This dataset accompanies the study: "Bone-marrow oedema estimation with a microwave imaging OECP array in ex-vivo bovine long bones: comparison with CT ground truth"
Objective: Evaluation of a 16-element open-ended coaxial probe (OECP) microwave imaging array for non-ionising estimation of bone-marrow edema in ex-vivo bovine samples, validated against CT-based measurements.
The dataset contains microwave imaging (MWI), computed tomography (CT), and quantitative analysis results for seven ex-vivo bovine bone samples (A-G) from a single 13-month-old animal.
Samples:
- 4 femur halves (distal and proximal from two femurs)
- 3 humerus halves (from two humeri)
Experimental Design:
- Simulated bone-marrow edema via precision-drilled holes filled with iodine contrast/gel mixture
- Pre-injection and post-injection acquisition protocol
- Ground truth validation through differential weighing and CT imaging
| Specimen | Sample Type | Drilled Holes | Hole Diameter | Hole Depth | Nominal Volume per Hole |
|---|---|---|---|---|---|
| A | Femur | 16 | 2 mm | 8 mm | 25.13 µL |
| B | Femur | 16 | 2 mm | 8 mm | 25.13 µL |
| C | Femur | 25 | 2 mm | 8 mm | 25.13 µL |
| D | Femur | 25 | 2 mm | 8 mm | 25.13 µL |
| E | Humerus | 20 | 2 mm | 8 mm | 25.13 µL |
| F | Humerus | 20 | 2 mm | 8 mm | 25.13 µL |
| G | Humerus | 16 | 2 mm | 8 mm | 25.13 µL |
Hole Pattern: 4 mm center-to-center spacing in a regular matrix pattern.
- Array Configuration: 4×4 OECP array (16 probes)
- Frequency Range: 2.25-3.00 GHz
- Target Frequency: 2.5 GHz (analysis)
- Nominal Power: 0 dBm
- Calibration: OSL (Open-Short-Load) with air, aluminum foil, deionized water (22°C), and propylene glycol (22°C)
- Spatial Grid: 16×16 effective measurements (array scanned over 16 positions with 4 rotations per position)
- Acquisition Time: ~70 minutes per session (pre or post)
- Raw Data: S₁₁ scattering parameters
- Scanner: GE Medical Systems Revolution Ascend
- Parameters: 100 kVp, 80 mA
- Reconstruction: BONE kernel
- Slice Thickness: 3.0 mm
- Pixel Size: 0.94 mm
- CT Positivity Criterion: ≥50% voxels ≥250 HU
GALEAZZI DATA/
├── summary-quantitative-results.csv
├── README.md
├── A/
│ ├── ct-scan-pre-injection.png
│ ├── ct-scan-post-injection.png
│ ├── dielectric-permittivity-matrix-pre-injection.csv
│ ├── dielectric-permittivity-matrix-post-injection.csv
│ ├── fluid-content-matrix-post-injection.csv
│ ├── dielectric-permittivity-image-A-pre-injection.png
│ ├── dielectric-permittivity-image-A-post-injection.png
│ ├── dielectric-permittivity-fusion-image-A-pre-injection.png
│ ├── dielectric-permittivity-fusion-image-A-pre-injection-with-values.png
│ ├── dielectric-permittivity-fusion-image-A-post-injection.png
│ ├── dielectric-permittivity-fusion-image-A-post-injection-with-values.png
│ └── roi-coordinates.json
├── B/
│ └── [same structure as A]
├── C/
│ └── [same structure as A]
├── D/
│ └── [same structure as A]
├── E/
│ └── [same structure as A]
├── F/
│ └── [same structure as A]
└── G/
└── [same structure as A]
Each specimen folder contains the following files:
ct-scan-pre-injection.png- CT slice image before contrast injection (background removal applied)ct-scan-post-injection.png- CT slice image after contrast injection (background removal applied)
dielectric-permittivity-matrix-pre-injection.csv- 16×16 matrix of real relative permittivity values (εᵣ), pre-injection baselinedielectric-permittivity-matrix-post-injection.csv- 16×16 matrix of real relative permittivity values (εᵣ), post-injection
Processing: Derived from S₁₁ parameters via Stuchly open-ended coaxial probe inversion model. Spatial resolution enhanced through nonlinear sigmoidal interpolation based on probe sensitivity characterization.
fluid-content-matrix-post-injection.csv- 16×16 matrix of estimated aqueous fluid fraction (%)
Processing: Computed using CRIM (Complex Refractive Index Method) two-phase mixing model comparing post-injection permittivity against pre-injection baseline. Values weighted by probe penetration depth profile.
dielectric-permittivity-image-{X}-pre-injection.png- Standalone visualization of pre-injection permittivity distributiondielectric-permittivity-image-{X}-post-injection.png- Standalone visualization of post-injection permittivity distribution
dielectric-permittivity-fusion-image-{X}-pre-injection.png- CT-MWI fusion image, pre-injectiondielectric-permittivity-fusion-image-{X}-pre-injection-with-values.png- CT-MWI fusion with dielectric permittivity values annotated at grid pointsdielectric-permittivity-fusion-image-{X}-post-injection.png- CT-MWI fusion image, post-injectiondielectric-permittivity-fusion-image-{X}-post-injection-with-values.png- CT-MWI fusion with dielectric permittivity values annotated at grid points
Fusion Method: CT anatomical reference overlaid with MWI permittivity data for spatial co-registration.
Note: Images with "-with-values" suffix display the numerical dielectric permittivity values (εᵣ) from the corresponding permittivity matrix, mapped onto the fusion visualization at each grid point.
roi-coordinates.json- Rectangular subset coordinates defining the analysis region for fluid volume computation
summary-quantitative-results.csv- Aggregated volumetric measurements for all specimens
Columns:
Specimen- Sample identifier (A-G)Sample_Type- Anatomical origin (Femur/Humerus)Drilled_Holes- Number of contrast-injected holesMeasured_Injected_Volume_uL- Ground truth volume from differential weighing (µL)Measured_Injected_Uncertainty_uL- Measurement uncertainty (µL)Scaled_Injected_Volume_uL- Injected volume scaled to probe penetration depth (µL)CT_Estimated_Volume_uL- Volume estimated from CT positivity criterion (µL)CT_MWI_Scaled_Volume_uL- CT volume scaled to probe penetration depth (µL)MWI_Measured_Volume_uL- Volume estimated from MWI fluid content analysis (µL)
All CSV matrices are 16×16 grids representing the spatial distribution of measurements. Coordinate system:
- Origin: Top-left
- Row indexing: Top to bottom
- Column indexing: Left to right
- Missing values:
nan(outside measurement region) - Units:
- Permittivity matrices: Dimensionless (relative permittivity εᵣ)
- Fluid content matrix: Percentage (%)
ROI coordinates define rectangular analysis regions:
{
"row_start": <int>,
"col_start": <int>,
"height": <int>,
"width": <int>
}Coordinates reference the 16×16 grid indexing system (0-indexed).