Fat Quantification in Dual-Layer Detector Spectral Computed Tomography: Experimental Development and First In-Patient...

Area of Science:

• Medical Imaging
• Radiology
• Computed Tomography

Background:

• Dual-energy computed tomography (DECT) offers objective, contrast-independent fat quantification for prognostic parameters like hepatic steatosis and muscle quality.
• Current DECT fat quantification is limited to source-based techniques (e.g., fast kVp-switching, dual-source CT), requiring prospective dual-energy mode selection.

Purpose of the Study:

• To develop a material decomposition algorithm for fat quantification using dual-layer detector-based spectral CT (dlsCT).
• To validate this algorithm in phantoms and in vivo for patient liver and skeletal muscle.
• To assess the utility of dlsCT for retrospective spectral information acquisition independent of imaging mode.

Main Methods:

• Phantoms with varying fat and iodine concentrations were scanned using dlsCT and compared with MR relaxometry (MRR) and MR spectroscopy (MRS).
• A 3-material decomposition algorithm was applied to dlsCT data to quantify fat, iodine, and phantom material.
• In vivo validation involved 10 patients undergoing contrast-enhanced abdominal dlsCT and MRR, with non-contrast dlsCT datasets used for reference.

Main Results:

• Excellent agreement was found between dlsCT and MR techniques for phantoms (ICC 0.96-0.98) and skeletal muscle (ICC 0.96).
• Moderate agreement was observed for liver fat quantification (ICC 0.75).
• Bland-Altman analysis showed mean differences of -0.7% for liver and 0.5% for skeletal muscle, with excellent inter- and intraobserver agreement.

Conclusions:

• Fat quantification using a material decomposition algorithm on dlsCT data has been successfully developed and validated.
• dlsCT demonstrates good agreement with MR techniques for fat quantification in patient liver and skeletal muscle.
• This technique allows for the detection of hepatic steatosis and myosteatosis in routine clinical scans, retrospectively providing spectral information.