US 12,175,674 B1
Predicting paraspinal muscle sarcopenia using magnetic resonance imaging
Alexander P. Hughes, New York, NY (US); Frank Cammisa, Greenwich, CT (US); Andrew Sama, Greenwich, CT (US); Federico Girardi, New York, NY (US); Artine Arzani, Arcadia, CA (US); Kyle Finos, Bruce Township, MI (US); and Isaac Nathoo, Piscataway, NJ (US)
Assigned to Tissue Connect Systems, Inc., Garden City, NY (US)
Filed by Tissue Connect Systems, Inc., Garden City, NY (US)
Filed on Sep. 19, 2023, as Appl. No. 18/470,269.
Int. Cl. G06T 7/00 (2017.01); G06T 7/174 (2017.01)
CPC G06T 7/0012 (2013.01) [G06T 7/174 (2017.01); G06T 2207/10088 (2013.01); G06T 2207/20084 (2013.01)] 27 Claims
OG exemplary drawing
 
1. A method for measuring paraspinal muscle parameters, the method comprising:
obtaining a plurality of magnetic resonance imaging (MRI) 2-dimensional image slices of a spine of a subject;
identifying, by a slice selection model, one or more representative slices in the plurality of MRI slices, each of the one or more representative slices being located at or adjacent to a specified physiological region corresponding to a specific biomechanical environment in a respective spine level of the spine of the subject, wherein the slice selection model comprises a first stage and a second stage, and wherein the plurality of MRI slices includes one or more sets of MRI slices corresponding to one or more respective spine levels of the spine of the subject, and wherein identifying the one or more representative slices in the plurality of MRI slices comprises:
(i) generating, by the first stage of the slice selection model, respective encodings for each of the MRI slices in the plurality of MRI slices; and
(ii) for each of the one more sets of MRI slices, selecting, by the second stage of the slice selection model, a respective representative slice located at or adjacent to the specified physiological region corresponding to the specific biomechanical environment in the respective spine level of the spine of the subject, wherein the selecting of each representative slice in each respective spine level is based on the encodings of the MRI slices in the plurality of MRI slices;
segmenting, by a muscle segmentation model, each respective representative slice of the one or more representative slices, wherein segmenting each respective representative slice includes generating an image segmentation mask including a plurality of pixels in the respective representative slice corresponding to paraspinal muscle and excluding a plurality of pixels in the respective representative slice corresponding to metal artifacts and/or to biological matter of the subject other than paraspinal muscle; and
extracting, by a measurement extraction model, one or more measurements of one or more respective parameters of the paraspinal muscle of the subject based on the plurality of pixels included in the generated image segmentation mask of each of the one or more representative slices.