Background: Reducing radiation dose in Computed Tomography (CT) is a clinical imperative, yet current low-dose CT (LDCT) protocols are inherently limited by noise and artifacts that obscure critical diagnostic details. While traditional iterative reconstruction and standard deep learning models have been proposed, they often introduce undesirable "waxy" textures and loss of fine structural edges. Methods: This study presents a high-fidelity image enhancement framework based on Generative Adversarial Networks (GANs) specifically optimized for LDCT denoising. Our approach integrates a multi-scale U-Net generator with a discriminator that evaluates both global structure and local texture. To ensure clinical reliability, we utilize a composite loss function comprising pixel-wise MSE, adversarial loss, and VGG-based perceptual loss, which encourages the model to preserve biologically relevant high-frequency details. Results: Experimental evaluation on the AAPM-Mayo Clinic dataset demonstrates that our proposed model significantly outperforms state-of-the-art methods, achieving a Peak Signal-to-Noise Ratio (PSNR) of 32.56 dB and a Structural Similarity Index (SSIM) of 0.912. Qualitative assessments confirm that our framework effectively suppresses quantum noise while maintaining the natural stochastic texture of CT images, facilitating better visualization of subtle lesions compared to standard CNN-based denoising. Conclusion: The proposed GAN-based framework bridges the gap between dose reduction and image quality, providing a viable pathway for high-quality diagnostic imaging at a fraction of the standard radiation exposure. This method holds significant potential for routine clinical use, particularly in screening scenarios where repeated exposure is a concern.

.PurposeThe Medical Imaging and Data Resource Center Mastermind Grand Challenge of modified radiographic assessment of lung edema (mRALE) tasked participants with developing machine learning techniques for automated COVID-19 severity assessment via mRALE scores on portable chest radiographs (CXRs). We examine potential biases across demographic subgroups for the best-performing models of the nine teams participating in the test phase of the challenge.ApproachModels were evaluated against a nonpublic test set of CXRs (814 patients) annotated by radiologists for disease severity (mRALE score 0 to 24). Participants used a variety of data and methods for training. Performance was measured using quadratic-weighted kappa (QWK). Bias analyses considered demographics (sex, age, race, ethnicity, and their intersections) using QWK. In addition, for distinguishing no/mild versus moderate/severe disease, equal opportunity difference (EOD) and average absolute odds difference (AAOD) were calculated. Bias was defined as statistically significant QWK subgroup differences, or EOD outside [; 0.1], or AAOD outside [0; 0.1].ResultsThe nine models demonstrated good agreement with the reference standard (QWK 0.74 to 0.88). The winning model (QWK = 0.884 [0.819; 0.949]) was the only model without biases identified in terms of QWK. The runner-up model (QWK = 0.874 [0.813; 0.936]) showed no identified biases in terms of EOD and AAOD, whereas the winning model disadvantaged three subgroups in each of these metrics. The median number of disadvantaged subgroups for all models was 3.ConclusionsThe challenge demonstrated strong model performances but identified subgroup disparities. Bias analysis is essential as models with similar accuracy may exhibit varying fairness.

Artificial Intelligence (AI) is transforming the domain of diagnostic radiology, evolving beyond a supportive technology into a full-fledged Radiology Extender (RE). Rather than functioning merely as an auxiliary, contemporary AI systems now perform a spectrum of high-stakes radiological tasks, including lesion detection, quantitative imaging analysis, clinical triage, and semi-automated report generation. These capabilities— rooted in convolutional neural networks (CNNs), natural language processing (NLP), and real-time inference Models are fundamentally shifting how radiology is practiced. This paper presents a rigorous and systematic review of AI's operational integration in radiology, critically evaluating its technical performance, clinical roles, and future trajectory. It introduces a conceptual framework structured around three AI tiers: pre-screening, decision support, and workflow optimization. The work also confronts systemic challenges such as interoperability, algorithmic opacity, and medico-legal ambiguities. Through a detailed clinical vignette involving the AI-assisted diagnosis of an inferior hip dislocation, the paper illustrates the real-world clinical efficacy and transformative value of AI. By advancing a vision of collaborative intelligence, where AI augments rather than replaces human expertise; we Propose a sustainable, ethical, and scalable roadmap for next-generation diagnostic imaging systems. This paper aims to guide both clinical practitioners and technology developers toward effective, explainable, and responsible AI integration within radiological ecosystems.

.Thanks to reviewers who served the Journal of Medical Imaging in 2024.

.PurposeEye morphology varies significantly across the population, especially for the orbit and optic nerve. These variations limit the feasibility and robustness of generalizing population-wise features of eye organs to an unbiased spatial reference.ApproachTo tackle these limitations, we propose a process for creating high-resolution unbiased eye atlases. First, to restore spatial details from scans with a low through-plane resolution compared with a high in-plane resolution, we apply a deep learning-based super-resolution algorithm. Then, we generate an initial unbiased reference with an iterative metric-based registration using a small portion of subject scans. We register the remaining scans to this template and refine the template using an unsupervised deep probabilistic approach that generates a more expansive deformation field to enhance the organ boundary alignment. We demonstrate this framework using magnetic resonance images across four different tissue contrasts, generating four atlases in separate spatial alignments.ResultsWhen refining the template with sufficient subjects, we find a significant improvement using the Wilcoxon signed-rank test in the average Dice score across four labeled regions compared with a standard registration framework consisting of rigid, affine, and deformable transformations. These results highlight the effective alignment of eye organs and boundaries using our proposed process.ConclusionsBy combining super-resolution preprocessing and deep probabilistic models, we address the challenge of generating an eye atlas to serve as a standardized reference across a largely variable population.

.PurposeOur purpose is to develop a computer vision approach to quantify intra-arterial thickness on digital pathology images of kidney biopsies as a computational biomarker of arteriosclerosis.ApproachThe severity of the arteriosclerosis was scored (0 to 3) in 753 arteries from 33 trichrome-stained whole slide images (WSIs) of kidney biopsies, and the outer contours of the media, intima, and lumen were manually delineated by a renal pathologist. We then developed a multi-class deep learning (DL) framework for segmenting the different intra-arterial compartments (training dataset: 648 arteries from 24 WSIs; testing dataset: 105 arteries from 9 WSIs). Subsequently, we employed radial sampling and made measurements of media and intima thickness as a function of spatially encoded polar coordinates throughout the artery. Pathomic features were extracted from the measurements to collectively describe the arterial wall characteristics. The technique was first validated through numerical analysis of simulated arteries, with systematic deformations applied to study their effect on arterial thickness measurements. We then compared these computationally derived measurements with the pathologists’ grading of arteriosclerosis.ResultsNumerical validation shows that our measurement technique adeptly captured the decreasing smoothness in the intima and media thickness as the deformation increases in the simulated arteries. Intra-arterial DL segmentations of media, intima, and lumen achieved Dice scores of 0.84, 0.78, and 0.86, respectively. Several significant associations were identified between arteriosclerosis grade and pathomic features using our technique (e.g., intima-media ratio average [, ]) through Kendall’s tau analysis.ConclusionsWe developed a computer vision approach to computationally characterize intra-arterial morphology on digital pathology images and demonstrate its feasibility as a potential computational biomarker of arteriosclerosis.

.The editorial evaluates how the GenAI technologies available in 2024 (without specific coding) could impact scientific processes, exploring two AI tools with the aim of demonstrating what happens when using custom LLMs in five research lab workflows.

.PurposeVirtual reality (VR) technology has emerged as a promising tool for physicians, offering the ability to assess anatomical data in 3D with visuospatial interaction qualities. The last decade has witnessed a remarkable increase in the number of studies focusing on the application of VR to assess patient-specific image data. This systematic review aims to provide an up-to-date overview of the latest research on VR in the field of surgical planning.ApproachA comprehensive literature search was conducted based on the preferred reporting items for systematic reviews and meta-analyses covering the period from April 1, 2021 to May 10, 2023. It includes research articles reporting on preoperative surgical planning using patient-specific medical images in virtual reality using head-mounted displays. The review summarizes the current state of research in this field, identifying key findings, technologies, study designs, methods, and potential directions for future research.ResultsThe selected studies show a positive impact on surgical decision-making and anatomy understanding compared to other visualization modalities. A substantial number of studies are reporting anecdotal evidence and case-specific outcomes. Notably, surgical planning using VR led to more frequent changes in surgical plans compared to planning with other visualization methods when surgeons reassessed their initial plans. VR demonstrated benefits in reducing planning time and improving spatial localization of pathologies.ConclusionsResults show that the application of VR for surgical planning is still in an experimental stage but is gradually advancing toward clinical use. The diverse study designs, methodologies, and varying reporting hinder a comprehensive analysis. Some findings lack statistical evidence and rely on subjective assumptions. To strengthen evaluation, future research should focus on refining study designs, improving technical reporting, defining visual and technical proficiency requirements, and enhancing VR software usability and design. Addressing these areas could pave the way for an effective implementation of VR in clinical settings.

Precision medicine is a revolutionary medical approach that takes into account individual variations in genes, environment, and lifestyle to tailor treatment and prevention strategies for each patient. One crucial aspect of precision medicine is medical imaging, which plays a fundamental role in the accurate assessment and diagnosis of diseases. By utilizing various imaging modalities, such as MRI, CT scan, and PET scan, doctors can obtain detailed information about a patient's condition and make informed decisions about the most effective therapeutic options. Furthermore, imaging techniques like radiomics have emerged as a promising tool in precision medicine. By extracting quantitative data from medical images, radiomics allows for objective and quantitative analysis of the biological characteristics of diseases. This information can then be correlated and integrated with genomic data to further enhance our understanding of diseases, a concept known as radiogenomics. Imaging plays a vital role in the development of precision medicine. Radiomics, based on high-throughput medical imaging, allows for the extraction of vast amounts of data from medical images to perform objective and quantitative analysis of the biological characteristics of diseases. This is crucial for tumor diagnosis, differential diagnosis, prognosis evaluation, and prediction of treatment response. Medical imaging, specifically radiomics, has greatly advanced the field of precision medicine by providing a comprehensive analysis of tumors and extracting quantitative data from medical images. This data can be correlated and integrated with genomic information to enhance our understanding of diseases and enable personalized treatment plans. In summary, medical imaging, particularly radiomics, plays a central and vital role in the development of precision medicine.

.PurposeQuantification of perfusion in ml/100 g/min, rather than comparing relative values side-to-side, is critical at the clinical and research levels for large longitudinal and multi-center trials. Intravoxel incoherent motion (IVIM) is a non-contrast magnetic resonance imaging diffusion-based scan that uses a multitude of -values to measure various speeds of molecular perfusion and diffusion, sidestepping inaccuracy of arterial input functions or bolus kinetics. Questions remain as to the original of the signal and whether IVIM returns quantitative and accurate perfusion in a pathology setting. This study tests a novel method of IVIM perfusion quantification compared with neutron capture microspheres.ApproachWe derive an expression for the quantification of capillary blood flow in ml/100 g/min by solving the three-dimensional Gaussian probability distribution and defining water transport time (WTT) as when 50% of the original water remains in the tissue of interest. Calculations were verified in a six-subject pre-clinical canine model of normocapnia, induced hypercapnia, and middle cerebral artery occlusion (ischemic stroke) and compared with quantitative microsphere perfusion.ResultsLinear regression analysis of IVIM and microsphere perfusion showed agreement (slope = 0.55, intercept = 52.5, ) with a Bland–Altman mean difference of . Linear regression between dynamic susceptibility contrast mean transit time and IVIM WTT asymmetry in infarcted tissue was excellent (, intercept = 0.3, ). Strong linear agreement was found between IVIM and reference standard infarct volume (slope = 1.01, ). The simulation of cerebrospinal fluid (CSF) suppression via inversion recovery returned a blood signal reduced by 82% from combined T1 and T2 effects.ConclusionsThe accuracy and sensitivity of IVIM provides evidence that observed signal changes reflect cytotoxic edema and tissue perfusion and can be quantified with WTT. Partial volume contamination of CSF may be better removed during post-processing rather than with inversion recovery.