From Projections to Images: Complete CT Reconstruction & Dose-Noise Simulation in Python

Complete CT Reconstruction & Noise Simulation Toolkit

This package provides a full, end-to-end implementation of CT image formation — from projection generation to filtered backprojection reconstruction and realistic photon noise modeling.

Built entirely in Python, this toolkit allows you to understand not only how CT images are reconstructed, but why noise and dose affect image quality the way they do.

🔹 Part 1 – Fan-Beam Projection Generation

• Shepp–Logan phantom simulation
• Curved-detector fan-beam geometry
• Adjustable 2D acquisition parameters:
• Source-axis distance (default 570 mm)
• Detector arc (32.5° with 512 bins)
• 720 projections over 360°
• Full projection simulation code

🔹 Part 2 – Filtered Backprojection (FBP)

• Ramp filter implementation
• Explicit backprojection loop
• Multiple scan modes:
• Full 360° scan
• Short scan with Parker weighting
• Overscan with generalized Parker weighting

Experiment with geometry and scan parameters to see their direct impact on reconstruction quality.

🔹 Part 3 – Photon Noise & Dose Simulation

• Poisson photon statistics modeling
• Dose scaling (I₀ variation)
• Log-domain noise propagation
• High-dose vs low-dose reconstruction comparison

See how noise originates in the photon domain — and how it propagates into the reconstructed image.

🔬 What This Bundle Demonstrates

Phantom → Projection → Photon statistics → Log transform → Redundancy weighting → Reconstruction → Noise behavior

A complete CT physics pipeline.

📂 What’s Included

• Python scripts for fan-beam projection simulation
• Python implementation of FBP (full, short, overscan modes)
• Noise modeling and dose comparison scripts
• Fully adjustable acquisition parameters

⚡ Roadmap

Current version supports curved detectors.

Flat-panel geometry support will be added in a future update — free for existing buyers.