Obesity is well-known as a major current health problem being a risk factor for a variety of diseases including hypertension, myocardial infarction, stroke, joint disorders and hepatitis. In this context, the visceral fat of the abdominal cavity is of particular relevance. Standard methods used in clinical routine such as the body mass index, however, do not specifically reflect fat distribution and in particular, do not distinguish between visceral and subcutaneous fat. Magnetic resonance imaging is a powerful tool to separate water and fat. In abdominal imaging, however, respiratory movement often requires breath-holding which can be difficult to perform by patients and might be even impossible in infants and experimental animals. Here we took advantage of the less motion-sensitive radial encoding. The aim of the study was to investigate whether this method can be used to separate abdominal fat and water in anesthetized mice at 7 T without any gaiting technique. Fat and water images were calculated from gradient echo datasets obtain at 3 different echo times using the three-point Dixon (3PD) [1] method and IDEAL [2]. The results of these two separation methods were compared regarding image quality.

Healthy female mice (n = 9) were anesthetized with isoflurane (1.5-2% in ambient air) and intubated via an endotracheal tube. Measurements were performed every four weeks over a period of three months (total number of measurements = 27).

A single echo spoiled radial FLASH sequence (TR = 600, flip angle = 15°, 157 radial spokes, FOV = 20 × 20 mm², 100 transversal slices, spatial resolution = 0.2 × 0.2 × 0.3 mm³, 13 averages) was used to obtain gradient echo data sets with an echo time of 2.0, 2.5 and 3.0 ms, respectively. All measurements were performed at a magnetic field strength of 7 T (ClinScan^TM, Bruker BioSpin, Germany) using a whole mouse body coil (Bruker BioSpin, Germany) for excitation and reception. The total measurement time was about one hour. The raw data was phase corrected to compensate for residual eddy-current effects on the high-field system and reconstructed using a gridding routine. For fat-water separation three-point Dixon (3PD) and IDEAL were applied on the reconstructed images. Separation of visceral and subcutaneous fat was done semi-automatically using the software package AMIRA (FEI^TM, USA).

As exemplary shown in Fig. 1, the reconstructed images were free of significant motion artefacts. The first and third images shown here were obtained at in-phase condition, whereas the second image was acquired at opposed-phase condition. Fig. 2 shows the resulting fat and water images using 3PD and IDEAL, respectively. Compared to IDEAL, the 3PD method yielded a clearly superior fat-water contrast. However, in areas with strong B0 inhomogeneity such as air-filled intestinal loops both methods showed their limitations. The corresponding B0 map, generated by IDEAL, is shown in Fig. 3.

Occasionally, as demonstrated in Fig. 4 by another slice position of the same animal, severe image artifacts were observed on fat images generated by IDEAL, whereas the fat-water separation achieved by 3PD may be still acceptable. The B0 map of those regions showed about five-fold higher magnetic field inhomogeneities compared to artefact-free images.

Beside the mentioned limitations, when using 3PD, a sufficient segmentation of the visceral fat was possible in all analyzed mice as shown in Fig. 5 where the visceral fat of the lumbar region is visualized three-dimensionally.

Radial encoding enabled the acquisition of images of the entire mouse abdomen with high spatial resolution and without significant motion artefacts. No gating techniques were required. The achieved image quality at preclinical high field MRI was sufficient to separate fat and water. Thereby 3PD seemed to be more robust compared to IDEAL, particularly in the presence of magnetic field inhomogeneities which are common at high field systems.Abdominal water and fat separation without any gating technique in the mice is possible with radial encoding. To the best of our knowledge this is the first report of separating visceral and subcutaneous fat in a mouse model. This method will be used in future preclinical follow-up studies in mouse models of obesity and in other species as well.