Legendre Polynomials: A Fully Automatic Method for Noise Reduction in ^99mTc-Mercaptoacetyltriglycine Renogram...

Area of Science:

• Nuclear Medicine
• Medical Imaging Analysis
• Biomedical Engineering

Background:

• Quantitative analysis of dynamic renogram images is often hindered by high noise levels.
• Accurate assessment of renal function relies on precise time-activity curve (TAC) analysis.

Purpose of the Study:

• To develop a fully automatic noise-removal method for dynamic mercaptoacetyltriglycine (99mTc-MAG3) renogram images.
• To evaluate the performance of the proposed Legendre polynomial method against existing filtering techniques.

Main Methods:

• A novel method utilizing Legendre polynomials was developed to fit and filter TACs.
• The method was validated on both Monte Carlo (MC)-simulated renal studies and real patient data.
• Performance was compared to the 1-2-1 filter (F121) and commercial Hermes software for parameters like relative function, Tmax, and T1/2.

Main Results:

• The Legendre method demonstrated approximately 2 times lower root mean squared error compared to F121 on noisy TACs.
• Statistically equivalent left relative function was achieved across all methods (Hermes, Legendre, F121).
• Legendre polynomials significantly improved Tmax and T1/2 determination in MC studies and showed greater robustness against noise for T1/2 in patient data compared to Hermes.

Conclusions:

• Legendre polynomials offer a promising, fully automatic, and robust tool for noise reduction in dynamic nuclear medicine imaging.
• The method is accurate and suitable for routine clinical application in renal imaging analysis.