Feat: add ignore_index support to DiceLoss

monai/losses/dice.py

@@ -67,45 +67,11 @@ def __init__(
         batch: bool = False,
         weight: Sequence[float] | float | int | torch.Tensor | None = None,
         soft_label: bool = False,
+        ignore_index: int | None = None,
     ) -> None:
         """
-        Args:
-            include_background: if False, channel index 0 (background category) is excluded from the calculation.
-                if the non-background segmentations are small compared to the total image size they can get overwhelmed
-                by the signal from the background so excluding it in such cases helps convergence.
-            to_onehot_y: whether to convert the ``target`` into the one-hot format,
-                using the number of classes inferred from `input` (``input.shape[1]``). Defaults to False.
-            sigmoid: if True, apply a sigmoid function to the prediction.
-            softmax: if True, apply a softmax function to the prediction.
-            other_act: callable function to execute other activation layers, Defaults to ``None``. for example:
-                ``other_act = torch.tanh``.
-            squared_pred: use squared versions of targets and predictions in the denominator or not.
-            jaccard: compute Jaccard Index (soft IoU) instead of dice or not.
-            reduction: {``"none"``, ``"mean"``, ``"sum"``}
-                Specifies the reduction to apply to the output. Defaults to ``"mean"``.
-
-                - ``"none"``: no reduction will be applied.
-                - ``"mean"``: the sum of the output will be divided by the number of elements in the output.
-                - ``"sum"``: the output will be summed.
-
-            smooth_nr: a small constant added to the numerator to avoid zero.
-            smooth_dr: a small constant added to the denominator to avoid nan.
-            batch: whether to sum the intersection and union areas over the batch dimension before the dividing.
-                Defaults to False, a Dice loss value is computed independently from each item in the batch
-                before any `reduction`.
-            weight: weights to apply to the voxels of each class. If None no weights are applied.
-                The input can be a single value (same weight for all classes), a sequence of values (the length
-                of the sequence should be the same as the number of classes. If not ``include_background``,
-                the number of classes should not include the background category class 0).
-                The value/values should be no less than 0. Defaults to None.
-            soft_label: whether the target contains non-binary values (soft labels) or not.
-                If True a soft label formulation of the loss will be used.
-
-        Raises:
-            TypeError: When ``other_act`` is not an ``Optional[Callable]``.
-            ValueError: When more than 1 of [``sigmoid=True``, ``softmax=True``, ``other_act is not None``].
-                Incompatible values.
-
+        Args follow standard MONAI DiceLoss with the addition of:
+            ignore_index: Specifies a target value that is ignored and does not contribute to the input gradient.
         """
         super().__init__(reduction=LossReduction(reduction).value)
         if other_act is not None and not callable(other_act):
@@ -126,29 +92,13 @@ def __init__(
         self.register_buffer("class_weight", weight)
         self.class_weight: None | torch.Tensor
         self.soft_label = soft_label
+        self.ignore_index = ignore_index
 
     def forward(self, input: torch.Tensor, target: torch.Tensor) -> torch.Tensor:
         """
         Args:
             input: the shape should be BNH[WD], where N is the number of classes.
             target: the shape should be BNH[WD] or B1H[WD], where N is the number of classes.
-
-        Raises:
-            AssertionError: When input and target (after one hot transform if set)
-                have different shapes.
-            ValueError: When ``self.reduction`` is not one of ["mean", "sum", "none"].
-
-        Example:
-            >>> from monai.losses.dice import *  # NOQA
-            >>> import torch
-            >>> from monai.losses.dice import DiceLoss
-            >>> B, C, H, W = 7, 5, 3, 2
-            >>> input = torch.rand(B, C, H, W)
-            >>> target_idx = torch.randint(low=0, high=C - 1, size=(B, H, W)).long()
-            >>> target = one_hot(target_idx[:, None, ...], num_classes=C)
-            >>> self = DiceLoss(reduction='none')
-            >>> loss = self(input, target)
-            >>> assert np.broadcast_shapes(loss.shape, input.shape) == input.shape
         """
         if self.sigmoid:
             input = torch.sigmoid(input)
@@ -163,27 +113,42 @@ def forward(self, input: torch.Tensor, target: torch.Tensor) -> torch.Tensor:
         if self.other_act is not None:
             input = self.other_act(input)
 
+        # Create valid mask if ignore_index is specified and target is in index format
+        valid_mask = None
+        if self.ignore_index is not None and target.shape[1] == 1:
+            valid_mask = (target != self.ignore_index).to(input.dtype)
+
         if self.to_onehot_y:
             if n_pred_ch == 1:
                 warnings.warn("single channel prediction, `to_onehot_y=True` ignored.", stacklevel=2)
             else:
                 target = one_hot(target, num_classes=n_pred_ch)
 
+        # Create valid mask if target was already one-hot but ignore_index channel is specified
+        if self.ignore_index is not None and valid_mask is None:
+            if 0 <= self.ignore_index < target.shape[1]:
+                valid_mask = torch.ones_like(target)
+                valid_mask[:, self.ignore_index] = 0.0
+
         if not self.include_background:
             if n_pred_ch == 1:
                 warnings.warn("single channel prediction, `include_background=False` ignored.", stacklevel=2)
             else:
-                # if skipping background, removing first channel
                 target = target[:, 1:]
                 input = input[:, 1:]
+                if valid_mask is not None and valid_mask.shape[1] == n_pred_ch:
+                    valid_mask = valid_mask[:, 1:]
 
         if target.shape != input.shape:
             raise AssertionError(f"ground truth has different shape ({target.shape}) from input ({input.shape})")
 
-        # reducing only spatial dimensions (not batch nor channels)
+        # Apply mask to both predictions and targets to exclude ignored regions
+        if valid_mask is not None:
+            input = input * valid_mask
+            target = target * valid_mask
+
         reduce_axis: list[int] = torch.arange(2, len(input.shape)).tolist()
         if self.batch:
-            # reducing spatial dimensions and batch
             reduce_axis = [0] + reduce_axis
 
         ord = 2 if self.squared_pred else 1
@@ -198,7 +163,6 @@ def forward(self, input: torch.Tensor, target: torch.Tensor) -> torch.Tensor:
 
         num_of_classes = target.shape[1]
         if self.class_weight is not None and num_of_classes != 1:
-            # make sure the lengths of weights are equal to the number of classes
             if self.class_weight.ndim == 0:
                 self.class_weight = torch.as_tensor([self.class_weight] * num_of_classes)
             else:
@@ -209,16 +173,13 @@ def forward(self, input: torch.Tensor, target: torch.Tensor) -> torch.Tensor:
                     )
             if self.class_weight.min() < 0:
                 raise ValueError("the value/values of the `weight` should be no less than 0.")
-            # apply class_weight to loss
             f = f * self.class_weight.to(f)
 
         if self.reduction == LossReduction.MEAN.value:
-            f = torch.mean(f)  # the batch and channel average
+            f = torch.mean(f)
         elif self.reduction == LossReduction.SUM.value:
-            f = torch.sum(f)  # sum over the batch and channel dims
+            f = torch.sum(f)
         elif self.reduction == LossReduction.NONE.value:
-            # If we are not computing voxelwise loss components at least
-            # make sure a none reduction maintains a broadcastable shape
             broadcast_shape = list(f.shape[0:2]) + [1] * (len(input.shape) - 2)
             f = f.view(broadcast_shape)
         else: