A simplified ultrasound approach to diagnose testicular torsion and predict unsalvageable testis.

Abstract

To develop a decision tree model using US features to differentiate testicular torsion and other conditions of acutescrotum and to investigate predictive parameters of unsalvageable testis in testicular torsion. ScrotalUS was reviewed in patients aged <30 years who presented with acute scrotum from 2014 to 2020. US findings of whirlpoolsign, testicular volume ratio, heterogeneous echotexture, testicular vascularity, epididymis enlargement and/or hyperemia, andavascular nodule were evaluated and compared. A decision tree model was created using the conditional inference tree analysisand the accuracy was calculated. Univariate logistic regression analysis was performed to find out the predictive US features ofunsalvageable testes. Total of 381 patients (13.2±7.2 years old; range, 1 day-30 years) were included. Thirty-four patientswere diagnosed with testicular torsion, and the others with orchitis or epididymo-orchitis (n=59), epididymitis (n=264),and appendage torsion (n=24). In the conditional inference tree analysis, whirlpool sign, avascular nodule, and increasedtesticular vascularity were the most significant discriminators (p<0.001), and the whirlpool sign was the first discriminator.The overall accuracy of the conditional inference tree was 91.1% (95% confidence interval [CI], 87.8-93.7%). Heterogeneousechotexture (odds ratio [OR], 74.99; 95% CI, 2.75-2046.26; p=0.01) and symptom-to-operation time >24 h (OR, 49.28; 95%CI, 1.92-1262.03; p=0.02) were significant predictors of unsalvageable testis. Conditional inference tree analysisshowed that the whirlpool sign of the spermatic cord, avascular nodule, and altered testicular vascularity were significantdiscriminators. Heterogeneous echotexture and symptom-to-operation delay were important prognostic factors for unsalvageabletestis.