For CECT in the acute care setting, both quantitative and qualitative spermatic cord abnormalities are highly discriminatory for acute scrotal pathology, with AUC values of 0.87 and 0.86, respectively. While we found that NECT can also predict acute scrotal pathology, it is significantly less discriminatory with an AUC value of 0.67. Both CECT and NECT were predictive of a wide range of acute scrotal pathology, most being infectious, inflammatory, or neoplastic. Quantitative differential cord enhancement sub-analysis identified multiple discrete enhancement cutoffs and diagnostic accuracy rates as high as 89%, with the potential to choose thresholds based on a practice’s tolerable miss rate. A high degree of reader agreement was seen for both CECT and NECT (ICC = 0.981 and 0.963, respectively). However, readers reported significantly lower confidence in the diagnosis of cord abnormalities at NECT (4.33 vs 3.67 on a 5-point Likert scale, P < 0.001). Quantitative and qualitative performance also suffered in the setting of bilateral SSP, which accounted for up to 63% of readers’ false negatives at qualitative analysis.

While differential spermatic cord enhancement showed robust diagnostic utility in predicting SSP, choice of a diagnostic threshold may depend on the patient population and clinical acuity. In practice, cutoff selection could be tailored to best-address specific clinical priorities. For example, in an outpatient setting with a low pretest probability of SSP, a 35% differential enhancement level maximizing specificity may be preferred to avoid false positives and unnecessary follow-up imaging. Alternatively, in an urgent care setting, there may be a higher likelihood of acute pathology or confounding clinical presentations. In this environment, a differential enhancement level of 10% (emphasizing sensitivity) or 25% (emphasizing accuracy) may be more appropriate. Regardless of threshold value, our quantitative approach has the potential to improve patient care by alerting providers to unsuspected scrotal pathology, leading to more focused treatment, confirmatory imaging, timely urologic consultation, and fewer "bounce-back" care encounters for vague or non-localizing symptoms resulting from SSP.

High performance of any radiologic quantitative measure should also be weighed against reproducibility and ease-of-use. Here, our quantitative analysis was designed for simplicity, using HUmax from only two ROIs encompassing the spermatic cords. Employing HUmax should eliminate the impact of inclusion of fat surrounding the cord (unlike mean HU density measurements), decreasing variability introduced by the size or shape of a manually drawn ROI. Further, HUmax ROI placement may be more rapid than individual cord vessel diameter or intraluminal ROI measurements utilized in prior studies [12, 13]. Additionally, use of differential enhancement analysis may be more reliable than peak cord enhancement in the setting of non-uniform CECT protocols with use of variable iodine concentrations, injection rates and acquisition timing. While assessment of reproducibility and speed of use for quantitative measures is beyond the scope of this study, this could be addressed in a future prospective trial.

Qualitative assessment of the spermatic cord (relying on enlargement, hyperemia, or stranding) had excellent performance, only slightly behind quantitative analysis. While qualitative assessment of NECT was discriminatory for SSP, and its interobserver agreement was high, performance and radiologist confidence were significantly lower than with CECT. This is likely explained by lack of intravascular contrast and its additional diagnostic utility for qualitative assessment of the spermatic cords. As obstructing urolithiasis can present with referred scrotal pain, non-contrast imaging in the setting of SSP is likely common. While the inguinal canals are included on routine CECT and NECT, our findings reinforce the importance of incorporating the inguinal canal in radiologists’ routine search patterns, as both inguinal hernias and scrotal pathology may be prevalent in the acute care setting.

Two prior studies found a similar correlation between CECT spermatic cord abnormalities and unilateral scrotal pathology at ultrasound. In one, scrotal pathology was associated with asymmetric cord vessel size and hyperenhancement, however varicoceles accounted for 60% of positive cases [12]. In the second, scrotal pathology was associated with asymmetric cord enhancement in 62.5% of positive cases (primarily epididymo-orchitis and testicular neoplasm) versus 6.5% of negative cases. In our patient cohort, we also found fewer varicoceles, potentially due to differences in patient selection and the lower incidence of symptomatic varicoceles in the acute-care setting [14]. In that study, mean cord densities were associated with specific scrotal pathologies with differences in enhancement theorized to be related to differential blood supply [13, 15, 16]. We, however, found no significant differences in peak or differential cord enhancement between infectious, inflammatory, and neoplastic etiologies. While Gupta et al. posited that quantitative assessment of cord vessel attenuation and threshold ratios would likely be impractical or impossible at CT, we found differential cord enhancement to be both highly accurate and specific for SSP. Of particular note, neither prior study included cases of bilateral scrotal pathology, which we found in 12.5% of CECT and 25% of NECT positive cases. Unsurprisingly, SSP in these patients was harder to identify using either quantitative or qualitative analysis, likely due to greater symmetry of cord enhancement and morphology. Since bilateral scrotal pathology appears to be relatively common in the acute-care setting, its exclusion could have improved our quantitative and qualitative performance. However, its inclusion in our analysis more accurately reflects patient presentations seen in our clinical practice.

This study had limitations. First, this was a single-center retrospective study with inherent selection bias due to inclusion of only patients with contemporaneous ultrasound and CT. Similar to prior studies, our patient cohort did not include any cases of testicular torsion, likely due to its unambiguous clinical presentation [13]. Presumptively, torsion would show ipsilateral decreased spermatic cord enhancement, but, to our knowledge, this has yet to be reported outside of an experimental model [17]. Our cohort also did not include any cases of funiculitis, which would likely present with ipsilateral spermatic cord hyperemia. Second, there was a relatively small sample size for NECT cases. While our findings of lower NECT accuracy compared to CECT would be expected, a larger number of NECT exams could better confirm our findings. Third, our study CT protocols were heterogeneous, including post-contrast scan timing. While this reflects the day-to-day reality facing many practices, more uniform protocols may have shown better accuracy and/or reader agreement. Finally, while potentially dependent on individual practice parameters, diagnostic performance of defined quantitative thresholds would be better addressed with a future prospective trial.