Assessment of Deep Venous Thrombosis by Grayscale Median Analysis of Ultrasound Images

Abstract

This study aimed to determine whether grayscale median (GSM) ultrasound (US) analysis could differentiate acute from subacute femoropopliteal deep venous thrombosis (DVT) classified based on (a) the time of symptoms onset and (b) sonographers' evaluation of US images. Patients who had US because of suspected DVT were included. Patients with history of previous DVT, with normal US results or US with findings of isolated calf thrombosis were excluded. A total of 128 venous segments of 63 patients were analyzed. Thrombosed veins and adjacent arteries images were recorded. Thrombi GSM were calculated. Grayscale median was compared with (a) the time of symptoms onset and (b) US image thrombi classification as acute or subacute done by 2 sonographers blinded to patient information. Statistical analysis included GSM Pearson intraobserver variability and t test comparisons of GSM means for (a) short versus long onset periods and (b) acute versus subacute US class, χ comparison between onset history and US class, creation of GSM accuracy, predictive values, and sensitivity curves for acute versus subacute DVT. Intraobserver variability correlation varied between 0.92 and 0.95. Thrombus mean GSM was 23 ± 12 versus 31 ± 16 if symptoms onset was shorter or longer than 2 weeks (P = 0.004). Mean GSM was 20 ± 11 versus 35 ± 14 for acute versus subacute thrombi (P < 0.001). Acute thrombi correlated with onset shorter than 2 weeks (P < 0.001). Grayscale median less than 13, equal to 22, or 33 or less had 85% positive predictive value (17/20), 77% accuracy (98/128), and 90% sensitivity (53/59) to differentiate acute from subacute thrombi. Grayscale median of US images documenting femoropopliteal DVT correlated with (a) the time between symptoms onset and US examination and (b) the sonographers' interpretation of acute versus subacute thrombosis.