of neuroscience have adopted standardized beginner-friendly computational analyses/tools, the intracellular patch-clamp field still lacks standardized & beginner-friendly analyses. Many intracellular papers rely on manual analysis with proprietary tools/code. For this brainhack project, we will build a jupyter book outlining some basic analysis that can be done on patch-clamp data, with a focus on building pipelines for batch analysis of patch-clamp data. The tutorials will utilize open-source tools from various sources, including the Allen institute! An early form of the notebooks can be found here.
Quality Control Visualization App
Peter Van Dyken (@pvandyken)
Many automated pipelines include some sort of HTML based Quality Control output, allowing you to quickly scan through visualizations of the results. These are all in-house, custom solutions; there's no easy way to generate nice QC interfaces without manually creating html files. The goal of this project is to design a ReactJS based webapp that reads a JSON file containing a structured account of all the data to be displayed. For instance, the json would have paths to all the images to be displayed, organized by view, subject, contrast, task, or any other arbitrary dimension. The web app would automatically organize all of the data into a clean, modern web app interface, allowing the user to view the data along any dimension. In other words, you could easily view all the different contrasts from a single subject, with a given view, or view all the subjects across a given condition, or any other arbitray "slice" of the data.
An app of this nature will be fairly complicated, so the goal during Brainhack would be to build a very basic prototype. Knowledge of Typescript (a typed superset of Javascript) and ReactJS will be helpful, but anyone interested in learning web app development is welcome. Knowledge or interest in graphic design (especially user interfaces) would also be useful.
[Placeholder text]