Quantitative High-Frequency Ultrasound Identifies Spermatogenesis in Infertile Men with Non-Obstructive Azoospermia

ObjectiveTo determine whether quantitative ultrasound (QUS), which characterizes tissue microstructure using radiofrequency data, can identify regional heterogeneity within seminiferous tubules that corresponds to localized spermatogenesis in men with non-obstructive azoospermia (NOA). DesignTwo-cohort study using a biological extremes cohort to establish plausibility of a QUS biomarker, followed by an independent NOA-biopsy cohort with site-matched imaging and tissue sampling. SettingAcademic male fertility referral center. PatientsThe biological extremes cohort included fertile men with presumed intact spermatogenesis (n=15) and men with NOA and subsequent negative microdissection testicular sperm extraction (mTESE; n=10). The NOA-biopsy cohort consisted of 27 men with NOA undergoing site-matched testicular biopsy via testicular sperm aspiration (TESA) or testicular sperm extraction (TESE), yielding 12 sperm-positive and 36 sperm-negative biopsy sites. InterventionsHigh-frequency testicular ultrasound (36 MHz) with acquisition of raw radiofrequency data, allowing objective, quantitative analysis of tissue scattering patterns beyond conventional grayscale imaging. Regions of interest were manually annotated and, in the NOA-biopsy cohort, spatially matched to biopsy locations. Main Outcome MeasuresAssociation between sperm presence at biopsy sites and a pre-specified QUS measure of local tissue heterogeneity: the 75th percentile of a sliding window coefficient of variation map of the Nakagami k-factor within the superficial testicular parenchyma (K_Zone1_CV). This metric reflects the upper range of local variability in ultrasound backscatter, which is influenced by the underlying organization of seminiferous tubules. ResultsIn the biological extremes cohort, K_Zone1_CV distinguished fertile controls (median 1.79, IQR 1.64-1.85) from NOA men with globally negative mTESE (median 1.51, IQR 1.42-1.58; P < 0.001), with an area under the receiver operating characteristic curve (AUC) of 0.91 (95% CI 0.79-1.00). In the independent NOA-biopsy cohort, K_Zone1_CV discriminated sperm-positive from sperm-negative biopsy sites with an AUC of 0.93 (95% CI 0.85-0.99). At a threshold of 1.60, sensitivity was 100%, specificity was 86.1%, positive predictive value was 70.6%, and negative predictive value was 100%. Serum hormone levels, testicular volumes, and biopsy technique did not differ significantly between groups. ConclusionsRegional testicular tissue heterogeneity measured by quantitative ultrasound is associated with localized spermatogenesis in men with NOA. At the selected threshold, no sperm-positive biopsy site was misclassified as negative. These findings support the hypothesis that QUS can noninvasively detect the focal seminiferous tubule heterogeneity that predicts sperm retrieval success. This imaging approach could inform future image-guided sperm retrieval strategies. Further validation in larger cohorts and assessment of intra-patient variability are needed.