Addressing phase errors in fat-water imaging using a mixed magnitude/complex fitting method.

Area of Science:

• Medical Imaging
• Quantitative MRI
• Liver Disease Diagnosis

Background:

• Accurate, noninvasive liver fat measurements are crucial for diagnosing and staging nonalcoholic fatty liver disease (NAFLD).
• Current chemical shift-based fat quantification methods are susceptible to phase errors, particularly from eddy currents, biasing results.
• Existing solutions like magnitude fitting reduce signal-to-noise ratio (SNR), compromising image quality.

Purpose of the Study:

• To develop a novel reconstruction method for accurate liver fat quantification.
• To overcome phase errors in multiecho gradient echo sequences without SNR penalty.
• To improve the reliability of fat fraction measurements in NAFLD.

Main Methods:

• A new reconstruction technique was developed, discarding phase from the first echo while preserving phase from subsequent echoes.
• The method was tested on 104 liver datasets from 52 patients, with comparisons to spectroscopy measurements.
• This 'mixed fitting' approach addresses phase errors without the SNR degradation of magnitude fitting.

Main Results:

• The proposed mixed fitting method accurately measures liver fat fraction.
• High signal-to-noise ratio was maintained across various echo time combinations.
• Low bias was observed in fat fraction measurements compared to spectroscopy.

Conclusions:

• The developed reconstruction method provides accurate and reliable liver fat quantification.
• This technique offers an improved approach for diagnosing and staging NAFLD noninvasively.
• The method successfully overcomes phase errors while preserving SNR, enhancing diagnostic utility.