Add option to show zero for each channel

[cbrnr]

Describe the problem

Currently, it is rather difficult to determine if a given data point lies below or above zero, which can be important e.g. when working with EEG (ERPs). Each channel has a tick on the left y-axis indicating zero, but this only helps when the data point of interest is very close to this indicator.

Describe your solution

A continuous horizontal line (in light gray) would make that much easier. Since the zero point is not important for all kinds of signals, this should definitely be an option to toggle on and off.

Describe possible alternatives

I don't think there's any alternative.

Additional context

No response

[cbrnr]

PS: I think we'd also need to consider whether DC removal is on or off. When it's on, I think that indicator is not the "true" zero, it's the zero after removing DC. Not sure how to indicate this in the plot, but maybe a toggleable button in the toolbar for DC removal would be sufficient? That way, users would at least know if the visualized signals are the original signals or if they have been processed by removing their offsets.

[Radhika-Gajulapalli]

Hello,

My name is Radhika and I am a student interested in contributing to MNE-Python for Google Summer of Code.

I would like to start contributing and I am particularly interested in this issue. Could someone guide me on how to begin working on it?

Thank you.

[cbrnr]

Hello Radhika! Thanks for working on this issue. I would recommend to look into _pg_figure.py and add horizontal InfiniteLine objects (for each DataTrace) to the plot item. Once this is working, I would make it toggleable.

Regarding your development environment, you should have both mne and mne-qt-browser installed and then create a minimal example script that calls raw.plot() (where raw is some toy data that you can create with RawArray).

[Radhika-Gajulapalli]

Hello Clemens,

Thank you for the suggestion. I am currently exploring _pg_figure.py and trying to understand how the plot items work. I will try adding horizontal InfiniteLine objects for each DataTrace as you suggested.

I am also setting up a minimal example using raw.plot() to test the changes.

I will update you once I make progress. Thank you for your guidance.

[cbrnr]

Great! I just made an unrelated PR, but also needed to create a plot with toy data, so you can use and tweak this example:

from numpy.random import default_rng

import mne
from mne import create_info
from mne.io import RawArray

mne.viz.set_browser_backend("qt")


def create_toy_data(n_channels=3, duration=25, sfreq=250, seed=None):
    rng = default_rng(seed)
    data = rng.standard_normal(size=(n_channels, duration * sfreq)) * 5e-6
    info = create_info(n_channels, sfreq, "eeg")
    return RawArray(data, info)


raw = create_toy_data()
raw.set_annotations(
    mne.Annotations(onset=[1, 5], duration=[1, 2], description=["test", "bad"])
)
raw.plot(block=True)

[Radhika-Gajulapalli]

Hello Clemens,

Thank you for providing the example script. I was able to run it successfully and the plot with the toy data and annotations appears correctly.

I am now exploring _pg_figure.py to understand how the traces are drawn and where to add the horizontal InfiniteLine objects for each DataTrace.
I will try implementing this and test it using the example script.

Thank you for the guidance.

[Radhika-Gajulapalli]

Hi Clemens,

I was able to run the example script and explore the DataTrace implementation. I added a first attempt at creating a horizontal InfiniteLine for the trace and confirmed that the project runs correctly with the modification.

Next I will investigate how the channel vertical offsets are handled so the zero reference line aligns properly for each channel.

Thanks for the example and guidance.

[Radhika-Gajulapalli]

Hi Clemens,

I continued exploring the DataTrace implementation and tested adding the InfiniteLine for the trace. The example script runs correctly with the modification.

I am now investigating how the vertical channel offsets (ypos) are handled so the zero reference line aligns correctly for each channel.

Please let me know if this approach makes sense.

Thanks!

[cbrnr]

Sounds good! Feel free to share your progress in a PR - it doesn't have to be perfect, and that way it is easier to provide guidance.

Regarding the channel offsets, in contrast to what I mentioned above, I now think that the horizontal line should only be at zero. If the zero point is not visible due to a large offset, it's better not to show a line at the midpoint. This makes it much easier to understand that we're always showing a zero line.

[Radhika-Gajulapalli]

Hi Clemens,

Thank you for the clarification. That makes sense regarding the channel offsets and why the zero point may not always be visible.

I will try implementing the reference line at the midpoint of the trace so that it remains visible even when offsets are applied. I’ll test this approach and share my progress in a pull request soon for feedback.

Thanks again for the guidance!

[Radhika-Gajulapalli]

Hi Clemens,

I ran the browser locally and explored the visualization to better understand how the channel offsets affect the visibility of the zero point. I was able to run the test plot and verify how the traces are displayed with offsets. I'm currently looking into the relevant parts of _graphic_items.py to understand how the trace positioning works so I can implement the reference line at the midpoint as discussed. I'll share my progress soon.

Thanks!

[Radhika-Gajulapalli]

Hi Clemens,

I was able to run the browser locally and explore how the traces are displayed with channel offsets. I'm currently reviewing the relevant parts of _graphic_items.py to understand how the trace positioning works.

Before starting the implementation, I wanted to confirm whether adding a horizontal reference line at the midpoint of each trace would be the correct approach, or if there is a specific place in the code where you would prefer this to be handled.

Thanks!

[cbrnr]

I would not add the line at the midpoint, but exactly at the zero point. If you need more detailed advice I will have to look at the existing code, which I can do tomorrow.

[Radhika-Gajulapalli]

Hi Clemens,
I would not add the line at the midpoint, but exactly at the zero point. If you need more detailed advice I will have to look at the existing code, which I can do tomorrow.
I am waiting about your response.