Quantitative assessment of nerve echogenicity: comparison of methods for evaluating nerve echogenicity in ulnar neuropathy at the elbow

Objective: The objective is to evaluate different methods to assess nerve echogenicity in a quantitative way by comparing a group of patients with ulnar neuropathy at the elbow (UNE) and a healthy control group, subsequently selecting the best tests for quantitative assessment of nerve echogenicity.

Methods: We included 56 patients with UNE and 37 healthy controls. High-resolution ultrasonography images of the ulnar nerve at the level of the medial epicondyle were saved in JPEG, TIFF or DICOM format, with a 19 pixels/mm resolution. Hypoechoic fraction was calculated by using 1 manual and 16 automatic thresholding methods.

Results: A significant difference in mean hypoechoic fraction between patients and controls was found using the following automatic thresholding methods: MaxEntropy 82% versus 74% (p < 0.001), RenyiEntropy 80% versus 69% (p < 0.001), Shanbhag 76% versus 68% (p = 0.002), Triangle 45% versus 58% (p = 0.036) and Yen 79% versus 67% (p < 0.001). Of these five tests a significant correlation between hypoechoic fraction and the cross-sectional area was found for: MaxEntropy 0.542 (p < 0.001), RenyiEntropy 0.558 (p < 0.001), Shanbhag 0.219 (p = 0.035) and Yen 0.513 (p < 0.001). The manual thresholding method did not detect a significant difference in hypoechoic fraction between patients and controls, and inter-rater agreement in hypoechoic fraction for manual thresholding was poor.

Conclusion: Quantitative nerve echogenicity assessment can be successfully used to distinguish between a group of patients with UNE and a healthy control group, preferably by using the MaxEntropy, RenyiEntropy or Yen methods.

Significance: Automatic thresholding techniques using the MaxEntropy, RenyiEntropy or Yen methods are the best quantitative tests, and these quantitative measures can probably be used in further studies evaluating echogenicity in mono- and polyneuropathies.