Quantifying in vivo MR spectra with circles.

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Spectroscopy

Background:

• Accurate quantification of in vivo magnetic resonance spectroscopy (MRS) data is crucial for medical and research applications.
• Existing MRS analysis methods struggle with low signal-to-noise ratios, overlapping peaks, and artifacts common in in vivo studies.

Purpose of the Study:

• To introduce and evaluate a novel frequency-domain technique, CFIT, for improved in vivo MRS data analysis.
• To address limitations of current methods in handling noisy and artifact-laden MRS data.

Main Methods:

• Developed a new frequency-domain technique (CFIT) that fits circular trajectories of spectral peaks using active circle models.
• Incorporated prior knowledge and baseline artifact correction using active contour strategies (snakes).
• Compared CFIT's stability and accuracy against a standard time-domain fitting tool and a frequency-domain absorption-mode technique using simulated and real human 31P MRS data.

Main Results:

• CFIT demonstrated the lowest fitting failures on real human chest and heart 31P MRS data.
• CFIT achieved accuracy comparable to the standard frequency-domain method, both outperforming the time-domain approach.
• Simulations suggested that Cramer-Rao Bounds may not be a reliable performance indicator for in vivo MRS data.

Conclusions:

• CFIT is a stable and accurate method for analyzing in vivo magnetic resonance spectroscopy data.
• The technique effectively handles phasing and baseline artifacts, offering a robust alternative to existing methods.
• CFIT shows significant promise for advancing research and clinical applications of in vivo MRS.