Noise removal using factor analysis of dynamic structures: application to cardiac gated studies

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Factor Analysis of Dynamic Structures (FADS) extracts meaningful physiologic data from dynamic image sets.
• The inverse process (iFADS) aims to reconstruct original data by excluding residual noise.
• Assessing FADS-iFADS efficiency on gated cardiac images is crucial for improving diagnostic accuracy.

Purpose of the Study:

• To quantitate and assess the efficiency of the FADS-iFADS method for noise reduction in gated cardiac images.
• To evaluate the impact of FADS-iFADS on image quality and signal-to-noise ratio (SNR).
• To compare FADS-iFADS performance against traditional smoothing and frequency filters.

Main Methods:

• Computer simulation of planar cardiac gated studies with added noise.
• Application of the FADS-iFADS program to simulated and patient data.
• Comparison of SNRs between original and processed data.
• Analysis of subtracted data to evaluate residual noise characteristics.

Main Results:

• The FADS-iFADS process resulted in an approximate fivefold increase in SNR.
• The difference between original and processed data consisted solely of white noise.
• Simulations and patient data demonstrated significant noise reduction.

Conclusions:

• FADS-iFADS effectively removes a substantial portion of noise, enhancing cardiac image quality.
• The method preserves spatial resolution and details, unlike smoothing or frequency filters.
• FADS-iFADS is a robust, non-operator-dependent tool for improving cardiac imaging once the number of factors is determined.