SU-E-I-96: Realistic Synthetic CT Imaging of Small Numerical Lesions with Simple CT Simulation Model Incorporating...

Area of Science:

• Medical Imaging
• Computational Phantoms
• Image Simulation

Background:

• Quantitative analysis in CT imaging necessitates ground truth data, which is often difficult to obtain.
• Realistic synthetic CT images are needed for training and validating image analysis algorithms.

Purpose of the Study:

• To present a simple CT simulation model for generating realistic synthetic CT images of numerical phantoms.
• To enable the creation of ground truth datasets for small lesions with varying shapes and sizes.

Main Methods:

• A CT simulator was developed using basic CT parameters from DICOM headers.
• Projector calibration involved measuring Noise Power Spectrum (NPS) and iterative refinement.
• Modulation Transfer Function (MTF) was incorporated to account for blurring and reconstruction kernels.
• Validation was performed by comparing simulated images with physically scanned phantoms.

Main Results:

• Synthesized CT images closely matched physically scanned images in visual assessment.
• Size measurements and line profile comparisons showed strong agreement between simulated and real images.
• The simulation accurately reflected various phantom shapes, sizes, and reconstruction kernels.

Conclusions:

• The proposed CT simulation method generates realistic synthetic CT images of numerical phantoms.
• These synthetic images exhibit high visual and quantitative similarity to real CT scans.
• The model has potential for use as a ground truth dataset in quantitative CT image analysis.