Deep-learning enabled ultrasound based detection of gallbladder cancer in northern India: a prospective diagnostic study.

Background: Gallbladder cancer (GBC) is highly aggressive. Diagnosis of GBC is challenging as benign gallbladder lesions can have similar imaging features. We aim to develop and validate a deep learning (DL) model for the automatic detection of GBC at abdominal ultrasound (US) and compare its diagnostic performance with that of radiologists. Methods: In this prospective study, a multiscale, second-order pooling-based DL classifier model was trained (training and validation cohorts) using the US data of patients with gallbladder lesions acquired between August 2019 and June 2021 at the Postgraduate Institute of Medical Education and research, a tertiary care hospital in North India. The performance of the DL model to detect GBC was evaluated in a temporally independent test cohort (July 2021-September 2022) and was compared with that of two radiologists. Findings: The study included 233 patients in the training set (mean age, 48 ± (2SD) 23 years; 142 women), 59 patients in the validation set (mean age, 51.4 ± 19.2 years; 38 women), and 273 patients in the test set (mean age, 50.4 ± 22.1 years; 177 women). In the test set, the DL model had sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) of 92.3% (95% CI, 88.1-95.6), 74.4% (95% CI, 65.3-79.9), and 0.887 (95% CI, 0.844-0.930), respectively for detecting GBC which was comparable to both the radiologists. The DL-based approach showed high sensitivity (89.8-93%) and AUC (0.810-0.890) for detecting GBC in the presence of stones, contracted gallbladders, lesion size <10 mm, and neck lesions, which was comparable to both the radiologists (p = 0.052-0.738 for sensitivity and p = 0.061-0.745 for AUC). The sensitivity for DL-based detection of mural thickening type of GBC was significantly greater than one of the radiologists (87.8% vs. 72.8%, p = 0.012), despite a reduced specificity. Interpretation: The DL-based approach demonstrated diagnostic performance comparable to experienced radiologists in detecting GBC using US. However, multicentre studies are warranted to explore the potential of DL-based diagnosis of GBC fully. Funding: None.

Agreement of Gallbladder Reporting and Data System for Gallbladder Wall Thickening at Ultrasonography: A Multireader Validation Study.

Objective: This article aims to evaluate the intrareader and interreader agreement of ultrasound (US) gallbladder reporting and data system (GB-RADS) and validate the risk of malignancy in each GB-RADS category. Materials and methods: This retrospective study comprised consecutive patients with nonacute gallbladder wall thickening who underwent US evaluation between January 2019 and December 2022. Three radiologists independently read the static US images and cine-loops for GB-RADS findings and assigned GB-RADS categories. The intraobserver (static images) and interobserver (static images and cine-loops) agreement was calculated using kappa statistics and Krippendorff's alpha. Another radiologist assigned a consensus GB-RADS category. The percentage of malignancy in each GB-RADS category was calculated. Results: Static US images of 414 patients (median age, 56 years; 288 women, benign = 45.6% and malignant = 54.4%) and cine-loops of 50 patients were read. There was weak to moderate intrareader agreement for most GB-RADS findings and moderate intrareader agreement for the GB-RADS category for all readers. On static images, the interreader agreement was acceptable for GB-RADS categories. On cine-loops, the interreader agreement for GB-RADS findings and categories was better than static images. The percentage of malignancy was 1.2%, 37%, 71.1%, and 89.1% in GB-RADS 2, 3, 4, and 5 categories. Conclusion: GB-RADS has moderate intrareader for GB-RADS categories. As originally proposed, the risk of malignancy is negligible in GB-RADS 2 category and highest in GB-RADS 5 category. However, the discriminatory performance of GB-RADS 3 and 4 categories is low. Larger multicenter studies with more readers must assess the reader agreement and validate the GB-RADS systems for wider clinical utilization.

Sonographic "Cervix Sign": A New Ancillary Sign of Gallbladder Neck Malignancy.

Background: The differentiation of benign and malignant gallbladder wall thickening is challenging. The purpose of this study is to evaluate a new sonographic sign, "cervix sign" for differentiation of benign and malignant gallbladder neck thickening. Methods: This retrospective study comprised consecutive patients with gallbladder neck thickening who underwent sonography between August 2019 and December 2021. The presence of "cervix sign" was assessed by two radiologists independently. Results: Sixty-five patients had gallbladder neck thickening (28 malignant and 37 benign). The sonographic "cervix sign" was present in 18 (64%) patients with malignant thickening and in only one (2.7%) patient with benign thickening (P = 0.0001). The mean wall thickness was greater, and symmetric wall thickening and liver metastases were more common in malignant thickening with "cervix sign" (without reaching statistical significance). There was substantial agreement (kappa = 0.78) between the two observers for the cervix sign. Conclusion: Sonographic "cervix sign" is a useful ancillary feature of gallbladder neck cancer.

Gastrointestinal involvement in gallbladder cancer: Computed tomography findings and proposal of a classification system.

BACKGROUND: There is relatively scarce data on the computed tomography (CT) detection of gastrointestinal (GI) involvement in gallbladder cancer (GBC). We aim to assess the GI involvement in GBC on CT and propose a CT-based classification. METHODS: This retrospective study comprized consecutive patients with GBC who underwent contrast-enhanced computed tomography (CECT) for staging between January 2019 and April 2022. Two radiologists evaluated the CT images independently for the morphological type of GBC and the presence of GI involvement. GI involvement was classified into probable involvement, definite involvement and GI fistulization. The incidence of GI involvement and the association of GI involvement with the morphological type of GBC was evaluated. In addition, the inter-observer agreement for GI involvement was assessed. RESULTS: Over the study period, 260 patients with GBC were evaluated. Forty-three (16.5%) patients had GI involvement. Probable GI involvement, definite GI involvement and GI fistulization were seen in 18 (41.9%), 19 (44.2%) and six (13.9%) patients, respectively. Duodenum was the most common site of involvement (55.8%), followed by hepatic flexure (23.3%), antropyloric region (9.3%) and transverse colon (2.3%). There was no association between GI involvement and morphological type of GBC. There was substantial to near-perfect agreement between the two radiologists for the overall GI involvement (k = 0.790), definite GI involvement (k = 0.815) and GI fistulization (k = 0.943). There was moderate agreement (k = 0.567) for probable GI involvement. CONCLUSION: GBC frequently involves the GI tract and CT can be used to categorize the GI involvement. However, the proposed CT classification needs validation.

Impact of Radial Percentage K-Space Filling and Signal Averaging on Native Lung MRI Image Quality in 3D Radial UTE Acquisition: A Pilot Study.

RATIONALE AND OBJECTIVE: To assess the impact of radial percentage k-space filling and signal averaging on lung MRI image quality in 3D radial ultrashort echo-time (UTE) acquisition. MATERIALS AND METHODS: In this IRB approved prospective study, 25 patients (10-30 years) referred for MRI examination for indications other than related to lungs were enrolled from January 2021 to November 2021. All the patients underwent lung MRI, using three different UTE sequence parameters with radial (R) percentage of 100 or 200 and number of signal averages (NSA) of one or two. Two radiologists independently assessed the images for the outline of pleural and mediastinal surface, visibility of lung parenchyma, major bronchi, and segmental bronchi. The quality of the images was assessed based on the degree of motion artifacts. For objective assessment, signal-to-noise ratio, contrast-to-noise ratio, and contrast ratio were calculated. RESULTS: The outline of pleural and mediastinal surface, lung parenchyma, and segmental bronchi were best demonstrated on R100_NSA2 sequence. The major bronchi were best demonstrated on R100_NSA2 and R100_NSA1 sequences. The intersequence difference was statistically significant for evaluating the pleural and mediastinal surface and segmental bronchi only (p < 0.05). Overall, the best image quality with least artifacts was seen with R100_NSA2 sequence. The objective assessment showed no statistically significant difference between the three sequences (p > 0.05). Interobserver agreement for different findings was substantial to almost perfect for R100_NSA2 and R200_NSA1 sequences. CONCLUSION: R100_NSA2 UTE sequence performed best for the evaluation of the different findings and showed the best image quality.

RadFormer: Transformers with global-local attention for interpretable and accurate Gallbladder Cancer detection.

We propose a novel deep neural network architecture to learn interpretable representation for medical image analysis. Our architecture generates a global attention for region of interest, and then learns bag of words style deep feature embeddings with local attention. The global, and local feature maps are combined using a contemporary transformer architecture for highly accurate Gallbladder Cancer (GBC) detection from Ultrasound (USG) images. Our experiments indicate that the detection accuracy of our model beats even human radiologists, and advocates its use as the second reader for GBC diagnosis. Bag of words embeddings allow our model to be probed for generating interpretable explanations for GBC detection consistent with the ones reported in medical literature. We show that the proposed model not only helps understand decisions of neural network models but also aids in discovery of new visual features relevant to the diagnosis of GBC. Source-code is available at https://github.com/sbasu276/RadFormer.

Contrast-enhanced ultrasound of solid pancreatic head lesions: a prospective study.

OBJECTIVE: To evaluate the role of contrast-enhanced ultrasound (CEUS) in the differential diagnosis of solid pancreatic head lesions (SPHL). METHODS: This prospective study comprised consecutive patients with SPHL who underwent CEUS evaluation of the pancreas. Findings recorded at CEUS were enhancement patterns (degree, completeness, centripetal enhancement, and percentage enhancement) and presence of central vessels. In addition, time to peak (TTP) and washout time (WT) were recorded. The final diagnosis was based on histopathology or cytology. Multivariate analysis was performed to identify parameters that were significantly associated with pancreatic ductal adenocarcinoma (PDAC). RESULTS: Ninety-eight patients (median age 53.8 years, 59 males) were evaluated. The final diagnosis was PDAC (n = 64, 65.3%), inflammatory mass (n = 16, 16.3%), neuroendocrine tumor (NET, n = 14, 14.3%), and other tumors (n = 4, 4.1%). Hypoenhancement, incomplete enhancement, and centripetal enhancement were significantly more common in PDAC than non-PDAC lesions (p = 0.001, p = 0.031, and p = 0.002, respectively). Central vessels were present in a significantly greater number of non-PDAC lesions (p = 0.0001). Hypoenhancement with < 30% enhancement at CEUS had sensitivity and specificity of 80.6% and 67.7%, respectively, for PDAC. There was no significant difference in the TTP and WT between PDAC and non - PDAC lesions. However, the WT was significantly shorter in PDAC compared to NET (p = 0.011). In multivariate analysis, lack of central vessels was significantly associated with a PDAC diagnosis. CONCLUSION: CEUS is a useful tool for the evaluation of SPHL. CEUS can be incorporated into the diagnostic algorithm to differentiate PDAC from non-PDAC lesions. KEY POINTS: • Hypoenhancement and incomplete enhancement at CEUS were significantly more common in PDAC than in non-PDAC. • Central vessels at CEUS were significantly associated with PDAC. • There was no difference in TTP and WT between PDAC and non-PDAC lesions.

Role of dual energy computed tomography in evaluation of suspected wall thickening type of gallbladder cancer.

Introduction: Accurate characterization of gallbladder wall thickening is of paramount importance. This study evaluated the role of dual-energy computed tomography (DECT) in the evaluation of gallbladder wall thickening. Material and methods: Seven consecutive patients (mean age 52.3 years, 5 males) with suspected wall thickening type of gallbladder cancer (GBC) who underwent DECT between August 2020 and March 2021 were included. Independent radiologists blinded to the final diagnosis reviewed the iodine maps, 80 keV, and 140 keV images separately. The final diagnosis was based on histopathology or cytology. Results: Two patients had xanthogranulomatous cholecystitis and 5 had GBC. Asymmetrical mural thickening was identified in six patients on iodine maps and five patients at 80 keV. Non-layered mural thickening was seen in all patients on iodine maps, 80 keV, and 140 keV. Iodine maps identified heterogeneous enhancement in all patients. Conclusions: These preliminary results suggest that DECT has the potential to characterize gallbladder wall thickening.

Large collateral from the right coronary artery in a case of coarctation of the aorta with bilateral subclavian artery aneurysms in an adult.

Here, we present a case of a 26-year-old gentleman who presented with a rare combination of bilateral huge subclavian artery aneurysms, coarctation of the aorta with a large coronary artery collateral and a bovine aortic arch. Various mediastinal collaterals have been described in previous cases of coarctation of the aorta, but the association of coronary collaterals is a rare occurrence. To the best of our knowledge, this case has not been reported in literature yet.

A six-chambered heart: cor triatriatum sinister with double-chambered right ventricle in association with ventricular septal defect.

BACKGROUND: Cor triatriatum has been described as a heart with three atria in which the left atrium (cor triatriatum sinistrum) or right atrium (cor triatriatum dextrum) is divided into two compartments by a fold of tissue, a membrane, or a fibromuscular band. Double-chambered right ventricle, on the other hand, is identified by the presence of an anomalous muscle bundle dividing the right ventricle into two chambers. CASE PRESENTATION: Here, we describe the case of a child who had a combination of both of these rare entities, effectively creating a heart with six chambers. The child underwent a successful intracardiac repair. CONCLUSIONS: The association of CTS with DCRV forming a "6-chambered heart" is extremely rare. Awareness of its existence and accurate preoperative diagnosis has important implications in its surgical repair with all the components of this disease spectrum, further increasing the complexity of a successful surgical repair.

Isolated ventricular inversion with hypoplastic right ventricle and pulmonary stenosis in an adolescent with situs solitus: a rare combination.

Isolated ventricular inversion with situs solitus is a severe and rare congenital cardiac malformation characterised by an atrioventricular discordance but with ventriculo-arterial concordance. Here, we present the rare case of an adolescent with isolated ventricular inversion and hypoplasia of the left-sided morphological right ventricle and pulmonary stenosis, a first of its kind to be reported in the literature.