To study longitudinal changes of fat fraction and relaxation in brown adipose tissues in rats by multi-modal imaging, and validated by histology and UCP1.There are two types of fat tissues, white adipose tissue (WAT) and brown adipose tissue (BAT), which essentially perform opposite functions in energy metabolism.¹ There is a large interest in exploring brown adipose tissue as a therapeutic target for obesity and diabetes.² Cold exposure activates adrenergic receptors in brown adipose tissues and improves the separation of WAT and BAT.^{3, 4} In this study we have evaluated the longitudinal changes of fat fraction (FF) and relaxation in BAT, and validated with histology and UCP1.Seven, eleven and fifteen weeks old male Wistar rats (n = 6) were maintained at either thermoneutral body temperature (36 °C ± 0.5 °C) or short term cold exposure (26 ± 0.5 °C) prior to imaging. MRI experiments were performed using a 7T (ClinScan, Bruker, Germany) with a volume transmit coil and surface coil. Dixon imaging was performed on twenty slices with thickness of 1mm, in-plane resolution of 216x216 μm², TR = 8ms, number of averages = 1, flip angle = 6°. Spin echo and Gradient echo based imaging were performed on the same slices with identical matrix size, average and spatial resolution for quantitation of T₂ and T₂* respectively. Twelve spin echoes images with initial echo time of 6.9ms and step size of 6.9ms with TR = 4000ms, and ten gradient echo images with initial echo time of 2.0ms and a step size of 1.5ms with TR = 1030ms were acquired. Respiration gated static PET scans were acquired using a small-animal PET scanner (Inveon, Siemens) for visualization of activated BAT. CT scans were acquired immediately after the PET scan. A two-layer feed-forward neural network (NNet) having scaled conjugate gradient based learning/training, and cross-entropy based performance evaluation was adopted.⁵ Multi-parametric images including fat fraction, T₂, and T₂* were provided as inputs for NNet based segmentation and quantification. Isolated tissues were subjected to histology and UCP1 analysis.Figure 1a-c, shows the FF, T₂ and T₂* maps of the interscapular region of an eleven week rat which is very complimentary for differentiating the BAT, WAT and muscle. The H&E stained (Figure 1d) image shows the boundaries between BAT and WAT. Figure 1e shows the UCP1 gene expression which was significantly higher in BAT. Quantitative box plots of FF, T₂ and T₂* from 7, 11 and 15 week old rats are shown in Figure 2. Increasing trend of FF and T₂ was observed as a function of age. The FF, T₂ and T₂* were different for cold and thermoneutral groups. The FF of BAT was reduced significantly in cold exposed groups compared to thermoneutral groups. FF, T₂ and T₂* values had significant overlap (Figure 3a) between WAT, BAT and muscle tissues compared to cold exposed condition (Figure 3b). This separation increased the specificity for NNet based classification of BAT, WAT and muscle tissues. There was a decrease in the BAT volume (~10%) at 11 week as compared to 7 week old rats in both cold exposed and thermoneutral groups. Cold exposed group at 7 and 11 weeks showed increase in BAT volume by ~5%. Figure 4a-c shows co-registered PET-CT images under cold stimulus. FF, T2 and T₂* images provided a clear contrast in differentiating the BAT, WAT and muscle. T₂ and T₂* images offer complimentary contrast for muscle in relation to fat tissues. T₂ relaxation is sensitive to tissue composition, structure, water content, and iron levels and eliminates the fluctuations due to magnetic field inhomogeneity and suscepibility effects. T₂* relaxation measures the decay of the transverse magnetization caused by both spin-spin relaxation and magnetic field inhomogeneities. Blood perfusion and oxygenation tend to increase in cold activated BAT and increases the levels of deoxyhemoglobin contributing to relaxation differences between WAT and BAT.⁴ In our current study, we have observed lower FF in BAT for both thermoneutral and short term cold stimulated animals compared to WAT.⁶ This observation was supported by mRNA analysis showing significant increase of UCP1 in BAT compared to WAT. The FF increased significantly in 11 and 15 weeks old rats compared to seven week animals which might be due to increased WAT in different compartments. T₂ relaxation increased in eleven and fifteen week old rats due to changes in microenvironment including fat, water and vasculature. Better separation of WAT and BAT with cold exposure improved the classification and NNet based segmentation with high accuracy.