Resample

The Resample tool changes the voxel dimensions and spacing of volume and mask data. Use resampling to upsample (increase resolution) or downsample (decrease resolution) images for specific processing requirements.

Accessing the Tool

1. Navigate to the Image tab in the ribbon.
2. Click Resample in the Transform section.

Concepts

Term	Description
Upsampling	Increasing the number of voxels (smaller voxel size)
Downsampling	Decreasing the number of voxels (larger voxel size)
Interpolation	Method for calculating new voxel values

Parameters

Interpolation Method

Select interpolation methods separately for volumes and masks:

Method	For volumes	For masks
Nearest	Fast, blocky results	Preserves label values exactly
Linear	Smooth gradients	Introduces intermediate values that are not valid labels
Cubic	Smoothest results	Not suitable for masks

The defaults are Linear for volumes and Nearest for masks, which is the right pairing for most work.

tip

Keep Nearest for masks. Any other method blends neighbouring labels and produces values that belong to no label, which turns a clean segmentation into a fringe of invalid voxels along every boundary.

Resampling Method

Choose the method to define target resolution:

Method	Description
Voxel dimensions (pixels)	Specify the target number of voxels in each direction
Pixel spacing (mm)	Specify the target voxel size in millimetres

Only the parameter group belonging to the selected method is shown.

Parameters (pixels)

Parameter	Description
X	Target number of voxels along X
Y	Target number of voxels along Y
Z	Target number of voxels along Z

Parameters (mm)

Parameter	Description
X	Target voxel size along X
Y	Target voxel size along Y
Z	Target voxel size along Z

Information

The Information group updates as you type, so the effect of a change is visible before you apply it:

Row	Description
Input physical dimensions (mm)	Size of the volume before resampling
Output physical dimensions (mm)	Size after resampling
Input pixel spacing (mm)	Voxel size before resampling
Output pixel spacing (mm)	Voxel size after resampling
Input voxel dimensions (pixels)	Voxel counts before resampling
Output voxel dimensions (pixels)	Voxel counts after resampling

The physical dimensions stay effectively the same through a resample: what changes is how many voxels cover that same physical space, and therefore how large each voxel is.

Workflow

1. Open the Resample tool from the Image tab.
2. Select the resample method (Voxel dimensions (pixels) or Pixel spacing (mm)).
3. Choose interpolation methods for volumes and masks.
4. Enter target dimensions or spacing values.
5. Review the information display to verify the resampling effect.
6. Click Apply to resample the data.

Use Cases

Reducing Memory Usage

Downsample large datasets for:

• Faster visualization and interaction
• Reduced processing time for preliminary analysis
• Lower memory consumption

Matching Resolution Between Datasets

Align resolution when combining datasets:

1. Identify the target resolution (usually the coarser dataset).
2. Resample finer datasets to match.
3. Enables direct comparison and fusion.

Preparing for Specific Algorithms

Some processing algorithms require specific resolutions:

• Neural network inputs may expect fixed dimensions
• Simulation meshes may have element size requirements
• Resample to meet these requirements

Isotropic Resampling

Convert anisotropic data (different spacing in each axis) to isotropic:

1. Note the smallest spacing value.
2. Set all three spacing values to this minimum.
3. Results in cubic voxels for uniform analysis.

Affected Objects

Resampling affects:

• All volume objects
• All mask objects

All datasets are resampled to the same grid to maintain spatial alignment.

Quality Considerations

Upsampling Limitations

info

Upsampling cannot recover detail that doesn't exist in the original data. Interpolation creates smooth transitions but does not add real information.

Downsampling Information Loss

warning

Downsampling permanently reduces detail. High-frequency features and fine structures may be lost. Consider keeping the original data if full resolution may be needed later.

Memory Impact

Resampling significantly affects memory usage:

$$\text{Memory ratio} = \frac{X_{new} \times Y_{new} \times Z_{new}}{X_{old} \times Y_{old} \times Z_{old}}$$

Example: Doubling resolution in all dimensions increases memory by 8×.

Scripting

This operation is available via the Python scripting API using resample().

import ScriptingApi as api

app = api.Application()
volume_operations = app.get_volume_operations()

volume_operations.resample(
    ["Volume1"],
    [256, 256, 128],              # Target dimensions [x, y, z]
    [0.5, 0.5, 1.0],              # Target spacing [x, y, z] in mm
    api.Interpolation.Linear,     # Volume interpolation: Nearest, Linear, or Cubic
    api.Interpolation.Nearest     # Mask interpolation: Nearest recommended for masks
)

See the VolumeOperations API Reference for details.