docs: address CodeRabbit review on #8826

- vista3d/inferer.py: clarify spatial-axis indices for [N,C,W,H,D] tensor
  (axes 2-4, not 0-2) in point_based_window_inferer inputs docstring.
- visualize/utils.py: fix matshow3d docstring typos ('dim and reshape' -> 'then
  reshaped'; 'dimensionas' -> 'dimension as').

Signed-off-by: Soumya Snigdha Kundu <soumya_snigdha.kundu@kcl.ac.uk>

Author: aymuos15

monai/apps/vista3d/inferer.py

@@ -45,9 +45,10 @@ def point_based_window_inferer(
     patch inference and average output stitching, and finally returns the segmented mask.
 
     Args:
-        inputs: [1CWHD], input image to be processed (axis 0 is columns/Width,
-            axis 1 is rows/Height, axis 2 is Depth, matching arrays returned by
-            MONAI's NIfTI/ITK readers).
+        inputs: [1CWHD], input image to be processed (spatial axes are in
+            Width, Height, Depth order; i.e., for [N, C, W, H, D], Width is
+            axis 2, Height is axis 3, Depth is axis 4, matching arrays
+            returned by MONAI's NIfTI/ITK readers).
         roi_size: the spatial window size for inferences.
             When its components have None or non-positives, the corresponding inputs dimension will be used.
             if the components of the `roi_size` are non-positive values, the transform will use the

monai/visualize/utils.py

@@ -65,10 +65,10 @@ def matshow3d(
         figsize: size of the figure.
         frames_per_row: number of frames to display in each row. If None, sqrt(firstdim) will be used.
         frame_dim: for higher dimensional arrays, which dimension from (`-1`, `-2`, `-3`) is moved to
-            the `-3` dimension. dim and reshape to (-1, spatial0, spatial1) shape to construct frames,
+            the `-3` dimension, then reshaped to (-1, spatial0, spatial1) to construct frames,
             default to `-3`.
         channel_dim: if not None, explicitly specify the channel dimension to be transposed to the
-            last dimensionas shape (-1, spatial0, spatial1, C). this can be used to plot RGB color image.
+            last dimension as shape (-1, spatial0, spatial1, C). this can be used to plot RGB color image.
             if None, the channel dimension will be flattened with `frame_dim` and `batch_dim` as
             shape (-1, spatial0, spatial1).
             note that it can only support 3D input image. default is None.