Abstract
Objectives
Computed tomography (CT) is frequently the first cross-sectional imaging modality obtained in patients presenting with suspected biliary obstruction, yet its sensitivity for direct detection of common bile duct (CBD) stones is known to be variable. This study aimed to determine the rate of CT-missed findings in patients with magnetic resonance cholangiopancreatography (MRCP)-confirmed choledocholithiasis and to identify stone and protocol characteristics associated with CT nondetection.
Methods
In this single-center retrospective study, 33 adults with MRCP-confirmed choledocholithiasis who had undergone abdominal CT within seven days of MRCP were included. CT images were independently re-evaluated by two blinded radiologists; discordant readings were resolved by consensus. MRCP served as the reference standard. CT sensitivity and its 95% confidence intervals (CI) were calculated using the Wilson method. Intergroup comparisons were performed using the Mann-Whitney U test and Fisher’s exact test. Agreement between MRCP and CT measurements of CBD diameter was assessed by Spearman correlation and Bland-Altman analysis.
Results
CT sensitivity was 48.5% (16/33; 95% CI, 32.5-64.8%), corresponding to a missed finding rate of 51.5%. MRCP stone diameter was significantly larger in the CT-detected group (median 9.0 mm vs. 5.0 mm; p=0.001), and CT-measured CBD diameter was likewise larger (median 14.0 mm vs. 10.0 mm; p=0.017). Stones measuring ≤5 mm were missed on CT in 83.3% of cases versus 33.3% for stones larger than 5 mm (odds ratio, 10.00; p=0.010). MRCP and CT CBD diameters were strongly correlated (Spearman r=0.793; p<0.001), with negligible bias on Bland-Altman analysis (+0.18 mm). The Median stone attenuation in CT-detected cases was 78 hounsfield units (HU) (interquartile range, 56-114 HU).
Conclusion
CT failed to identify CBD stones in more than half of patients with MRCP- confirmed choledocholithiasis. Stone diameter ≤5 mm was the strongest predictor of CT non-detection. These findings support a low threshold for MRCP referral when CT does not directly demonstrate a CBD stone in patients with clinically suspected biliary obstruction.
Introduction
Cholelithiasis affects approximately 10-15% of the adult population in Western countries, and choledocholithiasis the presence of stones within the common bile duct (CBD) develops in 10-20% of patients with gallbladder stones at some point during their disease course.1, 2 Left unrecognised or inadequately managed, CBD stones predispose to serious biliary complications including acute cholangitis, acute biliary pancreatitis, and secondary biliary cirrhosis, all of which are associated with substantial morbidity and, in severe cases, mortality.3 Timely identification and appropriate triage of patients for further evaluation by magnetic resonance cholangiopancreatography (MRCP), endoscopic ultrasound (EUS), or therapeutic endoscopic retrograde cholangiopancreatography (ERCP) is therefore a clinical priority.4
In current clinical practice, transabdominal ultrasonography is the first line imaging modality for suspected biliary pathology; however, its sensitivity for direct CBD stone visualisation is limited, ranging from 22% to 55% in published series, largely owing to acoustic shadowing from overlying bowel gas and operator dependence.5, 6
Abdominal CT is widely obtained in emergency department patients presenting with non-specific abdominal pain, right upper quadrant discomfort, jaundice, or suspected acute pancreatitis, owing to its broad anatomical coverage and rapid acquisition.7 While CT can simultaneously evaluate a wide differential diagnosis and detect indirect biliary signs notably CBD dilatation the direct visualisation of a CBD stone on CT is considerably less reliable and is influenced by a complex interplay of stone composition, size, location, and technical imaging factors.8-10
The reported sensitivity of CT for choledocholithiasis varies widely across the literature, from approximately 40% to 80% depending on the study cohort, CT protocol, and reference standard used.8-11 Isoattenuating stones, stones smaller than 5 mm, and stones situated in the distal CBD are particularly prone to being overlooked.8, 10 MRCP provides non-invasive high resolution depiction of the biliary tree and has been validated as a highly accurate reference standard for choledocholithiasis, with pooled sensitivity and specificity exceeding 90%.12 Current European Society of Gastrointestinal Endoscopy (ESGE) guidelines endorse MRCP or EUS as the preferred diagnostic tools for intermediate-probability patients before therapeutic ERCP.4
Despite these well-established limitations, CBD stones are frequently not reported on CT examinations that precede or coincide with an eventual MRCP diagnosis of choledocholithiasis. The aim of this study was to determine the frequency of missed CBD stones on abdominal CT in a cohort of patients with MRCP-confirmed choledocholithiasis, and to compare CT-detected and CT-missed cases with respect to stone size and attenuation, CBD diameter, and CT acquisition protocol characteristics.
Methods
This was a single-center retrospective observational study conducted at Kastamonu Training and Research Hospital. Ethical approval was granted by the Kastamonu University Non-Interventional Clinical Research Ethics Committee (approval no: 2026-18, date 19.02.2026). The requirement for individual informed consent was waived given the retrospective nature of the study and the use of anonymised archival data. All data were handled in accordance with applicable data protection legislation.
Electronic medical records and the picture archiving and communication system were searched to identify patients who underwent MRCP between 1 January 2023 and 31 December 2025 for whom choledocholithiasis was confirmed. Eligibility required age ≥18 years, MRCP findings consistent with CBD stones, and at least one abdominal CT performed within seven days before or after MRCP in DICOM format. Patients were excluded if ERCP or surgical stone extraction had been performed between CT and MRCP, if biliary stenting or an artefact precluded CBD assessment, or if image quality was insufficient. The full criteria are summarised in Table 1. Thirty-three patients constituted the final cohort.
All CT examinations were performed for routine clinical indications on a 64-detector, 128-slice scanner (Revolution EVO; GE Healthcare, Chicago, IL, USA). Both contrast-enhanced (portal venous phase) and unenhanced acquisitions were represented. Axial images were acquired at 1.25 mm; multiplanar reconstructions were reviewed when available. MRCP was performed on a 1.5 Tesla system (SIGNA Victor; GE Healthcare). MRCP served as the reference standard.
CT images were independently reevaluated by two blinded radiologists with 1 and 11 years of abdominal imaging experience, respectively. Discordant readings were resolved by consensus. CT positivity was defined as direct visualization of a CBD stone, either as a hyperdense intraluminal focus or as a filling defect (target/crescent sign). A missed CBD stone was defined as a case in which MRCP confirmed choledocholithiasis and retrospective CT rereading identified findings compatible with a stone that had not been reported on the original interpretation.
Statistical Analysis
All analyses were performed using IBM SPSS Statistics (version 26.0). Normality was assessed using the Shapiro-Wilk test. Results are reported as mean ± standard deviation (SD) or median [interquartile range (IQR)] as appropriate. CT sensitivity and 95% confidence intervals (CIs) were calculated using the Wilson method. Continuous variables were compared using the Mann-Whitney U test, and categorical variables were compared using Fisher’s exact test. MRCP and CT CBD diameter correlation was assessed by Spearman coefficient; agreement by Bland Altman analysis.12 The association between stone diameter (≤5 mm vs. >5 mm) and CT non-detection was evaluated by Fisher exact test with odds ratio (OR) and 95% CI. p<0.05 was considered statistically significant.
Results
The cohort comprised 33 patients (17 men, 16 women; mean age, 68.1±17.3 years; range, 32-93 years). Baseline characteristics are shown in Table 2. Intrahepatic ductal dilatation was present in 27 patients (81.8%). Contrast-enhanced CT was performed in 20 (60.6%) patients and unenhanced CT in 13 (39.4%) patients. Periduodenal gas was identified on CT in 7 patients (21.2%). Median MRCP stone count was 1 (IQR, 1-2), median stone diameter was 7.0 mm (IQR, 4.0-9.0), and mean MRCP CBD diameter was 11.7±3.7 mm.
CT sensitivity was 48.5% (16/33; 95% CI, 32.5-64.8%), corresponding to a missed finding rate of 51.5% (17/33; 95% CI, 35.2-67.5%). Specificity, positive predictive value, and negative predictive value could not be derived because the cohort consisted exclusively of MRCP-positive patients.
Group comparisons are shown in Table 3. MRCP stone diameter was significantly larger in the CT-detected group [median: 9.0 mm (IQR, 7.8-11.2) vs. 5.0 mm (IQR, 3.0-7.0); U=232; p=0.001]. CT-measured CBD diameter was likewise larger [median, 14.0 mm (IQR, 10.8-16.0) vs. 10.0 mm (IQR, 8.0-12.0); U=202; p=0.017]. Stone count, MRCP CBD diameter, and age did not differ significantly between groups. Intrahepatic ductal dilatation (OR, 2.15; p=0.656), contrast-enhanced CT (OR, 1.17; p=1.000), and periduodenal gas (OR, 3.41; p=0.225) were comparable between groups.
MRCP and CT CBD diameters were strongly correlated (Spearman r=0.793; p<0.0001). Bland-Altman analysis showed a mean bias of +0.18 mm, SD of differences of 2.34 mm, and 95% LoA of -4.40 to +4.76 mm (Figure 1).
Stratification by MRCP stone diameter showed that CT missed stones ≤5 mm in 10 of 12 patients (83.3%) versus 7 of 21 patients (33.3%) with stones >5 mm (OR, 10.00; p=0.010) (Figure 2).
Among the 16 CT- detected stones, the median attenuation was 78 hounsfield units (HU) (IQR, 56-114 HU; range, 13-505 HU; Shapiro-Wilk p<0.001). Eight stones (50.0%) showed mixed or partially calcified attenuation (20-99 HU), 6 (37.5%) were densely calcified (≥100 HU), and 2 (12.5%) were near isoattenuating (<20 HU) (Table 4, Figure 3-5).
Discussion
The principal finding of this study is that CT failed to identify CBD stones in more than half of the patients with MRCP-confirmed choledocholithiasis, yielding an overall sensitivity of 48.5%. This figure falls within the range reported in the published literature, toward the lower end. Anderson et al.7 found a sensitivity of 77.2% for multidetector CT in a dedicated prospective series, while Soto et al.8 reported sensitivities as low as 46% for unenhanced helical CT. A more recent comparative analysis by Al-Dulaimi et al.11 confirmed that CT performs substantially worse than magnetic resonance imaging/MRCP for CBD stone detection in a real world emergency cohort. The comparatively low sensitivity observed in the present study is consistent with real- world retrospective series and likely attributable to the stone characteristics of the study population, in which small and mixed- attenuation calculi predominated, rather than to systematic technical inadequacy. The heterogeneity across published estimates underscores that CT sensitivity for choledocholithiasis is not a fixed diagnostic property but is heavily dependent on stone characteristics, patient selection, and protocol optimization.
Stone diameter on MRCP emerged as the most robust discriminator between CT-detected and CT-missed cases. The CT-detected group had a significantly larger median stone diameter (9.0 mm vs. 5.0 mm; p=0.001), and when a clinically relevant threshold of 5 mm was applied, the odds of CT non-detection were ten times higher for small stones (OR, 10.00; p=0.010). This finding is consistent with the well established principle that CBD stones smaller than 5 mm are substantially harder to identify on axial CT owing to partial volume averaging and the limited attenuation contrast between small calcifications and surrounding bile at standard reconstruction thicknesses.7, 10 Moreover, small CBD stones carry significant clinical risk microlithiasis is a recognised precipitant of acute biliary pancreatitis and cholangitis even in the absence of substantial ductal dilatation3, 13, 14 making their CT invisibility a clinically consequential limitation. These results support a low threshold for MRCP referral in patients with symptoms consistent with biliary obstruction, in whom CT shows no direct evidence of stones but cannot confidently exclude small calculi.
The attenuation profile of CT-detected stones was predominantly mixed or partially calcified, with a median of 78 HU (IQR, 56-114 HU). Only 37.5% of CT visible stones were densely calcified (≥100 HU), while 12.5% exhibited near isoattenuating characteristics. This distribution reflects the known compositional heterogeneity of CBD stones, which range from purely cholesterol (low attenuation, often isoattenuating with bile) to pigment dominant varieties.7, 8 The two near isoattenuating stones identified on CT rereading represent cases in which the diagnosis was recoverable only through systematic scrutiny, careful biliary windowing, and attention to subtle intraluminal filling defects. The 17 stones missed entirely in the original reporting presumably included a disproportionate number of isoattenuating or very small calcifications not visible under standard reporting conditions, which reinforces the conclusion that the diagnostic gap reflects both a physical limitation of CT and a partially addressable reporting behaviour.
The CT-measured CBD diameter was significantly larger in the CT-detected group (median, 14.0 mm vs. 10.0 mm; p=0.017). A plausible physiological explanation is that a larger stone is more likely to cause upstream biliary dilatation, increasing duct caliber and improving the conspicuity of an intraluminal filling defect. Conversely, smaller stones may cause only mild or intermittent dilatation. MRCP and CT CBD diameters were strongly correlated (Spearman r=0.793; p<0.0001), and Bland-Altman analysis demonstrated a negligible mean bias of +0.18 mm with 95% LoA of -4.40 to +4.76 mm.12 Although the wide limits of agreement indicate that individual measurements may deviate by up to approximately 5 mm, the absence of systematic bias confirms that CT reliably tracks CBD diameter at the population level. The broad LoA most likely reflects differences in measurement planes, distension state, and respiratory phase, rather than any fundamental inaccuracy of CT choledochometry.
Neither the use of contrast-enhanced CT (OR, 1.17; p=1.000) nor the presence of periduodenal gas (OR, 3.41; p=0.225) was associated with a significantly higher CT detection rate. These null results must be interpreted with caution, given the limited statistical power of this pilot study. The existing literature suggests that portal venous phase CT can meaningfully improve stone detection by optimising the attenuation contrast between CBD contents and surrounding structures.10 Systematic use of multiplanar coronal and sagittal reformations has similarly been shown to improve detection of biliary and abdominal pathology compared with axial-only review.15-19 The failure to detect protocol-related effects in the present series most likely reflects an insufficient sample size rather than a genuine absence of effect, and should not be construed as evidence that contrast enhancement and MPR review are unimportant in biliary CT evaluation.
The rate of missed findings documented here reflects a combination of physical imaging limitations and reporting practices. CBD stones represent one of the more common sources of missed or delayed diagnoses on abdominal CT, partly because the interpreter’s attention is directed toward the primary clinical question and the biliary tree receives only secondary scrutiny, a phenomenon analogous to satisfaction of search errors described in other CT contexts.20 The disparity in experience between the two rereading radiologists (1 year vs. 11 years) reflects the heterogeneity of expertise in real clinical settings.21 In the present consensus-based design, the potential negative influence of inexperience was mitigated by joint resolution of discordant cases, but in routine single-reader reporting, which characterises most emergency CT workloads, the risk of perceptual oversight remains a real, modifiable component of the diagnostic gap.
The clinical implications of these findings are relevant to several intersecting practice decisions. Current ESGE guidelines recommend MRCP or EUS for patients with intermediate pretest probability of choledocholithiasis before proceeding to ERCP.4 EUS has demonstrated high sensitivity and specificity for CBD stones in systematic reviews, with particular utility for small stones and borderline ductal dilatation.13-17 The results of the present study support the view that a CT showing no direct stone evidence should not confidently exclude choledocholithiasis, particularly in patients with stones ≤5 mm, mildly dilated ducts around 10 mm, or a syndrome consistent with biliary obstruction. In these cases, proceeding directly to MRCP or EUS where available is likely to reduce diagnostic delay, prevent unnecessary repeat imaging, and facilitate timely biliary drainage.4, 14
Study Limitations
This study has several limitations. The sample size of 33 patients is small, and the study should be considered a pilot and hypothesis-generating study rather than definitive. The absence of an MRCP negative control group precluded the calculation of specificity, positive predictive value, or negative predictive value. The retrospective single-centre design limits generalisability. Formal interobserver kappa statistics were not calculated as a primary endpoint, as the consensus design precluded separate analysis of individual reader performance; future prospective studies should incorporate inter rater agreement as a designated outcome.21 The heterogeneous CT protocols reflect real- world practice, but limit protocol- specific subgroup analyses. Finally, laboratory covariates prespecified in the original study design were not available in a sufficiently complete form for inclusion. Despite these limitations, the study provides quantitative real-world data on CT missed-finding rates in MRCP-confirmed choledocholithiasis and identifies stone size as the primary determinant of CT detectability, generating specific hypotheses for adequately powered prospective investigation.
Conclusion
CT failed to identify CBD stones in more than half of patients with MRCP confirmed choledocholithiasis in this single centre retrospective cohort, yielding a sensitivity of 48.5%. Stone diameter ≤5 mm was the strongest predictor of CT non-detection, with an OR of 10.00 compared with stones larger than 5 mm. MRCP and CT measurements of CBD diameter were strongly correlated, with negligible systematic bias, supporting the reliability of CT choledochometry at the population level. These results reinforce current guideline recommendations in favour of early MRCP referral for patients with a clinical presentation consistent with biliary obstruction, in whom CT does not directly demonstrate a CBD stone. Larger prospective multicentre studies are warranted to confirm these findings, to evaluate the independent contribution of CT protocol factors such as phase selection and slice thickness, and to develop evidence-based reporting standards for biliary CT evaluation.
