Individual muscle segmentation in MR images: A 3D propagation through 2D non-linear registration approaches

Abstract

Abstract-Manual and automated segmentation of individual muscles in magnetic resonance images have been recognized as challenging given the high variability of shapes between muscles and subjects and the discontinuity or lack of visible boundaries between muscles. In the present study, we proposed an original algorithm allowing a semi-automatic transversal propagation of manually-drawn masks. Our strategy was based on several ascending and descending non-linear registration approaches which is similar to the estimation of a Lagrangian trajectory applied to manual masks. Using several manually-segmented slices, we have evaluated our algorithm on the four muscles of the quadriceps femoris group. We mainly showed that our 3D propagated segmentation was very accurate with an averaged Dice similarity coefficient value higher than 0.91 for the minimal manual input of only two manually-segmented slices.

Spatial difference can occur between activated and damaged muscle areas following electrically‐induced isometric contractions

“…This delineation was performed for one every two slices and missing slices were automatically interpolated (Ogier et al . ). Using the truncated cone formula, muscle volume was calculated by summing the areas of all the slices, taking into account the slice thickness and the gaps between slices.…”

Quantitative muscle MRI study of patients with sporadic inclusion body myositis

“…A total of 26 muscle groups for both thighs and LL were semi‐automatically segmented on the T 1 W images as previously described . At the thigh level: vasti (VAS) (vastus lateralis, medialis, and intermedius), rectus femoris (RF), semimembranosus and semitendinosus (SM/ST), biceps femoris (BF), sartorius (SA), gracilis (GR), adductors (ADD) (adductor brevis, longus and magnus).…”

“…Muscles were manually segmented using FSLview in five slices and the corresponding masks were propagated in the z‐direction for the whole set of slices . As previously described, slice selection was based on three independent criteria: (i) a muscle is appearing, (ii) a muscle is disappearing, (iii) a muscle shape is changing significantly. These criteria ensured that the semi‐automated process generated an accurate segmentation.…”

In vivo characterization of skeletal muscle function in nebulin‐deficient mice

“…The border of the anatomic cross‐sectional area of the whole hindlimb muscles was manually delineated by using FSLview in the two slices located on the proximal and distal parts to separate the muscle region as a whole and the SAT. Segmentation of the missing intermediate slices was automatically performed on the basis of registration procedures as previously described . From the muscle region, muscle tissue, IMAT, and bone/vessels/connective tissues were classified according to their respective signal intensities.…”

“…Segmentation of the missing intermediate slices was automatically performed on the basis of registration procedures as previously described. 24 From the muscle region, muscle tissue, IMAT, and bone/vessels/connective tissues were classified according to their respective signal intensities. The volume of the hindlimb muscles, IMAT, and SAT (mm 3 ) was calculated as the sum of the volume of the 11 consecutive largest slices.…”