Value of grayscale histogram analysis based on ultrasound images in diagnosing sarcopenia.

BACKGROUND

Sarcopenia, an age-related condition marked by progressive muscle loss and dysfunction, is a growing clinical and public health challenge. While current diagnostic methods involve limitations in cost, accessibility, and assessment of muscle quality, ultrasound offers a practical alternative. This study examined grayscale histogram analysis of gastrocnemius muscle ultrasound images as a novel quantitative method for diagnosing sarcopenia by evaluating its ability to detect textural changes associated with intramuscular fat infiltration and fibrosis, with the ultimate aim of establishing an accurate, accessible diagnostic approach.

METHODS

A retrospective case-control study was conducted on 101 patients diagnosed with sarcopenia who were admitted to the Department of Endocrinology at Shandong Provincial Hospital between March and December 2024. Additionally, 101 healthy volunteers who underwent health examinations in our hospital during the same period were recruited as the control group. Grayscale histogram parameters, including the minimum gray value, maximum gray value, median gray value, mean gray value, standard deviation of gray values, skewness, kurtosis, and the gray values corresponding to seven percentile points (quantile 5, quantile 10, quantile 25, quantile 50, quantile 75, quantile 90, quantile 95) were extracted from the ultrasound images of the participants' gastrocnemius muscles. Statistical methods were used to analyze the differences between the sarcopenia and control groups. Receiver operating characteristic (ROC) curves were used to compare the differential diagnostic efficacy of each parameter and their combinations. Linear regression and least absolute shrinkage and selection operator (LASSO) were used to predict the probability of sarcopenia, with model performance evaluated with R values and the mean square error.

RESULTS

The grayscale histogram parameters of the gastrocnemius ultrasound images in the sarcopenia group, including the minimum gray value, maximum gray value, median gray value, mean gray value, standard deviation of gray values, and the gray values corresponding to seven percentile points, were significantly higher than those in the control group (P<0.001), while both the skewness and kurtosis were smaller than those in the control group (P<0.001). The gray value corresponding to quantile 75 demonstrated the best diagnostic efficacy [area under the curve (AUC) =0.988, sensitivity =96%, specificity =95%] at a cutoff of 132.5. The LASSO regression model outperformed linear regression (test set: R =0.769 0.727; mean square error =0.057 0.068).

CONCLUSIONS

The grayscale histogram parameters extracted from ultrasound images may be able to quantitatively reflect the differences between patients with sarcopenia and healthy individuals to some extent. Grayscale histogram analysis based on ultrasound images could be valuable for the diagnosis of sarcopenia.