Strategies for optimizing the phase correction algorithms in Nuclear Magnetic Resonance spectroscopy

Table 5 Results of the analysis of 27 brain phantom spectra obtained for two different phase correction algorithms: original Automics and Automics with proposed tunning applied.

Metabolite	not tuned				tuned				paired t-test p-values
	$\bar{x}$	s	95% CI	CV [%]	$\bar{x}$	s	95% CI	CV [%]
NAA	12.32	0.87	(12.25; 12.39)	7.48	12.51	0.24	(12.33; 12.69)	2.33	0.0023
Creatine	9.51	0.79	(9.45; 9.57)	8.36	10.04	0.09	(9.95; 10.13)	1.15	0.0048
Choline	2.87	0.14	(2.86; 2.88)	4.93	3.00	0.06	(2.85; 3.14)	1.89	0.0023
myo-Inositol	6.67	0.73	(6.61; 6.73)	10.81	7.49	0.06	(7.38; 7.59)	1.36	0.0048
Lactate	4.72	0.58	(4.67; 4.76)	12.07	5.02	0.12	(4.86; 5.17)	2.00	0.0023

The presented descriptive statistics are: sample mean ( $\bar{x}$ ), sample standard deviation (s), 95% confidence interval for population mean value (95%CI) and coefficient of variation (CV). Right column presents paired t test p-value resulting from testing the hypothesis on no accuracy improvement by algorithm tuning.

ISSN: 1475-925X