Effect of noise on MTF calculations using different phantoms

Area of Science:

• Medical Imaging Physics
• Image Quality Assessment
• Quantitative Imaging

Background:

• The Modulation Transfer Function (MTF) is a critical metric for evaluating the spatial resolution and image quality of imaging systems.
• Noise is an inherent factor in medical imaging that can significantly affect the accuracy and precision of MTF calculations.
• Different phantom designs and analysis techniques exist for MTF assessment, each with potential susceptibility to noise.

Purpose of the Study:

• To investigate the impact of image noise on the precision of Modulation Transfer Function (MTF) calculations.
• To compare the performance of three distinct MTF calculation techniques (edge, bar, and star bar phantoms) under varying noise conditions.
• To derive theoretical expressions for MTF standard deviation and signal-to-noise ratio (SNR) and validate them with Monte Carlo simulations.

Main Methods:

• Developed theoretical expressions for the standard deviation and SNR of MTF for edge, bar, and star bar phantoms.
• Utilized Monte Carlo simulations to generate artificial phantom images mimicking real-world sampling, blurring, and noise characteristics.
• Applied oversampling techniques to obtain presampled MTFs and analyzed results across different noise levels and frequencies.

Main Results:

• MTF standard deviation is directly proportional to image noise across all techniques.
• The bar phantom technique demonstrates superior precision compared to edge and star bar techniques across all tested frequencies and noise levels.
• Edge technique exhibits significant bias errors at higher frequencies under high noise conditions, while star bar performance varies with system isotropy.

Conclusions:

• The bar phantom technique provides more precise MTF calculations than the edge technique, with differences increasing at higher frequencies.
• High noise levels in the edge technique lead to poor SNR in the Line Spread Function (LSF), increasing MTF erroneously and reducing accuracy.
• The bar technique is recommended for robust MTF assessment, particularly in noisy imaging environments, due to its superior precision and accuracy.