CT differentiation of 1–2-cm gallbladder polyps: benign vs malignant

The value of CT radiomics combined with deep transfer learning in predicting the nature of gallbladder polypoid lesions

BACKGROUND

Computed tomography (CT) radiomics combined with deep transfer learning was used to identify cholesterol and adenomatous gallbladder polyps that have not been well evaluated before surgery.

PURPOSE

To investigate the potential of various machine learning models, incorporating radiomics and deep transfer learning, in predicting the nature of cholesterol and adenomatous gallbladder polyps.

MATERIAL AND METHODS

A retrospective analysis was conducted on clinical and imaging data from 100 patients with cholesterol or adenomatous polyps confirmed by surgery and pathology at our hospital between September 2015 and February 2023. Preoperative contrast-enhanced CT radiomics combined with deep learning features were utilized, and t-tests and least absolute shrinkage and selection operator (LASSO) cross-validation were employed for feature selection. Subsequently, 11 machine learning algorithms were utilized to construct prediction models, and the area under the ROC curve (AUC), accuracy, and F1 measure were used to assess model performance, which was validated in a validation group.

RESULTS

The Logistic algorithm demonstrated the most effective prediction in identifying polyp properties based on 10 radiomics combined with deep learning features, achieving the highest AUC (0.85 in the validation group, 95% confidence interval = 0.68-1.0). In addition, the accuracy (0.83 in the validation group) and F1 measure (0.76 in the validation group) also indicated strong performance.

CONCLUSION

The machine learning radiomics combined with deep learning model based on enhanced CT proves valuable in predicting the characteristics of cholesterol and adenomatous gallbladder polyps. This approach provides a more reliable basis for preoperative diagnosis and treatment of these conditions.

Gallbladder polyps and adenomyomatosis

Incidental findings are commonly detected during examination of the gallbladder. Differentiating benign from malignant lesions is critical because of the poor prognosis associated with gallbladder malignancy. Therefore, it is important that radiologists and sonographers are aware of common incidental gallbladder findings, which undoubtedly will continue to increase with growing medical imaging use. Ultrasound is the primary imaging modality used to examine the gallbladder and biliary tree, but contrast-enhanced ultrasound and MRI are increasingly used. This review article focuses on two common incidental findings in the gallbladder; adenomyomatosis and gallbladder polyps. The imaging features of these conditions will be reviewed and compared between radiological modalities, and the pathology, epidemiology, natural history, and management will be discussed.

Conventional Ultrasound Combined With Contrast-Enhanced Ultrasound in Differential Diagnosis of Gallbladder Cholesterol and Adenomatous Polyps (1-2 cm)

OBJECTIVES

This study aimed to determine ultrasonic image characteristics that enable differentiation between cholesterol and adenomatous polyps and to assess the diagnostic efficacy of combining conventional ultrasound (CUS) with contrast-enhanced ultrasound (CEUS).

METHODS

Eighty-nine patients with gallbladder polyps of 1-2 cm in diameter were enrolled and examined by CUS and CEUS before cholecystectomy. The appearances on CUS and CEUS were recorded and analyzed. The receiver operating characteristic (ROC) curve was used to calculate the optimal size threshold for distinguishing cholesterol from adenomatous polyps. A logistic regression analysis was performed to identify diagnostic variables. ROC analysis was performed to evaluate the diagnostic efficacy of the size, the independent variables, and the combined factors.

RESULTS

There were differences in size, number, vascularity on CUS and intralesional vascular shape, wash-out, and area under the curve on CEUS between the two groups (P < .05). ROC analysis indicated that a maximum diameter of 1.45 cm was the optimal threshold for the prediction of adenomatous polyps. The logistic regression analysis proved that the single polyp, presence of vascularity, and intralesional linear vessels were associated with adenomatous polyps (P < .05). ROC analysis showed that the area under the ROC curve, sensitivity, and specificity for the combination of the three independent variables were 0.858, 87.3%, and 67.6%. The number combined with intralesional vascular shape had the highest diagnostic sensitivity of 91.2%.

CONCLUSIONS

The combination of CUS and CEUS demonstrated great significance in the differential diagnosis of cholesterol and adenomatous polyps.

Triphasic dynamic enhanced computed tomography for differentiating cholesterol and adenomatous gallbladder polyps

BACKGROUND

Triphasic dynamic enhanced computed tomography (CT) scans were acquired to identify cholesterol and adenomatous gallbladder (GB) polyps that were inaccurately diagnosed before surgery.

PURPOSE

To evaluate the CT findings of 1.0- to 2.0-cm GB polyps for differentiating between cholesterol and adenomatous polyps.

METHODS

Fifty-two patients with GB polyps were treated surgically from December 2017 to July 2020 and were retrospectively divided into 2 groups according to the postoperative pathologic results: a cholesterol group with 30 patients and an adenomatous group with 22 patients. Unenhanced and triphasic dynamic enhanced CT scans were performed for all the patients within 2 weeks before surgery. The CT image parameters were measured and analyzed by 2 senior radiologists blinded to the pathological diagnoses.

RESULTS

Of the 22 patients in the adenomatous group, 77.3% were female and 22.7% were male, with a mean age of 53.5 years; among the 30 patients in the cholesterol group, 66.7% were female and 33.3% were male, with a median age of 50.1 years. The CT image parameters of all 52 patients with GB polyps were analyzed. Significant differences were found in the arterial phase CT values, portal venous phase CT values, delayed phase CT values, ∆CT1 values (portal venous phase CT minus delayed phase CT values), and ∆CT2 values (arterial phase CT minus delayed phase CT values) between the cholesterol and adenomatous polyp groups (p < 0.05). In differentiating the two groups, the ∆CT1 and ∆CT2 values were superior to the arterial, portal venous and delayed phase CT values regarding both sensitivity and specificity.

CONCLUSION

The arterial phase CT values, portal venous phase CT values, delayed phase CT values, and ∆CT values (including ∆CT1 and ∆CT2) from triphasic dynamic enhanced CT scans can differentiate the nature of gallbladder polypoid lesions, with the ∆CT values having the highest sensitivity and specificity.

Current update on gallbladder carcinoma

Gallbladder (GB) carcinoma is a relatively rare malignancy and is associated with poor prognosis. Numerous risk factors have been associated with the development of GB carcinoma. GB carcinomas may present as mass lesions replacing the GB, focal or diffuse thickening of the GB wall, and intraluminal mass in the GB. Various benign conditions can mimic GB carcinoma. This article reviews the epidemiology, pathology, clinical findings, imaging features, and management of GB carcinomas.

Gallbladder polyps: Correlation of size and clinicopathologic characteristics based on updated definitions

BACKGROUND

Different perspectives exist regarding the clinicopathologic characteristics, biology and management of gallbladder polyps. Size is often used as the surrogate evidence of polyp behavior and size of ≥1cm is widely used as cholecystectomy indication. Most studies on this issue are based on the pathologic correlation of polyps clinically selected for resection, whereas, the data regarding the nature of polypoid lesions from pathology perspective -regardless of the cholecystectomy indication- is highly limited.

METHODS

In this study, 4231 gallbladders -606 of which had gallbladder carcinoma- were reviewed carefully pathologically by the authors for polyps (defined as ≥2 mm). Separately, the cases that were diagnosed as "gallbladder polyps" in the surgical pathology databases were retrieved.

RESULTS

643 polyps identified accordingly were re-evaluated histopathologically. Mean age of all patients was 55 years (range: 20-94); mean polyp size was 9 mm. Among these 643 polyps, 223 (34.6%) were neoplastic: I. Non-neoplastic polyps (n = 420; 65.4%) were smaller (mean: 4.1 mm), occurred in younger patients (mean: 52 years). This group consisted of fibromyoglandular polyps (n = 196) per the updated classification, cholesterol polyps (n = 166), polypoid pyloric gland metaplasia (n = 41) and inflammatory polyps (n = 17). II. Neoplastic polyps were larger (mean: 21 mm), detected in older patients (mean: 61 years) and consisted of intra-cholecystic neoplasms (WHO's "adenomas" and "intracholecystic papillary neoplasms", ≥1 cm; n = 120), their "incipient" version (<1 cm) (n = 44), polypoid invasive carcinomas (n = 26) and non-neoplastic polyps with incidental dysplastic changes (n = 33). In terms of size cut-off correlations, overall, only 27% of polyps were ≥1 cm, 90% of which were neoplastic. All (except for one) ≥2 cm were neoplastic. However, 14% of polyps <1 cm were also neoplastic. Positive predictive value of ≥1 cm cut-off -which is widely used for cholecystectomy indication-, was 94.3% and negative predictive value was 85%.

CONCLUSIONS

Approximately a third of polypoid lesions in the cholecystectomies (regardless of the indication) prove to be neoplastic. The vast majority of (90%) of polyps ≥1 cm and virtually all of those ≥2 cm are neoplastic confirming the current impression that polyps ≥1 cm ought to be removed. However, this study also illustrates that 30% of the neoplastic polyps are <1 cm and therefore small polyps should also be closely watched, especially in older patients.

Quantitative contrast-enhanced US helps differentiating neoplastic vs non-neoplastic gallbladder polyps

OBJECTIVES

To differentiate between large (≥ 1 cm in diameter) gallbladder (GB) non-neoplastic and neoplastic polyps using quantitative analysis of contrast-enhanced ultrasound (CEUS) findings.

METHODS

From September 2017 to May 2018, 29 patients (10 males; median age, 63 years) with GB polyps of ≥ 1 cm in diameter who were undergoing cholecystectomy were consecutively enrolled. All patients underwent preoperative conventional US and CEUS examinations. Quantitative analysis of CEUS findings using time-intensity curves between the two groups was independently performed by two radiologists. The interobserver agreement for the quantitative analysis of the CEUS results was measured using the intraclass correlation coefficient. Receiver operating characteristic analysis was performed to evaluate the diagnostic performance of CEUS examination.

RESULTS

After the cholecystectomy, the patients were classified into the non-neoplastic polyp group (n = 12) and the neoplastic polyp group (n = 17) according to the pathological results. The interobserver agreement for quantitative assessment between the two radiologists was near perfect to substantial. Quantitative assessment of the CEUS findings revealed that the rise time, mean transit time, time to peak, and fall time of non-neoplastic GB polyps were significantly shorter than those of neoplastic polyps (p < 0.001, p = 0.008, p = 0.013, and p = 0.002, respectively). The sensitivity and specificity of the quantitative CEUS parameters for the differentiation between the two groups were 76.5-100% and 75%, respectively, with an area under the curve of 0.765-0.887.

CONCLUSIONS

Quantitative analysis of CEUS findings could be valuable in differentiating GB neoplastic polyps from non-neoplastic polyps.

KEY POINTS

• Quantitative analysis of CEUS findings could be valuable in differentiating gallbladder neoplastic polyps from non-neoplastic polyps. • Quantitative analysis of CEUS findings in gallbladder polyps provides cut-off values for differentiation between neoplastic polyps and non-neoplastic polyps with near-perfect to substantial interobserver agreement.

Differential diagnosis of gallbladder polypoid lesions using contrast-enhanced ultrasound

PURPOSE

The purpose of the study is to evaluate the utility of contrast-enhanced ultrasound (CEUS) for the differential diagnosis of gallbladder polypoid lesions (GPLs).

METHODS

Thirty-six patients with GPLs (17 with gallbladder cancer, 19 with benign polyps) who underwent CEUS were enrolled in the study. The mean age of patients was 65.7 ± 12.6 years. Perflubutane-based contrast agent and high-mechanical index mode, which can eliminate the background B-mode and provide precise visualization of tumor vessels, were used for CEUS, and two blinded readers evaluated the images, retrospectively.

RESULTS

Patient age and size of malignant GPLs (72.4 ± 9.4 years and 23.4 ± 7.5 mm) were significantly greater than those for benign lesions (59.6 ± 12.3 years and 12.4 ± 2.9 mm) (P < 0.01, respectively), and the receiver operating characteristic analysis showed the cut-off value as over 65 years and 16 mm. Univariate analysis showed that heterogeneity in B-mode (80% [12/15]), sessile shape (76% [13/17]), dilated vessel (71% [12/17]), irregular vessel (82% [14/17]), and heterogeneous enhancement (59% [10/17]) on CEUS were significantly correlated with malignant GPLs (P < 0.01, respectively). On CEUS, the diagnostic criterion for malignant GPLs was defined as having one or more of the above four features because of the highest accuracy. Sensitivity, specificity, and accuracy for malignant GBLs were 88%, 68%, and 78% for patient age; 76%, 89%, and 83% for size of GPLs; 80%, 68%, and 74% for B-mode; and 94%, 89%, and 92% for CEUS, respectively.

CONCLUSIONS

CEUS is useful for the differential diagnosis of malignant and benign GPLs.

Comet tail artifact on ultrasonography: is it a reliable finding of benign gallbladder diseases?

Purpose

The aim of this study was to evaluate whether the comet tail artifact on ultrasonography can be used to reliably diagnose benign gallbladder diseases.

Methods

This retrospective study reviewed the clinical findings, imaging findings, preoperative ultrasonographic diagnoses, and pathological diagnoses of 150 patients with comet tail artifacts who underwent laparoscopic cholecystectomy with pathologic confirmation. The extent of the involved lesion was classified as localized or diffuse, depending on the degree of involvement and the anatomical section of the gallbladder that was involved. This study evaluated the differences in clinical and imaging findings among pathologic diagnoses.

Results

All gallbladder lesions exhibiting the comet tail artifact on ultrasound examination were confirmed as benign gallbladder diseases after cholecystectomy, including 71 cases of adenomyomatosis (47.3%), 74 cases of chronic cholecystitis (49.3%), two cases of xanthogranulomatous cholecystitis (1.3%), and three cases of cholesterolosis (2.0%); there were two cases of coexistent chronic cholecystitis and low-grade dysplasia. There were no statistically significant differences in any of the clinical and ultrasonographic findings, with the exception of gallstones (P=0.007), among the four diseases. There were no significant differences in the average length, thickness, or number of comet tail artifacts among the four diagnoses. No malignancies were detected in any of the 150 thickened gallbladder lesions.

Conclusion

The ultrasonographic finding of the comet tail artifact in patients with thickened gallbladder lesions is associated with the presence of benign gallbladder diseases, and can be considered a reliable sign of benign gallbladder disease.

Diagnostic nomogram for gallbladder wall thickening mimicking malignancy: using contrast-enhanced ultrasonography or multi-detector computed tomography?

OBJECTIVES

To establish a diagnostic nomogram using contrast-enhanced ultrasonography (CEUS) in gallbladder wall thickening mimicking malignancy and compare with multi-detector computed tomography (MDCT).

METHODS

Seventy-two patients with gallbladder wall thickening on B-mode ultrasonography (BUS) were examined by CEUS to develop independent predictors for diagnosing gallbladder carcinoma. Among the 72 cases, 48 patients underwent both CEUS and MDCT. The diagnostic performances of different sets of CEUS criteria and MDCT were compared. A prediction model of malignancy using CEUS was developed. The performance of the nomogram was assessed with respect to its calibration, discrimination, and clinical usefulness.

RESULTS

Multivariate logistic regression indicated that inhomogeneous enhancement in the arterial phase was the strongest independent predictor of malignancy (odds ratio, OR 51.162), followed by interrupted inner layer (OR 19.788), washout time ≤40 s (OR 16.686), and wall thickness >1.6 cm (OR 3.019), which were all selected into the nomogram. Combined with the above significant features, the diagnostic performance of CEUS (AUC = 0.917) was higher than that of MDCT (AUC = 0.788, P = 0.070). The predictive model using CEUS showed good discrimination, with a concordance index of 0.974 (0.950 through internal validation), and good calibration. Decision curve analysis demonstrated that the nomogram was clinically useful.

CONCLUSIONS

CEUS could accurately differentiate between malignant and benign gallbladder wall thickening with equivalent efficacy compared to MDCT. The proposed nomogram could be conveniently used to facilitate the preoperative individualized prediction of malignancy in patients with gallbladder wall thickening.

Photoacoustic Imaging for Differential Diagnosis of Benign Polyps versus Malignant Polyps of the Gallbladder: A Preliminary Study

OBJECTIVE

To investigate the feasibility of ex vivo multispectral photoacoustic (PA) imaging in differentiating cholesterol versus neoplastic polyps, and benign versus malignant polyps, of the gallbladder.

MATERIALS AND METHODS

A total of 38 surgically confirmed gallbladder polyps (24 cholesterol polyps, 4 adenomas, and 10 adenocarcinomas) from 38 patients were prospectively included in this study. The surgical specimens were set on a gel pad immersed in a saline-filled container. The PA intensities of polyps were then measured, using two separate wavelength intervals (421-647 nm and 692-917 nm). Mann-Whitney U test was performed for the comparison of normalized PA intensities between the cholesterol and neoplastic polyps, and between the benign and malignant polyps. Kruskal-Wallis test was conducted for the comparison of normalized PA intensities among the cholesterol polyps, adenomas, and adenocarcinomas.

RESULTS

A significant difference was observed in the normalized PA intensities between the cholesterol and neoplastic polyps at 459 nm (median, 1.00 vs. 0.73; p = 0.032). Comparing the benign and malignant polyps, there were significant differences in the normalized PA intensities at 765 nm (median, 0.67 vs. 0.78; p = 0.013), 787 nm (median, 0.65 vs. 0.77; p = 0.034), and 853 nm (median, 0.59 vs. 0.85; p = 0.028). The comparison of the normalized PA intensities among cholesterol polyps, adenomas, and adenocarcinomas demonstrated marginally significant differences at 765 nm (median, 0.67 vs. 0.66 vs. 0.78, respectively; p = 0.049).

CONCLUSION

These preliminary results indicate that benign versus malignant gallbladder polyps might exhibit different spectral patterns on multispectral PA imaging.

Diffusion-weighted magnetic resonance imaging of gallbladder adenocarcinoma: analysis with emphasis on histologic grade

PURPOSE

To investigate the value of diffusion-weighted imaging (DWI) for differentiating gallbladder adenocarcinoma from adenoma, and predicting histologic grades of gallbladder adenocarcinoma.

MATERIALS AND METHODS

Fourty-three gallbladder adenocarcinomas and 8 adenomas were included. We compared apparent diffusion coefficient (ADC) values between adenocarcinoma and adenoma, and ADC values of gallbladder adenocarcinoma among the histologic grade.

RESULTS

Gallbladder adenocarcinoma (1.041×10(-3)mm(2)/s) showed significantly lower ADC values than adenoma (2.039×10(-3)mm(2)/s) (P<.001). Well-differentiated adenocarcinoma (1.290×10(-3)mm(2)/s) showed significantly higher ADC values than higher-grades (1.104×10(-3) and 0.915×10(-3)mm(2)/s in moderately- and poorly-differentiated, respectively) (P<.001).

CONCLUSION

DWI can help to differentiate gallbladder adenocarcinoma from adenoma, and well-differentiated from higher-grade adenocarcinoma.