Anatomical Modeling of Skeletal Muscles
Figure 3.7 shows the correspondence of the anatomical features. These features are utilized by an anatomical modeling method. These three features details described below.
Landmarks It is important to recognize the origin and insertion points of a muscle in the bone as landmarks. A landmark is a one-dimensional datum indicating one point in three-dimensional space. Various methods exist for landmark recognition. As skeletal muscle is attached to planar configurations, if the selection of the bone is correct, some positional error is acceptable in the shape model described below.
Fig. 3.6 Representative features of skeletal muscle in CT images
Fig. 3.7 Correspondence of the anatomical features on computer
For the recognition of landmarks, we use skeleton images classified automatically using an atlas derived from non-contrast CT images. This classified skeleton image is generated by a method considering the connectivity of the skeleton .
Landmarks are selected by first selecting the bone from the atlas based on the anatomical definition and then finding the most proximal point on the selected bone for the origin and the most distal point for the insertion. Classified skeletal image atlases and the virtual image unfolding-based method by this author and other studies [134, 216] describe this.
Anatomical Centerlines Anatomical centerlines indicate the direction of the long axes of the muscle fibers. Therefore, in each skeletal muscle, a pair of origin and insertion landmarks are connected by a straight line or curve. For use as an approximate running direction of the muscle fibers that supports recognition, there is no serious problem even if using a straight line or low-dimensional mathematical function.
Because of differences in height, body shape, gender, subcutaneous fat distribution, and other factors, there are large individual differences in the shape of the human body. In our experience, the geometric centerline is different from the anatomic one. Thus, in surface skeletal muscles, it is preferable to generate a centerline passing through multiple points placed as additional landmarks between the origin and insertion. We select additional landmarks using the surface of the subcutaneous fat, which is relatively easy to recognize by its low-density values.
Figure 3.8 shows anatomic centerlines generated in the identification of surface and deep muscle.
Model Generation The modeling of skeletal muscle can involve displaying the distribution in three-dimensional space or using an SSM to indicate the outline of the muscle. In the recognition process, it is necessary to select whether to use both models or to use the distribution model alone. Many muscles, such as the temporalis, have a complicated shape that requires the distribution model. Muscles with well- defined insertion points, belly, and attachment points, such as the biceps, can be analyzed with the shape model.
Fig. 3.8 Anatomic centerlines of the surface muscle and deep muscle