Start to use RAVE on your own data
This section tells you how to configure RAVE for your own projects.
- Data Repository (Add your own data to RAVE repository)
- Preprocess (Convert raw data to RAVE format)
After installing RAVE, if you type init_app(), RAVE will launch an example module called Condition Explore in your default browser, with a demo subject Subject_RAVE_Demo. If you ask questions such as "how can I use RAVE on my data?" or "How am I going to mount a module?". Then you probably want to read this section.
Go to your RAVE root folder. If you just installed RAVE It's usually under ~/
- in MacOS, you can type
Command(⌘)+Spaceto open your spotlight, type~/and pressEnter. - In Windows, it's probably located at
C:\Users\[username]\orC:\Users\[username]\Documents\)
There are two folders rave_modules and rave_data. rave_data is your data repository.
Open rave_data, and you will see two sub-folders data_dir, raw_dir.
This folder stores subject processed data, ready for RAVE to use. It will be handled by RAVE and you don't need to worry too much about it. The only two situations when you will need to change it is when you want to add SUMA files, or when you are adding epoch files. There are two articles telling you how to enable SUMA and how to create epochs.
raw_dir is the folder where you put your raw subject files. You need to manage your original ECoG data here with proper file hierarchy. The following paragraphs will be introducing you how to manage this folder.
If you look inside, there should be a Subject folder in it. The folder name is your subject code. Inside of this folder, there is an block folder 008 and within block folder, you have six sample files with name SubjectDatafile008_ch[1-6].mat
Let's have an example, say you have a subject YAY, it has three experiment blocks (008, 010, and 012) and 126 electrodes (channels):
- Project name:
Demo - Subject code:
YAY - Experiment blocks:
008,010,012 - Number of channels: 126
Folder structure (you need to change some of the names):
raw_data/
| ... (other subjects)
|
+ ---- YAY/
|
+ ---- 008/
| |
| \ ---- YAYDatafile008_ch1.mat
| YAYDatafile008_ch2.mat
| ...
| YAYDatafile008_ch126.mat
|
+ ---- 010/
| |
| \ ---- YAYDatafile010_ch1.mat
| YAYDatafile008_ch2.mat
| ...
| YAYDatafile008_ch126.mat
|
+ ---- 012/
|
\ ---- YAYDatafile008_ch1.mat
YAYDatafile008_ch2.mat
...
YAYDatafile008_ch126.mat
Important:
- Your Matlab file format should be
[SUBJECT_CODE]Datafile[BLOCK]_ch[CHANNEL].mat - Channel number must starts from
1
Once settled, open preprocess module by type
rave::rave_pre_process()
Enter your project name Demo, and subject code YAY and press button Load.
You can use R script rave::rave_pre_process(), or
require(rave)
rave_pre_process()
to enter preprocess app. There are three major steps here:
- Notch filter: filter out 60 Hz (±1), 120 Hz (±2), 180 Hz (±2)
- Common average reference: Select channels that are valid, calculate mean as reference signal
- Wavelet transformation: Perform wavelet, generating both power and phase blocks
- Epoching: Generate epoch files, telling RAVE when the trails start
Before starting any process, you need to load subject data first:
Step 1: Enter subject code and project name
Step 2: Press "Load" button
Step 3: Enter channel, blocks, sample rate
This information will freeze once Notch filter is applied. Please double check, and press "Update Changes" button.
Important: channels must go from 1 consecutively to the maximum channel number.
Go to Notch Filter section, press "Apply Notch Filter" once you saved channel & block information in the previous steps. Confirm the dialogue, and have half cup of coffee, then you will see the following UI. You can go through all the blocks and channels to find bad/epilepsy channels. If you run RAVE on a local machine, you can also export all the channels as PDF files.
Once finished preliminary screening, go to the next step (CAR)
Load notch filtered signals (this might take a while for the first time), and you will see the CAR plot with every channels lined up:
TODO: Add descriptions about this module
Once finished screening, you can press button Calculate Average to generate CAR signal, then you will see a select box appear, with your current reference channels as follows. Also this CAR signal will be rendered in the CAR plot area, with azure color.
* In this case, there are only seven channels and I didn't exclude any one
After Apply CAR, go to Post-CAR tab, and you can see pre- vs. post-CAR comparisons
* If you are unsatisfied with some channels, you can always go to Globals and change settings. However, you need to Re-run CAR to calculate again after changing settings.
Once you are satisfied with CAR results, go to Wavelet and perform wavelet in preprocessing module.
Before applying wavelet, you need to check the channels to be applied by looking at top part of wavelet input box.
There are a few options for wavelet:
- Frequency range: Range of frequencies to apply wavelet
- Frequency step size: By default 2, this means if your frequency range is [2,200], your final power frequencies will be 2, 4, 6, ..., 200
- Number of wavelet cycles: Default 7, for phase resolution
- Target sample rate: Default 100. The result will be down-sampled to 100 Hz.
- Parallel, number of cores: Default is number of cores your CPU has, you can adjust this number to any integer from 1 to the default number. For example, your CPU has 4 cores, you can set 3 cores here and leave one core to edit your paper, watch movies etc.
After all set, press Run Wavelet and wait, or go home :) Once wavelet is finished, a pop-up message will inform you. Then you can go to Post-Wavelet tab to view the results
You need to edit epoching files on your own. TODO: define epoching file format here, and write example script(?)