Citations

Radiology is undergoing a paradigm shift from traditional single-function AI systems to sophisticated multi-agent networks capable of autonomous reasoning, coordinated decision-making, and adaptive workflow management. These agentic AI systems move beyond simple pattern recognition to encompass complex radiological workflows including image analysis, report generation, clinical communication, and care coordination. While multi-agent radiological AI promises enhanced diagnostic accuracy, improved workflow efficiency, and reduced physician burden, it simultaneously amplifies the long-standing “black box” problem. Traditional explainable AI methods, which are adequate for understanding isolated diagnostic predictions, fail when applied to multi-step reasoning processes involving multiple specialized agents coordinating across imaging interpretation, clinical correlation, and treatment planning. This paper examines how agentic AI systems in radiology create “compound opacity” layers of inscrutability from agent interactions and distributed decision-making processes. We analyze the autonomy–transparency paradox specific to radiological practice, where increasing AI capability directly conflicts with interpretability requirements essential for clinical trust and regulatory oversight. Through examination of emerging multi-agent radiological workflows, we propose frameworks for responsible implementation that preserve both diagnostic innovation and the fundamental principles of medical transparency and accountability.

This review explores how multimodal foundation models (MFMs) are transforming biliary tract cancer (BTC) research. BTCs are aggressive malignancies with poor prognosis, presenting unique challenges due to difficult diagnostic methods, molecular complexity, and rarity. Importantly, intrahepatic cholangiocarcinoma (iCCA), perihilar cholangiocarcinoma (pCCA), and distal bile duct cholangiocarcinoma (dCCA) represent fundamentally distinct clinical entities, with iCCA presenting as mass-forming lesions amenable to biopsy and targeted therapies, while pCCA manifests as infiltrative bile duct lesions with challenging diagnosis and primarily palliative management approaches. MFMs offer potential to advance research by integrating radiological images, histopathology, multi-omics profiles, and clinical data into unified computational frameworks, with applications tailored to these distinct BTC subtypes. Key applications include enhanced biomarker discovery that identifies previously unrecognizable cross-modal patterns, potential for improving currently limited diagnostic accuracy—though validation in BTC-specific cohorts remains essential—accelerated drug repurposing, and advanced patient stratification for personalized treatment. Despite promising results, challenges such as data scarcity, high computational demands, and clinical workflow integration remain to be addressed. Future research should focus on standardized data protocols, architectural innovations, and prospective validation studies. The integration of artificial intelligence (AI)-based methodologies offers new solutions for these historically challenging malignancies. However, current evidence for BTC-specific applications remains largely theoretical, with most studies limited to proof-of-concept designs or related cancer types. Comprehensive clinical validation studies and prospective trials demonstrating patient benefit are essential prerequisites for clinical implementation. The timeline for evidence-based clinical adoption likely extends 7–10 years, contingent on successful completion of validation studies addressing current evidence gaps.

Key Points Automated cyst segmentation in autosomal dominant polycystic kidney disease could transform care by enabling precise monitoring of cyst progression with minimal manual effort. This technology can facilitate more consistent assessments of cyst burden across patients. Translating these segmentation models from magnetic resonance imaging to computed tomography is crucial for providing comprehensive, modality-independent evaluations. Background Total kidney and liver volumes are key image-based biomarkers to predict the severity of kidney and liver phenotype in autosomal dominant polycystic kidney disease (ADPKD). However, magnetic resonance (MR) imaging-based advanced biomarkers like total cyst number and cyst parenchyma surface area have been shown to more accurately assess cyst burden and improve the prediction of disease progression. The main aim of this study was to extend the calculation of advanced biomarkers to other imaging modalities; thus, we propose a fully automated model to segment kidney and liver cysts in computed tomography (CT) images. Methods Abdominal CTs of patients with ADPKD were gathered retrospectively between 2001 and 2018. A three-dimensional deep learning method using the nnU-Net architecture was trained to learn cyst edges-cores and the noncystic kidney/liver parenchyma. Separate segmentation models were trained for kidney cysts in contrast-enhanced CTs and liver cysts in noncontrast CTs using an active learning approach. Two experienced research fellows manually generated the reference standard segmentation, which were reviewed by an expert radiologist for accuracy. Results Two-hundred CT scans from 148 patients (mean age, 51.2±14.1 years; 48% male) were used for model training (80%) and testing (20%). In the test set, both models showed good agreement with the reference standard segmentations, similar to the agreement between two independent human readers (model versus reader: TCN kidney/liver r =0.96/0.97 and CPSA kidney r =0.98, inter-reader: TCN kidney/liver r =0.96/0.98 and CPSA kidney r =0.99). Conclusions Our study demonstrates that automated models can segment kidney and liver cysts accurately in CT scans of patients with ADPKD.

Abstract The advances in large language models (LLMs) have demonstrated sophisticated potential for automating complex tasks within the radiology workflow. From radiology report generation and report summarization to data collection for research trials, these models have proven to be powerful tools. However, optimal implementation of these models requires careful adaptation to the specialized medical domain. In addition, these models tend to generate information that is not truthful or factual, which can adversely affect patient care and clinical decisions. Strategies such as fine-tuning and prompt optimization have been shown to be impactful in eliminating these errors. Although these models undergo rapid updates and improvements, understanding the principles of prompt engineering and fine-tuning provides a foundation for evaluating and maintaining the performance of any LLM deployment. The current article aims to review the recent advancements in radiology using fine-tuning and prompt optimization to leverage LLMs’ capabilities. It delves into various techniques within each strategy, their advantages and limitations, and presents a framework to facilitate the practical integration of LLMs into radiology settings.

Abstract Background The CCDC26 germline variant rs55705857 is causal for development of IDH mutant (IDHmut) adult glioma. However, ~60% of IDHmut patients do not carry the rs55705857 risk allele. We aimed to identify variants associated with developing IDHmut glioma among patients that do not have the rs55705857 risk allele and to further understand development of IDHwt glioma. Methods We used three datasets that included 1216 IDHmut and 1442 IDHwt glioma patients and a case–case design to perform genome-wide association (GWAS) analyses comparing IDHmut versus IDHwt glioma. Analyses were performed overall and stratified by rs55705857 genotype and sex. Multivariable logistic regression and regression trees were used to develop models to predict IDH status using germline variants, age, and contrast enhancement on MRI. Results Three regions were identified comparing IDHmut versus IDHwt: rs55705857 (meta P-value [P] = 1.35 × 10−43), PHLDB1 (rs7125115, P = 3.46 × 10−17), and D2HGDH (rs71430382, P = 2.43 × 10−12). When analyzing only patients that do not have the rs55705857 risk allele, PHLDB1 (rs7125115, P = 1.73 × 10−13) and D2HGDH (rs71430382, P = 8.86 × 10−10) were identified. Among patients who have the rs55705857 risk allele, four variants between ROBO1 and ROBO2 (rs4680975, P = 4.65 × 10−7) increased the likelihood of having an IDHwt tumor. Tumor expression of ROBO1 was associated with rs4680975 genotype in IDHwt patients that have the rs55705857 risk allele (P = 0.034). Multivariable logistic analysis demonstrated that rs55705857, rs71430382 (D2HGDH), and age predicted IDH mutation status. Conclusions To understand the development of adult glioma, we demonstrate that D2HGDH and PHLDB1 should be prioritized for functional studies in IDHmut tumors. The ROBO1 region warrants further investigation in IDHwt tumors.

Recent advancements in compact MRI scanners, portable low-field-strength MRI systems, and high-performance gradient inserts address the challenges of conventional MRI for brain imaging, providing more accessible and versatile technology.

Purpose Spine-related pain disorders commonly require interventions such as spinal injections, which are often image-guided and increase in demand as populations age. This study investigates the impact of lifestyle changes following retirement on the incidence of these procedures, hypothesizing that retirement significantly influences the requirement for such treatments. Methods A review of medical records from XXX County between 2000 and 2023 was conducted, focusing on patients who received spinal injections, including epidural steroid, facet, and sacroiliac joint injections. Data were stratified by age, employment status pre- and post-procedure, and type of injection. Statistical analysis, including segmental linear regression and t-tests, was used to assess changes in incidence rates over time and across demographic subsets. Results Out of 134,318 patients analyzed, a total of 359,224 spinal injections were performed. There was a significant ( p -value <.01) increase in the frequency of procedures among individuals around the age of 60, which coincides with the typical retirement age in the United States. The study also highlighted a subsequent decline in the incidence of these procedures between the ages of 70 and 74, suggesting potential adaptation to chronic conditions or the effectiveness of earlier treatments. Conclusion The findings of this study confirm the hypothesis that retirement significantly impacts the demand for spinal injections, as evidenced by the marked increase in procedures at typical retirement ages. This underscores the importance of integrating preventive care and personalized medical guidance into healthcare strategies for individuals nearing retirement to effectively manage spine-related pain disorders and reduce the necessity for invasive treatments.

INTRODUCTION: The risk of progression to esophageal adenocarcinoma in Barrett's esophagus (BE) increases with advancing degrees of dysplasia. There is a critical need to improve the diagnosis of BE dysplasia, given substantial interobserver variability and overcalls of dysplasia during manual community pathologist reads. We aimed to externally validate a previously cross-validated BE dysplasia diagnosis deep learning model (BEDDLM) that predicts dysplasia grade on whole slide images (WSIs). METHODS: We digitized nondysplastic BE (NDBE), low-grade (LGD), and high-grade dysplasia (HGD) histology slides from 3 external academic centers. A consensus read by 2 expert study pathologists was used as the criterion standard. Slide stain characteristics were normalized using cycle-generative adversarial networks. WSIs were assessed by BEDDLM using an ensemble approach, combining a “You Only Look Once” model followed by a ResNet101 classifier model. RESULTS: We included 489 WSIs. Consensus histopathology revealed 232 NDBE, 117 LGD, and 140 HGD WSIs. The mean age (SD) was 66.9 (11.4) years; 413 (84.7%) were men. Using the BEDDLM ensemble model, sensitivity and specificity for NDBE were 73.3% (95% confidence interval [CI]: 67.09%–78.85%) and 93.4% (95% CI: 89.62%–96.10%); for LGD, 84.6% (95% CI: 76.78%–90.62%) and 80.6% (95% CI: 76.26%–84.54%); and for HGD, 80.7% (95% CI: 73.19%–86.89%) and 94.8% (95% CI: 91.97%–96.91%), respectively. The F1 score was 0.81, 0 0.69, and 0.83 for NDBE, LGD, and HGD, respectively. DISCUSSION: Our externally validated deep learning model demonstrates substantial accuracy for the diagnosis of BE dysplasia grade on WSIs.