Spectral Phasor Explorer
The Spectral Phasor Explorer transforms each hyperspectral pixel spectrum into a point in two-dimensional phasor space.
Overview
The Spectral Phasor Explorer transforms each hyperspectral pixel spectrum into a point in two-dimensional phasor space.
Pixels with similar spectral shapes tend to occupy similar regions of the phasor plot.
The phasor representation provides a model-free way to explore spectral populations and map them back to their spatial locations in the hyperspectral image.
1. Select Original or Subtraction mode.
2. Choose the Reconstruction Wavelength Range if needed.
4. Inspect the phasor scatter.
5. Select an ROI in phasor space to reconstruct corresponding image pixels.
6. Or select an ROI in image space to highlight corresponding phasor points.
7. Export the phasor plot, coordinates, reconstruction image, or mask as needed.
For every image pixel, the hyperspectral spectrum is transformed into two coordinates:
These coordinates are calculated using sinusoidal functions across the spectral channels.
The result is a compact two-dimensional representation of the spectral shape.
The Phasor Space panel displays one point for each hyperspectral image pixel.
Pixels with similar spectral characteristics tend to appear near one another.
Groups or clusters in phasor space may therefore correspond to different materials, tissues, spectral states, or mixed spectral populations.
The phasor plot includes a unit-circle reference.
The exact location of a spectral population within the circle depends on the spectral distribution of intensity across the hyperspectral channels.
The phasor coordinates are analytical coordinates and should not be interpreted directly as wavelength or intensity.
Original mode computes the phasor transform directly from the hyperspectral cube.
Use this mode as the standard starting point.
Subtraction mode subtracts one selected spectral band from all spectral bands before calculating the phasor transform.
The Subtract reference band control determines which band is used as the reference.
This can help emphasize spectral differences relative to a selected baseline band.
Interpretation should consider that subtraction changes the spectral representation before the phasor calculation.
Click Compute Phasor to calculate the spectral phasor coordinates for all image pixels.
The resulting scatter is displayed in Phasor Space.
The Spatial Reconstruction panel displays the mean image across the selected reconstruction band range.
Reconstruction Display Range
The Min and Max band controls determine which spectral bands are averaged to create the background reconstruction image.
These controls affect visualization only.
They do not change the phasor coordinates after the phasor has been calculated.
Click Select ROI in Phasor and draw a polygon around a population of phasor points.
IDCubePro identifies all pixels whose phasor coordinates fall inside the selected polygon.
The corresponding pixels are then mapped back to their original spatial locations in the hyperspectral image.
This provides direct spectral-population-to-image reconstruction.
Click Select ROI in Image and draw a rectangular region on the reconstruction image.
The corresponding pixels are identified in the hyperspectral image.
Their phasor coordinates are then highlighted in the Phasor Space panel.
This provides the reverse mapping from a known spatial region to its spectral phasor distribution.
The most useful feature of the Spectral Phasor Explorer is the bidirectional relationship between phasor space and image space.
Phasor ROI -> spatial reconstruction Image ROI -> highlighted phasor population This allows spectral and spatial information to be explored together.
When Overlay selection on original image is enabled, the selected phasor population is displayed as an overlay on the reconstruction image.
Disabling the overlay can be useful when only the selected mask is of interest.
Clear Selection removes the current phasor and image ROI selections.
The calculated phasor coordinates are retained.
You can therefore immediately create a new selection without recomputing the phasor transform.
Export Plot saves the current phasor axes as an image or PDF.
Export Coordinates saves the numerical phasor coordinates for all image pixels.
- Phasor imaginary coordinate
- Whether the pixel belongs to the current selection
CSV, Excel, and MATLAB export formats are supported.
Export Image saves the current reconstruction display.
Export Mask saves the selected spatial population as a binary mask.
Selected pixels are stored as foreground and unselected pixels as background.
PNG, TIFF, and MATLAB formats are supported.
The spectral phasor is a dimensionality-reduction and visualization method.
Proximity in phasor space indicates similarity according to the phasor representation.
It does not automatically identify the chemical, biological, or material identity of a population.
Interpretation should be supported by spectral references, experimental controls, known spectral features, or independent measurements when appropriate.
Pixels containing mixtures of multiple spectral components may appear between phasor populations.
Continuous trajectories or elongated clusters in phasor space can sometimes indicate spectral mixing or gradual transitions.
Phasor results depend on the spectral data supplied to the tool.
Appropriate preprocessing may include:
- Removal of noisy wavelengths
- Normalization when scientifically appropriate
Use preprocessing suitable for the imaging system and scientific objective.
1. Load and preprocess the hyperspectral dataset.
2. Open Spectral Phasor Explorer.
3. Start with Original mode.
5. Inspect the distribution of phasor points.
6. Select a phasor population and inspect its spatial reconstruction.
7. Select representative image regions and inspect their phasor distributions.
8. Repeat ROI selection to explore spectral populations.
9. Use Subtraction mode when comparison relative to a reference band is useful.
10. Export coordinates and masks for downstream analysis if needed.
Phasor points appear tightly compressed The hyperspectral spectra may be highly similar or dominated by common background/intensity structure.
Many points appear outside the expected distribution Inspect the dataset for noisy pixels, saturated regions, invalid values, or unusual preprocessing.
Spatial reconstruction appears noisy The selected phasor ROI may contain heterogeneous or low-signal populations.
The selected phasor region maps to unexpected image locations The selected phasor population may represent a shared spectral pattern present in multiple spatial regions.
Image ROI produces a broad phasor population The spatial ROI may contain multiple spectral populations or mixed pixels.
Subtraction mode produces very different results This is expected because subtraction changes every spectrum before the phasor transformation.
Important
Spectral phasor analysis is an exploratory method.
Clustering or separation in phasor space should not be interpreted as definitive identification without independent scientific validation.