Performance of explainable artificial intelligence in guiding the management of patients with a pancreatic cyst.

BACKGROUND/OBJECTIVES: Pancreatic cyst management can be distilled into three separate pathways - discharge, monitoring or surgery- based on the risk of malignant transformation. This study compares the performance of artificial intelligence (AI) models to clinical care for this task. METHODS: Two explainable boosting machine (EBM) models were developed and evaluated using clinical features only, or clinical features and cyst fluid molecular markers (CFMM) using a publicly available dataset, consisting of 850 cases (median age 64; 65 % female) with independent training (429 cases) and holdout test cohorts (421 cases). There were 137 cysts with no malignant potential, 114 malignant cysts, and 599 IPMNs and MCNs. RESULTS: The EBM and EBM with CFMM models had higher accuracy for identifying patients requiring monitoring (0.88 and 0.82) and surgery (0.66 and 0.82) respectively compared with current clinical care (0.62 and 0.58). For discharge, the EBM with CFMM model had a higher accuracy (0.91) than either the EBM model (0.84) or current clinical care (0.86). In the cohort of patients who underwent surgical resection, use of the EBM-CFMM model would have decreased the number of unnecessary surgeries by 59 % (n = 92), increased correct surgeries by 7.5 % (n = 11), identified patients who require monitoring by 122 % (n = 76), and increased the number of patients correctly classified for discharge by 138 % (n = 18) compared to clinical care. CONCLUSIONS: EBM models had greater sensitivity and specificity for identifying the correct management compared with either clinical management or previous AI models. The model predictions are demonstrated to be interpretable by clinicians.

A Confirmed Extrarenal Birt-Hogg-Dubé-Associated Oncocytic Neoplasm.

Birt-Hogg-Dubé (BHD) syndrome is a rare inherited disease characterized by a variety of renal epithelial tumors and oncocytosis, with extrarenal manifestations primarily consisting of pulmonary cysts and cutaneous fibrofolliculomas. Here we report a unique case of a primary extrarenal BHD-associated oncocytic epithelial neoplasm which arose between the duodenum and head of the pancreas. The unusual morphology and immunoprofile of this lesion defied classification as any previously reported entity, despite an extensive diagnostic workup. The immunohistochemical and molecular features indicate the tumor was driven by FLCN loss, and thus a consequence of the underlying germline mutation with a somatic second hit. This tumor is the first reported example of an extrarenal BHD-associated oncocytic epithelial tumor driven by FLCN loss.

Radiomics machine learning algorithm facilitates detection of small pancreatic neuroendocrine tumors on CT.

PURPOSE: The purpose of this study was to develop a radiomics-based algorithm to identify small pancreatic neuroendocrine tumors (PanNETs) on CT and evaluate its robustness across manual and automated segmentations, exploring the feasibility of automated screening. MATERIALS AND METHODS: Patients with pathologically confirmed T1 stage PanNETs and healthy controls undergoing dual-phase CT imaging were retrospectively identified. Manual segmentation of pancreas and tumors was performed, then automated pancreatic segmentations were generated using a pretrained neural network. A total of 1223 radiomics features were independently extracted from both segmentation volumes, in the arterial and venous phases separately. Ten final features were selected to train classifiers to identify PanNETs and controls. The cohort was divided into training and testing sets, and performance of classifiers was assessed using area under the receiver operator characteristic curve (AUC), specificity and sensitivity, and compared against two radiologists blinded to the diagnoses. RESULTS: A total of 135 patients with 142 PanNETs, and 135 healthy controls were included. There were 168 women and 102 men, with a mean age of 55.4 ± 11.6 (standard deviation) years (range: 20-85 years). Median PanNET size was 1.3 cm (Q1, 1.0; Q3, 1.5; range: 0.5-1.9). The arterial phase LightGBM model achieved the best performance in the test set, with 90 % sensitivity (95 % confidence interval [CI]: 80-98), 76 % specificity (95 % CI: 62-88) and an AUC of 0.87 (95 % CI: 0.79-0.94). Using features from the automated segmentations, this model achieved an AUC of 0.86 (95 % CI: 0.79-0.93). In comparison, the two radiologists achieved a mean 50 % sensitivity and 100 % specificity using arterial phase CT images. CONCLUSION: Radiomics features identify small PanNETs, with stable performance when extracted using automated segmentations. These models demonstrate high sensitivity, complementing the high specificity of radiologists, and could serve as opportunistic screeners.

Diminishing calcifications as a potential predictor of pancreatic ductal adenocarcinoma arising in association with IPMN in patients with chronic pancreatitis.

Chronic pancreatitis (CP) is a progressive benign fibroinflammatory condition involving repeated episodes of pancreatic inflammation, which lead to fibrotic tissue replacement and subsequent pancreatic insufficiency. A lifetime risk of developing pancreatic ductal adenocarcinoma (PDAC) in patients with chronic pancreatitis is reported to be 1.5%-4%. However, diagnosis of PDAC in patients with CP can be challenging, in part due to overlapping imaging features. In rare instances, pancreatic parenchymal calcifications that are typically associated with chronic pancreatitis may diminish in the case of a developing PDAC. In this article, we present a patient with chronic pancreatitis in whom calcifications decreased at the time of pancreatic ductal adenocarcinoma diagnosis, as compared to prior CT imaging. The unique imaging features of "diminishing calcifications" associated with a hypoattenuating lesion can potentially be a useful sign of pancreatic ductal adenocarcinoma and may aid in early diagnosis and prompt treatment intervention.

Preoperative Prediction of Lymph Node Metastases in Nonfunctional Pancreatic Neuroendocrine Tumors Using a Combined CT Radiomics-Clinical Model.

BACKGROUND: PanNETs are a rare group of pancreatic tumors that display heterogeneous histopathological and clinical behavior. Nodal disease has been established as one of the strongest predictors of patient outcomes in PanNETs. Lack of accurate preoperative assessment of nodal disease is a major limitation in the management of these patients, in particular those with small (< 2 cm) low-grade tumors. The aim of the study was to evaluate the ability of radiomic features (RF) to preoperatively predict the presence of nodal disease in pancreatic neuroendocrine tumors (PanNETs). PATIENTS AND METHODS: An institutional database was used to identify patients with nonfunctional PanNETs undergoing resection. Pancreas protocol computed tomography was obtained, manually segmented, and RF were extracted. These were analyzed using the minimum redundancy maximum relevance analysis for hierarchical feature selection. Youden index was used to identify the optimal cutoff for predicting nodal disease. A random forest prediction model was trained using RF and clinicopathological characteristics and validated internally. RESULTS: Of the 320 patients included in the study, 92 (28.8%) had nodal disease based on histopathological assessment of the surgical specimen. A radiomic signature based on ten selected RF was developed. Clinicopathological characteristics predictive of nodal disease included tumor grade and size. Upon internal validation the combined radiomics and clinical feature model demonstrated adequate performance (AUC 0.80) in identifying nodal disease. The model accurately identified nodal disease in 85% of patients with small tumors (< 2 cm). CONCLUSIONS: Non-invasive preoperative assessment of nodal disease using RF and clinicopathological characteristics is feasible.

Flank pain, hypertension, and hematuria: CT and 3D cinematic rendering in the evaluation of renal artery emergencies-a pictorial essay.

Non-traumatic acute renal artery emergencies encompass a spectrum of etiologies, including renal artery stenosis, arteriovenous malformations, aneurysms and pseudoaneurysms, dissections, thrombosis, and vasculitis. Prompt and accurate diagnosis in the emergency setting is crucial due to the potential for significant morbidity and mortality. Computed tomography (CT) and CT angiography (CTA) are the mainstay imaging modalities, offering rapid acquisition and high diagnostic accuracy. The integration of 3D postprocessing techniques, such as 3D cinematic rendering (CR), improves the diagnostic workflow by providing photorealistic and anatomically accurate visualizations. This pictorial essay illustrates the diagnostic utility of CT and CTA, supplemented by 3D CR, through a series of 10 cases of non-traumatic renal artery emergencies. The added value of 3D CR in improving diagnostic confidence, surgical planning, and understanding of complex vascular anatomy is emphasized.

Instagram reels versus image posts in radiology education.

OBJECTIVE: In January 2016, we created an Instagram page for radiology education. Numerous publications in different fields have reported that Instagram "reels," introduced in 2020 as a short-form video feature, are more popular than image posts. These findings and our familiarity with Instagram prompted us to analyze our own data to better understand how image posts compared with reels when used in the context of radiology education. MATERIALS AND METHODS: For each post category, metric values were extracted from the Instagram platform and analyzed as continuous variables, reported as medians with interquartile ranges (IQR). Metrics were compared between image categories using the Kruskal-Wallis test, with resulting p-values adjusted for multiple comparisons using the Bonferroni correction. Corrected p-values of less than 0.05 were considered statistically significant. RESULTS: We included 128 images and 96 reels in the analysis. Images generally reached a larger audience, with a median of 18,745 [IQR: 13,478-27,243] impressions vs. 11,972 [IQR: 9,310.0-13,844.5] for reels (p < 0.01). Images also tended to be shared more frequently (median 19 vs. 20, p < 0.01), liked more often (median 480 vs. 296, p < 0.01), and saved more by users (median 138 vs. 84, p < 0.01) than reels, respectively. Both images and reels received a similar number of comments, with a median of 3 comments for both (p > 0.99). We also explored the performance differences of image post subcategories. Within images, our "You Make the Call!" (YMTC) questions (n = 23) displayed higher performance metrics across the board than the three other types of image posts combined (n = 105). When compared, the median number of impressions for YMTC images was 36,735 [IQR: 31,343-40,742] vs. 15,992 [IQR:12,774-21,873] for other types of images (p < 0.01). YMTC images were shared more often (median 25 vs. 17, p < 0.01), received more likes (median 809 vs. 445, p < 0.01) and saves (median 206 vs. 119, p < 0.01) than non-YMTC images, respectively. User engagement showed slightly different trends with YMTC reels being the most liked, while quiz reels receiving the most comments and talking clips being the most saved. CONCLUSION: Our findings on the use of Instagram in radiology education suggest that static images perform much better than reels. Consequently, we recommend to radiology educators seeking to establish an Instagram presence that using static image posts is an appropriate approach for reaching a radiology audience, particularly with image posts that engage an audience with participatory opportunities such as answering quiz-like questions aimed at making a diagnosis.

Deep pelvic endometriosis causing ureteral obstruction.

Endometriosis is a chronic disease characterized by the presence and growth of endometrial glands and stroma outside of the uterine cavity. The pathogenesis is unclear, but a common theory attributes the condition to retrograde menstruation into the peritoneal cavity via the fallopian tubes. Hormonal influence causes these ectopic tissues to undergo cyclical bleeding, resulting in subsequent inflammation and scar tissue formation; however, it can affect postmenopausal women. In rare instances, endometriotic lesions can obstruct the ureter and result in hydroureteronephrosis and subsequent loss of renal function. This condition presents with nonspecific symptoms and is known as an often-silent disease, resulting in challenging and delayed preoperative diagnosis. In this article, we report the case of an asymptomatic 65-year old female who was diagnosed with deep pelvic endometriosis, which obstructed the right distal ureter. We focus on optimizing diagnosis and management through the application of radiological imaging modalities, specifically computed tomography (CT) and magnetic resonance imaging (MRI).

Websites, mobile apps, and social media: Premier online educational tools for radiology.

Demand for online educational tools has risen steadily as technological innovations have evolved over the past several decades. Websites were the first platform to be introduced, and eventually used for online schooling, soon after the advent of the World Wide Web. Access to information and updated content in a short period of time on a wide-screen device such as a computer made websites popular early in their development. With the technological revolution of smart phones, mobile applications have been developed on various operating systems and, through this progress, a new form of educational platform was initiated. The portable features of mobile applications represent a pioneer era of educational tools for medical professionals. Online communications have transformed into social media over the last decade and have since been adopted by much of the world. All three of these educational platforms have created a significant impact on medical education communities, specifically in radiology. We describe the relative strengths of each platform and illustrate how our experience over more than two decades guides our recommendations.

Pancreatic Cancer Surveillance and Survival of High-Risk Individuals.

Importance: Pancreatic ductal adenocarcinoma (PDAC) is a deadly disease with increasing incidence. The majority of PDACs are incurable at presentation, but population-based screening is not recommended. Surveillance of high-risk individuals for PDAC may lead to early detection, but the survival benefit is unproven. Objective: To compare the survival of patients with surveillance-detected PDAC with US national data. Design, Setting, and Participants: This comparative cohort study was conducted in multiple US academic medical centers participating in the Cancer of the Pancreas Screening program, which screens high-risk individuals with a familial or genetic predisposition for PDAC. The comparison cohort comprised patients with PDAC matched for age, sex, and year of diagnosis from the Surveillance, Epidemiology, and End Results (SEER) program. The Cancer of the Pancreas Screening program originated in 1998, and data collection was done through 2021. The data analysis was performed from April 29, 2022, through April 10, 2023. Exposures: Endoscopic ultrasonography or magnetic resonance imaging performed annually and standard-of-care surgical and/or oncologic treatment. Main Outcomes and Measures: Stage of PDAC at diagnosis, overall survival (OS), and PDAC mortality were compared using descriptive statistics and conditional logistic regression, Cox proportional hazards regression, and competing risk regression models. Sensitivity analyses and adjustment for lead-time bias were also conducted. Results: A total of 26 high-risk individuals (mean [SD] age at diagnosis, 65.8 [9.5] years; 15 female [57.7%]) with PDAC were compared with 1504 SEER control patients with PDAC (mean [SD] age at diagnosis, 66.8 [7.9] years; 771 female [51.3%]). The median primary tumor diameter of the 26 high-risk individuals was smaller than in the control patients (2.5 [range, 0.6-5.0] vs 3.6 [range, 0.2-8.0] cm, respectively; P < .001). The high-risk individuals were more likely to be diagnosed with a lower stage (stage I, 10 [38.5%]; stage II, 8 [30.8%]) than matched control patients (stage I, 155 [10.3%]; stage II, 377 [25.1%]; P < .001). The PDAC mortality rate at 5 years was lower for high-risk individuals than control patients (43% vs 86%; hazard ratio, 3.58; 95% CI, 2.01-6.39; P < .001), and high-risk individuals lived longer than matched control patients (median OS, 61.7 [range, 1.9-147.3] vs 8.0 [range, 1.0-131.0] months; 5-year OS rate, 50% [95% CI, 32%-80%] vs 9% [95% CI, 7%-11%]). Conclusions and Relevance: These findings suggest that surveillance of high-risk individuals may lead to detection of smaller, lower-stage PDACs and improved survival.

Lessons learned from rebranding a radiology education facebook page.

OBJECTIVE: In May 2009, we created a Facebook page for radiology education. While we shared a host of learning materials such as case images, quiz questions, and medical illustrations, we also posted world news, music, and memes. In February 2023, we eliminated everything from the site not related to radiology education. Our aim was to determine how focusing on radiology education alone would affect audience growth for our Facebook page. MATERIALS AND METHODS: We exported our Facebook post data for the dates March 1, 2023 through February 29, 2024, to represent the full calendar year after we revised our content presentation, which we compared to data from November 1, 2020 to October 31, 2021. The mean and standard deviation for each post type's reach for 2023/24 were analyzed and compared against the 2020/21 statistics, and Wilcoxon rank sum tests were used to obtain p-values. Linear regressions for each year were performed to understand the relationship between reach and engagement. RESULTS: A total of 4,270 posts were included in our new analysis. Our average number of posts per day decreased from 24.8 to 11.71, reducing by more than half the amount of content shared to our social media page. Our posts had a mean overall reach of 4,660-compared to 1,743 in 2021 (p=0.0000). There was a statistically significant increase in reach for posts on artificial intelligence, case images, medical illustrations, pearls, quiz images, quiz videos, slideshow images, and both types of instructional videos (p<0.005). For both 2021 and 2024, the linear regression slopes were positive (y=0.0687x-65.0279 and y=0.006334x+21.3425, respectively). CONCLUSIONS: Facebook and other social media have been found to be helpful sources for radiology education. Our experience and statistics with radiology education via social media may help other radiology educators better curate their own pages. To optimize experiences for students, professionals, and other users, and to reach more people, we found that providing readily accessible radiology education is preferred to the social aspects of social media.

Adrenal cysts less than 10 cm can be safely observed.

BACKGROUND: Adrenal cysts are rare and appropriate management is unclear due to a lack of data on their natural history. Understanding adrenal cyst growth patterns would assist in clinical management. METHODS: This single-institution study included all adult patients diagnosed with simple adrenal cysts between 2004 and 2021. Baseline characteristics and outcomes of those who underwent resection (ADX) or observation (OBS) were compared using the chi-squared test, student's t-test, and Wilcoxon rank-sum test. Growth curves and sensitivity analysis were plotted for all patients who had follow-up imaging. RESULTS: We identified 77 patients with imaging-confirmed adrenal cysts. The majority were female (75.3%) and more than half were white (55.8%). One-third of patients underwent ADX, and the remaining were observed. ADX patients were younger (median age [IQR]: 55.5 y [45.0-68.2 y] vs. 44.2 y [38.7-55.0 y], p = 0.01) and more likely to be Hispanic (12% vs. 0%, p = 0.05). ADX patients presented with larger cysts (5.6 vs. 2.6 cm, p = 0.002). The median time from diagnosis to last follow-up was 1.1 y for ADX and 4.1 y for OBS. Average growth for OBS was 0.3 cm/y, while average growth for ADX was 3.9 cm/y. In ADX patients, cysts >10 cm grew significantly faster than cysts <10 cm (median growth rate 13.2 cm/y vs. 0.3 cm/y, p < 0.05). There was no adrenal malignancy diagnosis, hyperfunctionality, or observation-related complications (e.g., rupture). CONCLUSION: While size >4-6 cm has guided surgical referral for solid adrenal masses, this study demonstrates a size threshold of 10 cm, below which asymptomatic, simple adrenal cysts can safely be observed.

Outcomes Following a False-Positive Multi-Cancer Early Detection Test: Results from DETECT-A, the First Large, Prospective, Interventional MCED Study.

Guideline recommended standard of care screening is available for four cancer types; most cancer-related deaths are caused by cancers without standard of care screening. DETECT-A is the first prospective interventional trial evaluating a multi-cancer early detection (MCED) blood test (CancerSEEK) in women without a history of cancer, providing the first opportunity to assess the long-term outcomes of individuals with false-positive (FP) MCED results. This prospective analysis of DETECT-A participants with FP results evaluates the performance of an imaging-based diagnostic workflow and examines cancer risk following a FP result. This analysis included all DETECT-A participants with a positive CancerSEEK test and subsequent flourine-18 fluorodeoxyglucose positron emission tomography-IV contrast-enhanced computed tomography (18-F-FDG PET-CT) imaging and clinical workup indicating no evidence of cancer within 1 year of enrollment (n = 98). Medical records, study interactions, and study surveys were used to assess cancer incidence, treatments, and clinical outcomes through August 2023. Ninety-five of 98 participants with a FP result remained cancer-free with a median follow-up of 3.6 years (IQR: 2.5-4.1) from determination of FP status. Three incident cancers were observed over the follow-up period. One bilateral stage IIIC ovarian cancer was diagnosed 1.9 years after determination of FP status; two stage I breast cancers were diagnosed 0.1 and 1.6 years from determination of FP status. The annual incidence rate of cancer during follow-up from FP determination was 1.0% (95% confidence interval, 0.2%-2.8%). Participants with a positive CancerSEEK test who underwent 18-F-FDG PET-CT and clinical workup without cancer findings had low risk for cancer over the following several years. Prevention Relevance: This study provides multiyear clinical outcomes data following a false-positive multi-cancer early detection test for individuals participating in a prospective interventional trial. It provides a preliminary performance assessment of an imaging-based diagnostic workflow following a false-positive multi-cancer early detection test.

Cross-sectional imaging of mimics of inflammatory bowel disease: not everything is Crohn's disease or ulcerative colitis.

Acute and chronic bowel pathologies can often be mistaken for manifestations of inflammatory bowel disease (IBD), and there are many entities with imaging and clinical features that overlap with IBD, making diagnosis difficult. We describe multiple inflammatory, infectious, neoplastic, and vascular entities with imaging and clinical features that may mimic IBD, and highlight differentiating features to assist in diagnosis.