Proton Radiography
Proton radiography/CT uses high-energy proton beams (100-250 MeV) to image the relative stopping power (RSP) of tissue, which is the quantity directly needed for proton therapy treatment planning. Unlike X-rays which measure attenuation, proton imaging measures the energy loss and scattering of individual protons as they traverse the object. Each proton's entry/exit position and angle are tracked, and the residual energy is measured. The RSP is reconstructed from many proton histories using iterative algorithms. Challenges include multiple Coulomb scattering (which blurs the spatial resolution to ~1 mm) and the need for single-proton tracking at high rates.
Energy Loss Scattering
Gaussian
filtered back projection
PARTICLE_TRACKER
Forward-Model Signal Chain
Each primitive represents a physical operation in the measurement process. Arrows show signal flow left to right.
Π(proton) → D(g, η₁)
Benchmark Variants & Leaderboards
Proton Radiography
Proton Radiography
Π(proton) → D(g, η₁)
Standard Leaderboard (Top 10)
| # | Method | Score | PSNR (dB) | SSIM | Trust | Source |
|---|---|---|---|---|---|---|
| 🥇 | pCT-Former | 0.760 | 33.0 | 0.920 | ✓ Certified | Proton CT transformer, 2024 |
| 🥈 | ProtonNet | 0.712 | 31.0 | 0.890 | ✓ Certified | Proton CT DL, 2022 |
| 🥉 | DROP-TVS | 0.595 | 27.0 | 0.790 | ✓ Certified | Penfold et al., 2010 |
| 4 | FBP-MLP | 0.467 | 23.5 | 0.650 | ✓ Certified | Schulte et al., 2008 |
Mismatch Parameters (3) click to expand
| Name | Symbol | Description | Nominal | Perturbed |
|---|---|---|---|---|
| energy_loss | ΔS | Stopping power error (%) | 0 | 2.0 |
| scattering | Δθ_MCS | Multiple Coulomb scattering error (%) | 0 | 5.0 |
| range_straggling | ΔR | Range straggling error (%) | 0 | 3.0 |
Reconstruction Triad Diagnostics
The three diagnostic gates (G1, G2, G3) characterize how reconstruction quality degrades under different error sources. Each bar shows the relative attribution.
Model: energy loss scattering — Mismatch modes: multiple coulomb scattering, nuclear interactions, tracker resolution, energy straggling
Noise: gaussian — Typical SNR: 10.0–25.0 dB
Requires: beam energy, tracker alignment, energy detector calibration, water equivalent calibration
Modality Deep Dive
Principle
Proton radiography images the transmission and scattering of high-energy protons (50-800 MeV) through dense objects. Unlike X-rays, protons undergo significant multiple Coulomb scattering (MCS) in matter, which provides density and compositional contrast. Both transmission (energy loss) and scattering angle measurements contribute to image formation. Proton radiography can penetrate very dense materials (steel, depleted uranium) that are opaque to X-rays.
How to Build the System
Requires a high-energy proton accelerator facility (synchrotron or cyclotron delivering 200-800 MeV protons). The object is placed in the beam path between tracking detectors (silicon strip or GEM detectors) that measure each proton's position and angle before and after the object. A magnetic spectrometer (quadrupole lens system, e.g., at LANL pRad facility) focuses transmitted protons onto a scintillator + camera detector.
Common Reconstruction Algorithms
- Most Likely Path (MLP) estimation for proton CT reconstruction
- Filtered back-projection with scattering-angle weighting
- Algebraic reconstruction (ART) with MCS forward model
- Material discrimination from dual-parameter (transmission + scattering) analysis
- Deep-learning proton CT reconstruction for reduced view angles
Common Mistakes
- Ignoring multiple Coulomb scattering in the reconstruction model, causing blur
- Nuclear interaction losses (protons stopped or scattered out of detector acceptance)
- Insufficient proton statistics leading to noisy images
- Energy straggling not modeled, causing depth-of-field blur in radiography
- Detector alignment errors between upstream and downstream tracking systems
How to Avoid Mistakes
- Use MLP or cubic spline path estimation in iterative reconstruction algorithms
- Account for nuclear interaction losses in the forward model; filter outlier tracks
- Accumulate sufficient proton histories (>10⁶ for radiography, >10⁸ for proton CT)
- Include energy straggling in the forward model or use higher energy protons to reduce it
- Carefully align tracking detectors with survey or use track-based alignment algorithms
Forward-Model Mismatch Cases
- The widefield fallback applies Gaussian blur, but proton radiography measures energy loss and multiple Coulomb scattering (MCS) of high-energy protons traversing the object — the scattering angle distribution encodes areal density, not spatial blur
- Protons lose energy continuously (Bethe-Bloch formula: -dE/dx ~ Z/A * z^2/beta^2) and scatter via Coulomb interaction — the measurement combines transmission intensity, residual energy, and scattering angle, none of which are modeled by optical blur
How to Correct the Mismatch
- Use the proton radiography operator that models energy-dependent proton transport: energy loss via Bethe-Bloch stopping power and angular broadening via Highland MCS formula (theta_rms ~ 13.6 MeV/(p*v) * sqrt(t/X_0))
- Reconstruct water-equivalent path length (WEPL) maps from residual energy measurements, or use scattering radiography for material discrimination — essential for proton therapy treatment planning
Experimental Setup
Phase-II proton CT prototype (Loma Linda / NIU)
200
scintillating fiber tracker + residual energy calorimeter
4
256x256
360
1.0
proton therapy treatment planning verification
Signal Chain Diagram
Key References
- Schulte et al., 'Conceptual design of a proton computed tomography system for applications in proton radiation therapy', IEEE Trans. Nucl. Sci. 51, 866-872 (2004)
Canonical Datasets
- Simulated proton CT phantoms (Penfold et al.)