Physicians operate at the intersection of 2 conflicting imperatives: the clinical mandate to avoid missed diagnoses and the ethical requirement to avoid unnecessary interventions. While advanced imaging can reduce diagnostic uncertainty, overuse introduces systemic inefficiencies, financial waste, and physical harm. This article argues that the solution lies in transitioning from an information maximization mindset to a satisficing framework, rooted in the theory of bounded rationality. Beyond biological risks, additional imaging often identifies insignificant incidental findings, triggering diagnostic cascades and psychological distress. Over-ordering may be motivated by defensive medicine, patient satisfaction pressures, and financial conflicts of interest. A satisficing framework clarifies when additional information is no longer needed. When satisficing, one stops acquiring information once current information is sufficient for action, whereas under a Value of Information (VOI) framework, one stops when the expected incremental benefit of additional information no longer exceeds its incremental costs and harms. Both frameworks reflect that the relationship between clinical utility and imaging data volume is nonlinear; eventually, incremental contributions diminish while cumulative costs continue to rise. Putting this approach into practice requires leveraging Clinical Decision Support Systems (CDSS), minimalist protocols, and increased visibility regarding opportunity costs. Robust safety protocols, including departmental reviews of exception rates and peer reviews, can be used to ensure that satisficing does not lead to increased diagnostic errors. Ultimately, quality in radiology should be defined by whether imaging appropriately informs management, rather than by the volume of data gathered. The goal is to provide the information necessary to act safely and effectively while recognizing when to stop.
Primary chronic pain syndromes are increasingly understood through advanced neuroimaging techniques, which reveal consistent structural, functional, and molecular alterations in the central nervous system. Multi-modal Magnetic Resonance Imaging-including diffusion, structural, and functional approaches-demonstrates reduced gray matter, disrupted neural network connectivity, and altered activation in key pain-processing regions such as the insula, thalamus, and anterior cingulate cortex, with distinct patterns for pain anticipation and stimulus processing. Positron emission tomography/CT imaging further elucidates neurobiological mechanisms, identifying changes in glucose metabolism, neurotransmitter systems, and neuroinflammation, particularly through elevated Translocator Protein (marker of microglial activation) signals and altered opioid and dopaminergic pathways in chronic pain populations. Recent studies highlight the potential of imaging biomarkers for diagnosis, patient stratification, and prediction of treatment response, with machine learning and multivariate pattern analysis improving specificity and classification accuracy. Integrating imaging, molecular, and psychosocial data enables the creation of composite signatures for personalized pain management. Despite these advances, challenges remain in standardizing imaging, validating biomarkers, and implementing findings into routine clinical practice. Ongoing research for imaging pain syndromes focuses on harmonization efforts, large-scale multicenter collaborations, and the integration of artificial intelligence to optimize biomarker utilization and strengthen clinical decision-support systems. This review explores how advanced Magnetic Resonance Imaging and Positron emission tomography/CT techniques have transformed the understanding of primary chronic pain syndromes, facilitating precision diagnosis and targeted therapeutic strategies.
Emergency radiology operates in a high-acuity, time-sensitive environment where imaging is tightly integrated into real-time clinical decision-making. Growing imaging demand, increasing case complexity, and workforce constraints have intensified pressure on emergency radiologists. Artificial intelligence (AI) has emerged as a potential tool to support imaging prioritization, interpretation, and operational efficiency. However, to meaningfully advance care delivery, the role of AI must be considered beyond algorithm performance, including its implementation, reliability, and real-world clinical impact. In this narrative review, we examine the role of AI across the emergency radiology workflow through three lenses: current capabilities, limitations of the supporting evidence, and practical considerations for clinical implementation. We review applications spanning pre-image acquisition, image acquisition and reconstruction, computer-aided triage and detection, reporting, and follow-up, integrating published evidence with practical insights. Discrepancies between reported and real-world performance, the influence of human-AI interaction on clinical decision-making, and the potential for subtle errors and bias are also discussed. As national regulatory and local governance frameworks continue to evolve, including emerging challenges posed by large language models, gaps remain between reported and real-world AI performance. In emergency radiology, the true impact of AI will depend on how seamlessly and effectively these tools are integrated into existing clinical workflows. Local validation, ongoing performance monitoring, and multidisciplinary institutional oversight are essential to identify performance variability, mitigate biases, and support reliable use in a high-stakes clinical environment.
Neurodegenerative disorders have traditionally been classified according to clinical syndromes or patterns of anatomical involvement on neuroimaging. However, growing evidence demonstrates that similar clinical phenotypes may arise from distinct molecular pathologies, while a single pathogenic protein may manifest with diverse clinical and imaging presentations. This has led to the emergence of the proteinopathy paradigm, which conceptualizes neurodegeneration as a disorder of protein misfolding, aggregation, and consequent pathologic changes. This review provides an imaging-focused overview of the major central nervous system proteinopathies, including prion diseases, amyloid-β-related disorders, tauopathies, synucleinopathies, and TAR DNA-binding protein 43-associated diseases. We discuss the presence of distinct and often predictable radiological phenotypes in these conditions, which can help in diagnosis, predict clinical progression, and explain clinical phenotype. Conventional magnetic resonance imaging remains central to structural pattern recognition, while advanced techniques such as diffusion-weighted imaging, susceptibility-weighted imaging, perfusion imaging, and quantitative volumetry may enhance diagnostic confidence. Molecular imaging with fluorodeoxyglucose positron emission tomography and emerging amyloid and tau tracers further enables in vivo characterization of disease-specific metabolic and molecular signatures. By integrating molecular mechanisms with imaging findings, this review highlights the role of neuroimaging as a bridge between microscopic protein pathology and macroscopic disease expression. Understanding proteinopathy-specific imaging patterns allows a shift from symptom-led diagnosis toward a biology-driven framework, improving diagnostic accuracy, prognostication, and the potential for targeted therapeutic monitoring in neurodegenerative disease.
Interstitial lung disease (ILD) encompasses a diverse range of conditions that lead to inflammation and/or scarring of the lung interstitium, often affecting airspaces and resulting in a progressive decline in lung function. High-Resolution Computed Tomography (HRCT) is a crucial diagnostic tool for ILDs, and their characterization based on imaging. This article specifically focuses on ILD presentations characterized by small lung nodules on HRCT, defined as those measuring less than 10 mm. Small nodules on HRCT are analyzed based on size, distribution, borders, attenuation, associated findings, and temporal evolution to narrow diagnostic considerations. A key factor is the distribution pattern, which is placement within the secondary pulmonary lobule and axial interstitium. Based on their distribution pattern relative to the anatomical core of the lobule, small nodules are classified into 3 specific imaging phenotypes: perilymphatic, centrilobular, and random. Perilymphatic nodular phenotypes typically involve disease processes affecting the pulmonary lymphatics along the interlobular septa, pleura, fissures, and/or bronchovascular bundles. Common conditions include Sarcoidosis, Occupational lung diseases such as coal worker pneumoconiosis (CWP) and silicosis, Chronic beryllium disease, Granulomatous and Lymphocytic interstitial lung disease (GL-ILD), Pulmonary septal amyloidosis, Pulmonary alveolar microlithiasis, Diffuse Pulmonary Ossification. Centrilobular nodular phenotypes are centered on the core structures of the secondary pulmonary lobule, including bronchioles, pulmonary arterioles, and central lymphatic vessels. They can be nonbranching (solid or ground-glass attenuation) or branching, often appearing as a "tree-in-bud" morphology. Nonbranching centrilobular nodules are seen in conditions such as Nonfibrotic hypersensitivity pneumonitis (HP), Respiratory bronchiolitis (RB), Follicular bronchiolitis (FB), Lymphocytic interstitial pneumonitis (LIP), Metastatic pulmonary calcification (MPC), Pulmonary hemosiderosis, and Pulmonary Langerhans cell histiocytosis (LCH). Branching centrilobular nodules ("tree-in-bud") are typically not associated with ILDs and often indicate Infectious bronchiolitis, Aspiration and other disorders. Random pulmonary nodular phenotypes refer to nodules without a consistent relationship to the secondary lobule or other lung structures. While profuse perilymphatic nodules (e.g., in sarcoidosis and occupational lung diseases) can appear randomly distributed, true random patterns are characteristic of hematogenous infections or miliary metastases.
Psychoradiology is an emerging interdisciplinary subspecialty bridging psychiatry, neuroradiology, and neuroscience to explore the neurobiological basis of mental illness. Using multimodal imaging techniques such as magnetic resonance imaging, functional magnetic resonance imaging, diffusion tensor imaging, magnetic resonance spectroscopy, positron emission tomography, and single photon emission computed tomography, psychoradiology enables the investigation of structural, functional, and neurochemical abnormalities underlying psychiatric disorders. Although neuroimaging has advanced over 4 decades, its clinical diagnostic yield in primary psychiatric syndromes remains limited, serving mainly to exclude secondary structural causes. Recent research demonstrates disease-specific changes in brain morphology, connectivity, and metabolism across schizophrenia, depression, and bipolar disorder, supporting their characterization as intrinsic brain disorders. The integration of imaging findings with clinical and computational models presents emerging opportunities for precision psychiatry. As psychoradiology transitions from research to clinical application, it holds promise for developing imaging biomarkers that inform disease subtyping, prognosis, and individualized treatment response.
The usual interstitial pneumonia (UIP) pattern remains central to the diagnosis and management of fibrotic interstitial lung diseases (ILDs), particularly idiopathic pulmonary fibrosis (IPF). Thin-section computed tomography enables noninvasive characterization across the UIP spectrum, with structured pattern classification demonstrating strong correlation with histopathology and survival outcomes. In appropriate clinical settings, thin-section CT can obviate the need for surgical lung biopsy. Multidisciplinary discussion remains essential to integrate imaging, clinical, and pathologic findings for optimal therapeutic decision-making, including timely antifibrotic initiation and transplant referral. This review synthesizes current knowledge on the UIP imaging spectrum, emphasizing its diagnostic, prognostic, and clinical implications. We further outline recent advances in quantitative imaging, deep learning models, and molecular biomarkers that improve diagnostic precision and risk stratification. We also highlight emerging frontiers, including AI-driven spatiotemporal modeling and molecular imaging, that promise earlier detection, individualized prognostication, and precision-guided therapy.
Large total joint arthroplasties (TJA), including total shoulder arthroplasty (TSA; anatomic [aTSA] and reverse [rTSA]), total knee arthroplasty (TKA), and total hip arthroplasty (THA), are among the most commonly performed and cost-effective orthopedic procedures in the United States. Utilization continues to rise due to an aging population with increasing functional demands, younger age at implantation, expanded indications, favorable clinical outcomes, and improved implant longevity. Despite excellent implant survivorship, complications remain clinically important. Imaging plays a central role in the evaluation of the painful arthroplasty, particularly in the emergency setting. Familiarity with normal postoperative appearances and characteristic imaging findings of complications is essential for accurate diagnosis and timely management. This review describes contemporary TSA, THA, and TKA designs and presents an imaging-based approach for evaluation of the painful arthroplasty in the emergency setting. The imaging findings of common complications, such as periprosthetic joint infection (PJI), aseptic loosening, osteolysis, instability, periprosthetic fracture (PPF) and component failure, are reviewed.
Cerebrospinal fluid (CSF) leaks, whether spontaneous or iatrogenic, can lead to debilitating post-dural puncture or intracranial hypotension-related headaches characterized by orthostatic symptoms. Epidural blood patching (EBP) has become a mainstay treatment once conservative measures (bed rest, hydration, and caffeine) fail. An EBP involves injecting autologous blood into the epidural space to seal dural defects, with success rates ranging from ∼70%-90% for post-dural puncture (iatrogenic) leaks but only ∼30% for spontaneous leaks on the first attempt. For patients with persistent CSF leakage despite repeat EBPs (refractory cases), fibrin sealant ("fibrin glue") injections provide an alternative minimally invasive therapy. Fibrin sealants polymerize into a clot that can patch the dural tear and withstand normal CSF pressures. Several case series report that targeted fibrin glue therapy (alone or combined with blood) yields additional successes, especially in patients who have failed standard blood patches. However, outcomes vary widely, with success rates for fibrin patches in the literature ranging from as low as ∼12%-30% in complex spontaneous leaks to as high as 80%-90% in some cohorts. This review provides a comprehensive overview of CSF leak etiologies and management, focusing on the techniques, efficacy, and indications of EBPs and fibrin glue sealant patches, as well as current evidence and evolving strategies to optimize treatment of these challenging cases.
Pulmonary vein stenosis (PVS) is an uncommon but often progressive condition in children, associated with high morbidity and mortality despite advances in diagnosis and treatment. Etiologies include primary congenital disease and secondary causes such as postoperative total or partial anomalous pulmonary venous return repair, bronchopulmonary dysplasia, and anatomic compression. PVS can occur in isolation or in association with complex congenital heart disease and is frequently characterized by restenosis after intervention. Multimodality imaging is essential for diagnosis, treatment planning, and follow-up in pediatric PVS. Transthoracic echocardiography remains the first-line screening tool, although limited acoustic windows and small vessel caliber may reduce accuracy. Cross-sectional imaging with cardiac magnetic resonance and computed tomography angiography provides high-resolution anatomical assessment and is increasingly complemented by dual-energy computed tomography with iodine perfusion mapping to evaluate the functional impact of stenosis. Cardiac catheterization remains the reference standard for hemodynamic assessment and offers therapeutic capabilities. This review summarizes the epidemiology, anatomy, pathophysiology, and imaging features of PVS in children, with emphasis on the role of multimodality imaging in both congenital and acquired forms. We discuss current interventional, surgical, and medical treatment strategies, highlight challenges in pediatric imaging, and outline recommendations for long-term surveillance to detect restenosis and guide timely reintervention.
Despite growing clinical interest in targeted PET tracers, FDG remains a stalwart in PET imaging due to its widespread availability and applicability. Recent data support expanded clinical indications for FDG-PET in the evaluation of acutely ill patients. This review aims to summarize recent developments in PET imaging among acutely ill patients, emphasizing expanding benign indications. This review highlights updated guidelines and recent evidence regarding FDG-PET in the evaluation of infection, fever of unknown origin, vascular inflammation, and rheumatologic disorders. We underscore the results of recent studies published in the last 5 years from publication, primarily between 2019 and 2024. FDG-PET remains a cornerstone of PET imaging, and its value in benign and inflammatory conditions continues to grow. While FDG-PET is essential for outpatient oncology, it is increasingly used to image hospitalized patients, given its ability to diagnose and localize infection and inflammation.
Cystic lung diseases represent a heterogeneous group of conditions characterized by the presence of multiple pulmonary cysts. Accurate recognition and differentiation of true cysts from their mimics (such as cavities, emphysema, bullae, blebs, and honeycombing) are essential, as management strategies and prognoses vary significantly. This review provides a systematic approach to the radiologic evaluation of cystic lung diseases, beginning with the definition and pathogenesis of pulmonary cysts and progressing through a structured diagnostic algorithm. We detail the characteristic imaging findings and clinical associations for major cystic lung diseases, including lymphangioleiomyomatosis (LAM), Birt-Hogg-Dubé syndrome (BHD), light-chain deposition disease (LCDD), lymphocytic interstitial pneumonia (LIP), desquamative interstitial pneumonia (DIP), pulmonary Langerhans cell histiocytosis (PLCH), and amyloidosis. Additionally, we discuss uncommon etiologies, such as cystic metastases, hypersensitivity pneumonitis, and cystic changes associated with genetic syndromes like neurofibromatosis type 1 (NF-1) and trisomy 21. The review emphasizes key imaging clues, such as cyst distribution, wall characteristics, associated nodules, and parenchymal abnormalities, that aid in narrowing the differential diagnosis. With the growing use of CT imaging, cystic lung diseases are increasingly identified in both symptomatic and asymptomatic patients. Familiarity with their imaging patterns, clinical contexts, and distinguishing features is essential for radiologists and clinicians alike. By following a stepwise, pattern-based approach, early and accurate diagnosis can be achieved, potentially improving patient outcomes through timely surveillance and targeted management.
Hydrocephalus is defined by abnormal accumulation of cerebrospinal fluid (CSF) within the ventricles, resulting in ventriculomegaly with variable effects on intracranial pressure. Historically classified as obstructive or communicating, contemporary frameworks further categorize hydrocephalus by chronicity, age of onset, and etiology, particularly distinguishing idiopathic from secondary causes in adults. Hydrocephalus most commonly arises from impaired CSF circulation or absorption, with less frequent contribution from altered CSF production. Increasing attention has been directed toward alternative CSF clearance mechanisms, including the glymphatic system, which remain incompletely defined. Radiologic evaluation is central to the diagnosis and management of hydrocephalus, enabling accurate assessment of ventricular morphology, associated parenchymal changes, and potential underlying etiologies. Normal pressure hydrocephalus, a chronic communicating hydrocephalus of older adults, is characterized by a clinical triad of gait disturbance, cognitive decline, and urinary dysfunction. Conventional structural and advanced imaging markers may assist in diagnosis, prognostication, and selection of patients for CSF diversion, in conjunction with clinical assessment. This review summarizes fundamental physiologic concepts of CSF dynamics and imaging features of hydrocephalus, with particular emphasis on imaging in normal pressure hydrocephalus.
Acquired pulmonary venous disease is predominantly detected on postablation imaging for atrial fibrillation, but may also be encountered incidentally or during evaluation for acute cardiopulmonary events. These entities require a high index of suspicion, as they involve subtle findings in an often-overlooked region. Moreover, recognition of associated findings may be particularly difficult on nongated imaging, which is prone to artifacts. Cardiac-gated CT and magnetic resonance imaging remain the primary imaging modalities, though echocardiography and diagnostic catheterization may contribute to the diagnosis in select cases. This review outlines key acquired pulmonary venous abnormalities, including pulmonary vein stenosis, pulmonary vein thrombosis, tumor invasion of the pulmonary vein, and intrapulmonary venous collateralization ("meandering pulmonary vein"). For each, we review relevant pathophysiology, imaging features, and clinical implications. The objective is to highlight key imaging features and potential challenges to avoid misdiagnosis and guide appropriate management across clinical scenarios.
Artificial intelligence (AI) enhances the practice of chest imaging by improving diagnostic accuracy, streamlining workflows, and facilitating personalized patient care. As a powerful tool, AI augments the expertise of radiologists, enabling more precise evaluations and quicker decision-making. This article examines the barriers to AI adoption in chest imaging, focusing on challenges related to bias, transparency, accountability, and data privacy. We discuss the ethical implications of AI-driven decision-making, particularly concerning fairness, and propose strategies to address these concerns. Additionally, we explore regulatory obstacles, including the approval pathways for AI algorithms and the need for continuous learning and adaptability in clinical settings. We also address practical considerations, such as the integration of AI tools into existing workflows, model generalizability, and economic factors. The article concludes with recommendations for responsible AI adoption, emphasizing the importance of interdisciplinary collaboration, robust validation frameworks, and continuous education for radiologists. By navigating these challenges, the radiology community can effectively leverage AI's potential, ultimately leading to enhanced patient outcomes and improved diagnostic processes.
Interstitial lung abnormalities (ILA) refer to incidental changes seen on chest CT, usually in people without a formal diagnosis of interstitial lung disease (ILD). Interest in ILA has grown in recent years, partly because they may signal early fibrotic lung changes and partly because CT scans are being used more often for screening. Defined by the Fleischner Society and updated by the ATS, ILA encompasses 3 patterns: nonsubpleural, subpleural nonfibrotic, and subpleural fibrotic. Subpleural fibrotic ILA, marked by traction bronchiectasis and honeycombing, carries the greatest likelihood of progression and the poorest prognosis. Older age, smoking history, and certain genetic traits such as the MUC5B promoter variant increase the likelihood of finding ILA. The presence of ILA also appears to raise the risk of lung cancer. For management, current guidelines recommend tailoring follow-up based on individual risk, with closer surveillance for patients more likely to progress. Newer tools, including quantitative imaging and artificial intelligence, may help detect subtle disease earlier and refine risk assessment. Despite advances, challenges remain in defining progression thresholds and treatment strategies, highlighting the need for further research.
Intra-abdominal free air is a critical radiologic finding that commonly indicates hollow viscus perforation. Prompt recognition of this finding on plain radiography and computed tomography (CT), 2 of the most commonly used modalities in the setting of suspected hollow viscus perforation, is thus crucial for timely surgical intervention. Diagnosing intra-abdominal free air can be challenging, as it may result from a wide range of both pathologic and benign etiologies. Common pathologic causes include perforation due to peptic ulcer disease (PUD) and diverticulitis and common benign causes include iatrogenic sources such as recent surgery and peritoneal dialysis. Intrinsic limitations of various imaging modalities as well as the condition of the patient also complicate the detection of intra-abdominal free air. As such, it is important to not only detect free air on imaging but to also contextualize it, which can help differentiate surgical from non-surgical causes and guide appropriate management. The goals of this review are to highlight the many causes of free air and the utility of various imaging modalities in its diagnosis, to provide clues to differentiate benign versus surgical etiologies of free air, to suggest a management framework, and to review emerging techniques and future directions in free air detection.
Pulmonary alveolar proteinosis (PAP) is a rare airspace disease classically associated with the crazy-paving pattern on high-resolution computed tomography (HRCT). While highly suggestive, this imaging pattern is not pathognomonic and appears across a wide spectrum of pulmonary pathologies. In this review, we adopt a phenotype-first approach, using representative imaging cases to walk the reader through the differential diagnosis of crazy-paving, with attention to radiologic distribution, clinical context, and disease acuity. We emphasize distinguishing features between PAP and its mimics-including pulmonary edema, diffuse alveolar hemorrhage, organizing pneumonia, mucinous adenocarcinoma, exogenous lipoid pneumonia, acute fulminant PAP, and COVID-19 pneumonia-using side-by-side imaging and contextual pearls. Special attention is given to the radiologic clues favoring autoimmune versus secondary PAP, including geographic distribution of ground-glass opacities, subpleural sparing, and lower lobe predominance. The review concludes with a summary of diagnostic strategies, pathologic correlation, and treatment options, including insights from post-pandemic diagnostic pitfalls. This pattern-based framework is designed for the radiologist and serves as a practical guide for recognizing PAP within the broader spectrum of airspace diseases.
White matter diseases encompass a heterogeneous spectrum of central nervous system disorders, with neuroimaging serving a pivotal role in their diagnosis and evaluation. Recent advances in imaging techniques and diagnostic frameworks have refined the evaluation of both common and rare entities. Updated criteria for multiple sclerosis, neuromyelitis optica spectrum disorder, and myelin oligodendrocyte glycoprotein antibody-associated disease increasingly incorporate advanced MRI biomarkers, such as the central vein sign and paramagnetic rim lesions, are improving diagnostic specificity and enabling earlier diagnosis. The improved recognition of adult-onset leukodystrophies and other clinically significant conditions, including progressive multifocal leukoencephalopathy, CADASIL, and Susac syndrome, has been driven by characteristic MRI patterns and quantitative imaging approaches. In parallel, artificial intelligence and machine learning techniques, including automated lesion segmentation and radiomics, are emerging as valuable tools for objective lesion quantification, disease classification, and prediction of disease activity.
U.S. healthcare spending has remained persistently high despite repeated efforts at correction. This essay offers a structural explanation. Waste, excess prices, and administrative complexity matter, but much spending growth reflects durable features of the sector that cannot be readily eliminated. Baumol's cost disease provides the core framework: in labor-intensive services with limited productivity gains, costs rise because wages in them must keep pace with more productive sectors. Medical technology more often expands capacity, utilization, and clinical expectations than it reduces labor inputs. The U.S. physician training pathway is unusually long and expensive, and federal residency caps have artificially constrained physician supply, reinforcing a high compensation floor. The healthcare and social assistance sector functions as a de facto industrial policy, as it is the nation's largest employment sector and the top employer in 38 states, making aggregate cost compression politically costly in ways that are structural, not incidental. Domestic multiplier effects deepen that political durability. Five distinctively American features further limit centralized cost control: population scale and decentralization, higher per capita income, a heavier chronic disease burden, the absence of a national health technology assessment authority, and weaker redistributive institutions. Given the constraints, the aspiration to make American healthcare dramatically cheaper without major disruption is unrealistic. A more credible agenda is to foster local stewardship within a structurally high-cost system.