Feasibility of a Support Vector Machine Classifier for Myofascial Pain Syndrome: Diagnostic Case‐Control Study

Abstract

Objectives Myofascial pain syndrome (MPS) is the most common cause of chronic pain worldwide. The diagnosis of MPS is subjective, which has created a need for a robust quantitative method of diagnosing MPS. We propose that using a support vector machine (SVM) along with ultrasound (US) texture features can differentiate between healthy and MPS‐affected skeletal muscle. Methods B‐mode US video data were collected in the upper trapezius muscle of healthy (29) participants and patients with active (21) and latent (19) MPS, using an acquisition method outlined in previous works. Regions of interest were extracted and filtered to obtain a unique set of 917 images where texture features were extracted from each region of interest to characterize each image. These texture features were then used to train 4 separate binary SVM classifiers using nested cross‐validation to implement feature selection and hyperparameter tuning. The performance of each kernel was estimated on the data and validated through testing on a final holdout set. Results The radial basis function kernel classifier had the greatest Matthews correlation coefficient performance estimate of 0.627 ± 0.073 (mean ± SD) along with the largest area under the curve of 91.0% ± 3.0%. The final holdout test for the radial basis function classifier resulted in 86.96 accuracy, a Matthews correlation coefficient of 0.724, 88% sensitivity, and 86% specificity, validating our earlier performance estimates. Conclusions We have demonstrated that specific US texture features that have been used in other computer‐aided diagnostic literature are feasible to use for the classification of healthy and MPS muscle using a binary SVM classifier.

Machine Learning Diagnostic Modeling for Classifying Fibromyalgia Using B-mode Ultrasound Images

“…While computeraided diagnosis research in ultrasound has been heavily explored in areas such as cancer and liver diseases, [30][31][32] there has been little to no work in chronic pain. 18 In our previous work, we demonstrated that image texture features extracted from B-mode ultrasound images can be used to discriminate healthy asymptomatic individuals from patients with MPS. 10 This became the basis for the current work using machine learning modeling; the objective was to develop a statistical model with good predictive and diagnostic capabilities.…”

“…The SVM performance validation on the holdout test set demonstrated 84.1% accuracy, demonstrating similar discriminative capabilities to our previous findings. 10,18 Overall, to date, there has not been another study that has used machine learning models for binary classification of B-mode ultrasound images of the muscles of the upper trapezius in healthy and FM patients.…”

“…To solve this problem, the images were sorted using the complex wavelet structural similarity index measure (CW-SSIM) method that has been previously used by our group. 18 Briefly, the CW-SSIM is an extension of the original SSIM algorithm which was created to capture the visual impact of changes to contrast, structure, and luminance. As the SSIM is sensitive to various types of noise, the CW-SSIM is used instead which involves a complex version of a multiscale, multiorientation, steerable pyramid decomposition, 19 in which for our case we used four-scale and 16 orientations.…”

“…Much like our prior work on MPS and SVMs, 18 we approached the FM diagnostic problem in the same manner. We initially demonstrated that the image texture features extracted from both FM and healthy ultrasound data sets were statistically significantly different through a number of multivariate statistical techniques such as principal components analysis and multivariate analysis of variance.…”

Intelligent Algorithm-Based Picture Archiving and Communication System of MRI Images and Radiology Information System-Based Medical Informatization

“…At the patient level, the overall accuracy of the RBF classifier was 94.97%. Behr et al [ 18 ] carried out a case-control experiment to study the feasibility of a support vector machine classifier for myofascial pain syndrome. They found that the accuracy of the RBF classifier was 86.96, the Matthews correlation coefficient was 0.724, the sensitivity was 88%, and the specificity was 86%.…”