Table 4.

Select references from recent literature reporting fully automated approaches in segmenting subcutaneous, visceral, and muscle adipose tissue depots, using CT and MRI data in humans

Compartment	Modality	References	Remarks
Subcutaneous and visceral adipose tissues	CT	Zhao et al. [142]	−190 to −30 HU threshold range, radial profile approach to identify tissue boundaries, compared single-slice measurements at L4 and L5 vertebrae to abdominal volume measurements
		Ohshima et al. [143]	−190 to −30 HU threshold range, radial profile approach to identify abdominal tissue boundaries
		Makrogiannis et al. [144]	FCM to identify air, muscle, fat, and bone tissues, single-slice approach at L4 and L5 vertebrae, separation of subcutaneous and visceral compartments using gradient vector flow, ACM, manual removal of signals from food residues in gastrointenstinal tract
		Nemoto et al. [145]	−190 to −30 HU threshold range, multi-slice approach, data rescaling, removal of air voxels, identification of bone, fat, and muscle voxels, morphological and region-growing operations, validation against manual segmentation, in men and women
	MRI	Liou et al. [146]	T1W and T2W 1.5 T MRI. Four pulse sequences, SI HT, RG, EM, correlation with MA, consideration for motion artifacts and atypical anatomies
		Armao et al. [147]	Multi-slice FS 1.5 T MRI, SI HT, RG, correlation with MA
		Kullberg et al. [148]	3D 1.5 T CSE WFI with continuously moving table, multi- parametric analysis of water-only, fat-only, in-phase (water + fat), water fraction, and fat fraction data, SI HT, MO, lung segmentation, geometric models to exclude bone marrow in spine and pelvis, correlation with MA
		Kullberg et al. [149]	3D 1.5 T CSE WFI, exploits fat fraction data for HT, geometric model to exclude bone marrow and intermuscular adipose tissue, FCM, and MO, correlation with semi-automated analysis, correlation between single-slice and volume measurements, study in children
		Nakai et al. [150]	Multi-slice FS MRI, SI HT, template matching, correlation with MA, short-term longitudinal study
		Würslin et al. [151]	Multi-slice T1W 1.5 T MRI, whole-body analysis, SI HT, FCM, ACM, explicit detection of extremities, correlation with MA
		Zhou et al. [152]	Multi-slice FS 1.5 T MRI, with and without water suppression, SI HT, FCM, ACM and consideration of partial volume effects
		Wald et al. [153]	3D 1.5 T CSE WFI, whole-body analysis, SI HT, statistical shape and appearance models, correlation with MA, large n = 314 cohort
		Joshi et al. [154]	3D 3 T CSE WFI, atlas-based approach, correlation with MA
		Thörmer et al. [155]	Multi-slice 1.5 T CSE WFI, FCM, RG, ACM. Correlation with MA in obese cohort
		Addeman et al. [92]	3D 3 T CSE WFI, exploits fat fraction and T2* data, conversion from Cartesian to polar coordinates, surface fitting, correlation with MA, cross-sectional study
		Sadananthan et al. [156]	Multi-slice 3 T CSE WFI, EM, segmentation of superficial and deep subcutaneous depots using graph cut and level set methods, correlation with semi-automated analysis, cross- sectional study
Inter- and intra-muscular adipose tissues	CT	Senseney et al. [157]	Medical Imaging Processing, Analysis, and Visualization (MIPAV) software from the National Institutes of Health (http://mipav.cit.nih.gov/)
	MRI	Positano et al. [158]	Multi-slice T1W 1.5 T MRI, FCM, ACM, SI HT, EM algorithm, correlation with MA
		Prescott et al. [159]	Multi-slice T1W MRI, emphasis on interstitial adipose tissue in OA patients, N3 bias field correction, signal normalization, MO, RG, correlation with MA
		Makrogiannis et al. [160]	3D FS 3 T MRI, multi-parametric approach using water-suppressed and fat-suppressed complementary images, N3 bias field correction, K-means clustering, parametric deformable and ACM, correlation with CT
		Valentinitsch et al. [161]	3D 3 T CSE WFI, OA and type 2 diabetes cohort, comparison with MA, multi-parametric approach using water-only, fat- only, and in-phase (water + fat) images, K-means clustering, MO, and RG

Additional relevant citations are found therein

ACM active contour models, CSE WFI chemical-shift encoded water–fat MRI, EM expectation/maximization, FCM fuzzy C-means (clustering), FS frequency-selective, HT histogram thresholding, MA manual segmentation analysis, MO morphological operations, OA osteoarthritis, RG region growing, SI signal intensity, T1W/T2W T1-/T2-weighted