Fix SSIMLoss docstring examples: remove redundant 1- prefix

[haoyu-haoyu]

Fixes #8822.

Description

`SSIMLoss.forward()` already returns `1 - ssim_value` (see `monai/losses/ssim_loss.py` line 127):

```python
ssim_value = self.ssim_metric._compute_tensor(input, target).view(-1, 1)
loss: torch.Tensor = 1 - ssim_value
```

…yet the three docstring examples call `print(1-SSIMLoss(...)(x,y))`, which computes `1 - (1 - ssim) = ssim` — the structural similarity value, not the loss. A user copying the example would build a "loss" that increases as images grow more similar and train in the wrong direction.

Removing the `1-` prefix in the three `print(...)` calls makes the examples print the actual loss, consistent with the `Returns` docstring that already says "1 minus the ssim index (recall this is meant to be a loss function)".

Verified locally:

```

SSIMLoss(spatial_dims=2)(x, x) # identical inputs
tensor(0.) # correct loss behavior
```

Before the fix the docstring would have printed `1.0` (the similarity), after the fix it prints `0.0` (the loss).

Types of changes

• Non-breaking change (fix or new feature that would not break existing functionality).
• In-line docstrings updated.

Checks

• I've read the contribution guidelines.
• DCO `Signed-off-by` included.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 0c0a8169-3332-435e-bf08-272ee0cee673

📥 Commits

Reviewing files that changed from the base of the PR and between 089f02d and ee14040.

📒 Files selected for processing (1)

• monai/losses/ssim_loss.py

---

📝 Walkthrough

Walkthrough

This PR corrects three docstring examples in SSIMLoss.forward(). Previously, examples showed print(1-SSIMLoss(...)(x,y)), but since the loss function already computes 1 - ssim_value internally, this would negate the result to the similarity value instead of the loss. The fix removes the redundant 1- prefix from examples for 2D, pseudo-3D, and 3D cases.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	Title accurately describes the core fix: removing the redundant 1- prefix from SSIMLoss docstring examples.
Description check	✅ Passed	Description covers the issue, root cause, fix, and verification. Follows template with non-breaking change and docstring update checkboxes marked.
Linked Issues check	✅ Passed	Changes directly address issue #8822: all three docstring examples have the redundant 1- prefix removed, fixing the negation error that would invert the loss.
Out of Scope Changes check	✅ Passed	Only docstring example lines modified; no changes to API, logic, or unrelated code. All changes are in-scope to the issue.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[ericspod]

Thanks for the update @haoyu-haoyu but this is being addressed in #8821 which will be merged soon.

[haoyu-haoyu]

Hi @ericspod — thanks for the heads-up! Just to double-check, I think #8821 might be the wrong number — that PR is @Zeesejo's fix for the variable swap in JukeboxLoss.forward (issue #8820), and its diff only touches monai/losses/spectral_loss.py, not ssim_loss.py.

I scanned all PRs that touch ssim_loss.py and only #8828 (this PR) is open. Did you mean a different PR, or were you grouping this in with the broader monai/losses/ audit cluster (#8818, #8821, #8828) and planning to take all of them in together?

Happy to close this if there's an SSIMLoss fix landing elsewhere — just want to make sure we don't drop the docstring fix on the floor.