Novel Methodology for Muscle Volumization: 3D Laser Surface Scanning Meets CT

Segmentation is the term for identification of intrinsic structures from volumetric scans like computed tomography (CT) or magnetic resonance (MR). Materials offering high image contrast can be automatically segmented; however, segmentation of soft tissues is much more difficult due to tissue homogeneity and the resulting similarity in image greyscale at muscle boundaries. The need exists for complete musculoskeletal models wherein the musculature and bony skeleton can interact virtually using medical imaging software for use in finite element analysis – a computerized method for assessing how an object will react to physical forces. Additionally, future benefits of such models could have positive impacts on anatomical teaching, pre‐operative planning, and improve musculoskeletal computer modeling. We propose a methodology utilizing 3D laser surface scanning for semi‐automatic segmentation and registration of shoulder musculature to prescanned CT images. The 3D surface geometry and texture of a dissected shoulder specimen was undertaken with laser scanning (Artec Space Spider). Scanning took place before and after each muscle was manually dissected. Eleven geometrical surface models containing texture were collected (). Each model was exported as a stereolithographic (stl) mesh file containing only geometrical information. Meshes were imported into medical imaging software (Materialise 3‐Matic) where each mesh was registered to the initial CT using the bones as common registration fiducials. Next, each mesh was co‐registered to the pre‐scanned CT images. Each mesh was filled to create a volume that could be visualized both three dimensionally and on the CT images. A Boolean subtraction was used to subtract one 3D volume from another. Boolean subtractions were performed opposite the order of dissection so that deeper layers of meshes were subtracted from a more superficial layer, leaving one single muscle volume. The result is a 3D musculoskeletal model as depicted in . In order to validate the scanned muscle volume for accuracy, the mean of a three‐trial water displacement protocol was compared to our computer segmented muscle volumes. Of the 13 muscles measured, computer segmented error ranged from −10% to 18.2% (). Good agreement between muscle volumes were achieved between water displacement and surface laser computer segmented methods (). Future directions include manually segmenting musculature from the original CT images in order to compare with our current findings. Overall, the laser scanning approach and mesh workflow appears a valid methodology that may prove to be more expeditious than the current manual segmentation approaches. Support or Funding Information Department of Anatomy & Cell Biology, Schulich School of Medicine & Dentistry, Western University, London, ON, Canada 1 Roth | McFarlane Hand and Upper Limb Centre (HULC), St. Joseph's Health Care, London, ON, Canada 2 Muscle Computer Segmented Muscle Volume (mm3) Water Displacement Muscle Volume (mm3) Water Displacement Three Trial Standard Deviation % Error Trapezius 44621 44397 1010 0.5 Deltoid 294141 294769 2084 −0.2 Pectoralis Major 10904 9913 638 10.0 Pectoralis Minor 3230 2732 254 18.2 Triceps 51066 53510 2190 −4.6 Biceps 13399 13148 819 1.9 Coracobrachialis 27056 27656 1191 −2.2 Latissimus Dorsi 62328 69274 2138 −10.0 Teres Major 90518 82502 3102 9.7 Teres Minor 16584 16805 238 −1.3 Infraspinatus 104910 97372 1126 7.7 Subscapularis 161146 169024 1382 −4.7 Supraspinatus 50744 50819 1389 −0.1

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.