Date of Award
Master of Science (MS)
Electrical and Computer Engineering
Myofascial Pain Syndrome (MPS) is a common chronic muscle pain disorder that affects a large portion of the global population, seen in 85-93% of patients in specialty pain clinics . MPS is characterized by hard, palpable nodules caused by a stiffened taut band of muscle fibers. These nodules are referred to as Myofascial Trigger Points (MTrPs) and can be classified by two states: active MTrPs (A-MTrPs) and latent MtrPs (L-MTrPs). Treatment for MPS involves massage therapy, acupuncture, and injections or painkillers. Given the subjectivity of patient pain quantification, MPS can often lead to mistreatment or drug misuse. A deterministic way to quantify the pain is needed for better diagnosis and treatment.
Various medical imaging technologies have been used to try to find quantifiable and measurable biomarkers of MTrPs. Ultrasound imaging, with it’s accessibility and variety of modalities, has shown significant findings in identifying MTrPs. Elastography ultrasound, which is used for measuring stiffness in soft tissues, has shown that MTrPs tend to be stiffer than normal muscle tissue. Doppler ultrasound has shown that bloodflow velocities differ significantly in areas surrounding MTrPs. MTrPs have been identified in standard B-mode grayscale ultrasound, but have varying conclusions with some studies identifying them as dark hypoechoic blobs and other studies showing them as bright hyperechoic blobs. Despite these discoveries, there is a high variance among results with no correlations to severity or pain.
As a step towards quantifying the pain associated with MTrPs, this work aims to introduce a machine learning approach using image processing with texture recognition to detect MTrPs in Bmode ultrasound. A texture recognition algorithm called Gray Level Co-Occurrence Matrix (GLCM) is used to extract texture features from the B-mode ultrasound image. Feature maps are generated to emphasize these texture features in an image format in anticipation that a deep convolutional neural network will be able to correlate the features with the presence of a MTrP. The GLCM feature maps are compared to the elastography ultrasound to determine any correlations with muscle stiffness and then evaluated in the presence of MTrPs. The feature map generation is accelerated with a GPU-based implementation for the goal of real-time processing and inference of the machine learning model. Finally, two deep learning models are implemented to detect MTrPs comparing the effect of using GLCM feature maps of B-mode ultrasound to emphasize texture features for machine learning model inputs.
Formby, Benjamin, "Detection of Myofascial Trigger Points With Ultrasound Imaging and Machine Learning" (2023). All Theses. 4182.