Closed LeSasse closed 1 month ago
This also relates to #333.
We need to somehow address the issue that nilearn does not accept any other "method" than these ones here: https://github.com/nilearn/nilearn/blob/4f4730163097457cf9ddb5674ffd158ee8fa822e/nilearn/connectome/connectivity_matrices.py#L497
In my opinion, we might need a custom JuniferConnectivityMeasure
akin to JuniferNiftiSpheresMasker
, which would allow this.
In my opinion, we might need a custom
JuniferConnectivityMeasure
akin toJuniferNiftiSpheresMasker
, which would allow this.
In my opinion, this is reasonable, and would also allow us to set the default to EmpiricalCovariance which is what we use mainly anyways.
In my opinion, we might need a custom
JuniferConnectivityMeasure
akin toJuniferNiftiSpheresMasker
, which would allow this.In my opinion, this is reasonable, and would also allow us to set the default to EmpiricalCovariance which is what we use mainly anyways.
We definitely will have more upsides than downsides from what I understand. I'll get the skeleton ready if @juaml/junifer-core agrees on this and then you can get this in while I get #333 in?
In my opinion, we might need a custom
JuniferConnectivityMeasure
akin toJuniferNiftiSpheresMasker
, which would allow this.In my opinion, this is reasonable, and would also allow us to set the default to EmpiricalCovariance which is what we use mainly anyways.
We definitely will have more upsides than downsides from what I understand. I'll get the skeleton ready if @juaml/junifer-core agrees on this and then you can get this in while I get #333 in?
sounds good! ping me once i can start getting active
In my opinion, we might need a custom
JuniferConnectivityMeasure
akin toJuniferNiftiSpheresMasker
, which would allow this.In my opinion, this is reasonable, and would also allow us to set the default to EmpiricalCovariance which is what we use mainly anyways.
We definitely will have more upsides than downsides from what I understand. I'll get the skeleton ready if @juaml/junifer-core agrees on this and then you can get this in while I get #333 in?
sounds good! ping me once i can start getting active
@LeSasse You should be able to easily implement it by tweaking https://github.com/juaml/junifer/blob/main/junifer/external/nilearn/junifer_connectivity_measure.py
We need to define the value for kind, i guess, so maybe something like:
kind : {"covariance", "correlation", "partial correlation", "tangent", "precision"}
can turn into:
kind : {"covariance", "spearman", "pearson", "partial correlation", "tangent", "precision"}
What do you think
We need to define the value for kind, i guess, so maybe something like:
kind : {"covariance", "correlation", "partial correlation", "tangent", "precision"}
can turn into:
kind : {"covariance", "spearman", "pearson", "partial correlation", "tangent", "precision"}
What do you think
I'd actually keep "correlation" as is so that we're 100% compatible with nilearn but make the Spearman or any other correlation metric like so: "<name> correlation"
, so Spearman can become: "spearman correlation".
sounds good, then the list for implementing the spearman correlation will be:
kind : {"covariance", "spearman correlation", "correlation", "partial correlation", "tangent", "precision"}
sounds good, then the list for implementing the spearman correlation will be:
kind : {"covariance", "spearman correlation", "correlation", "partial correlation", "tangent", "precision"}
Exactly! :D
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 88.56%. Comparing base (
675ccd0
) to head (cc589c3
).
PR Preview Action v1.4.7 :---: Preview removed because the pull request was closed. 2024-08-06 12:51 UTC
Which marker do you want to include?
Currently it does not work out of the box, but some people like it and its a small enough thing to expect it out of the box and not really worth it have users implement their own markers for this.
Is there any publication or available code?
just rank time series and get pearson correlation
Do you have a sample code that implements this outside of junifer?
Anything else to say?
No response