PurposeAlzheimer’s disease (AD) is the leading cause of dementia, characterized by the accumulation of amyloid-β and tau aggregates in the brain. Current diagnostic approaches primarily detect mature plaques, but growing evidence suggests that pre-fibrillar aggregates are present earlier in normal-appearing tissue and may provide a valuable window for intervention. We propose the first proof-of-concept end-to-end deep learning (DL) model to identify AD signatures directly from normal-appearing tissue (i.e., background parenchyma) of raw spectral fluorescence data.ApproachTo develop and evaluate our DL model, we used one million signals extracted from spectral fluorescence images of brain cortex sections collected from wild-type and 5xFAD mice (a transgenic strain carrying five familial human AD mutations).ResultsOur results demonstrate that AD-related signatures can indeed be detected in normal-appearing brain cortex regions. Experiments revealed that model performance depends on input dimensions, training set size, and the preprocessing kernel size. The optimal configuration used 1000×32 input, 9000 training samples, and a 3×3 kernel, achieving ≥95% accuracy, sensitivity, specificity, F1-score, and area under the receiver operating characteristic curve. Importantly, no mice were misclassified, and saliency analysis indicated that a broad range of spectral fluorescence wavelengths contributed to the model’s decisions.ConclusionsThese findings highlight the potential of DL-based analysis of spectral fluorescence data to enable earlier detection of AD-related changes, supporting a shift toward a more proactive strategy for biomarker discovery.
PurposeWe aim to present a consolidated overview of current achievements in generative artificial intelligence (AI) in breast imaging, identify prevalent technical and clinical challenges alongside the strategies employed to overcome them, and highlight critical directions for future investigation.ApproachA comprehensive literature search has been conducted in PubMed, MEDLINE, and Embase for peer-reviewed journal publications since 2020, with the final search performed in October 2025. The review focuses on common methodologies, performance metrics, and strategies used to address existing hurdles in breast imaging.ResultsThe 120 included papers were categorized by application, ranging from technical and/or clinical tasks—image synthesis (including 24 data augmentation, 10 enhancement, 20 image-to-image translation, and 2 secure data sharing papers), 12 segmentation, 5 reconstruction, and 16 classification papers—as well as 17 papers in radiology workflow improvement, 13 papers in the development of medical education materials, and 7 papers on solving beyond our defined tasks (six papers were assigned more than one application). Papers in each category showed their own potential to solve the task of interests. However, we identified six common challenges and concerns, such as lack of standardized evaluation for synthetic outputs, across papers that the future research should be focused on.ConclusionsRecent works across the field of generative AI in breast imaging have demonstrated transformative potential in clinical workflows. However, there still exists technical and ethical gaps to be addressed for safe, reliable, and equitable deployment of the generative AI models to routine medical practice.
Purpose The transition from task-specific artificial intelligence toward general-purpose foundation models raises fundamental questions about their capacity to support the integrated reasoning required in clinical medicine, where diagnosis demands synthesis of ambiguous patient narratives, laboratory data, and multimodal imaging. Approach We provide the first controlled, cross-sectional evaluation of the GPT-5 family (GPT-5, GPT-5 Mini, and GPT-5nano) against its predecessor GPT-4o across a diverse spectrum of clinically grounded tasks, grounded tasks, including the United States Medical Licensing Examination, text-based, text-based reasoning benchmarks (MedQA and MedXpertQA Text), and visual question answering (VQA) in neuroradiology (BraTS), digital pathology (PathVQA and BreakHis), and mammography (EMBED, InBreast, Chinese Mammography Database, and CBIS-DDSM) using a standardized zero-shot chain-of-thought protocol. Results GPT-5 demonstrated substantial performance gains with absolute improvements exceeding 25 percentage points on MedXpertQA. When tasked with multimodal synthesis, GPT-5 effectively leveraged its enhanced training corpora to achieve state-of-the-art or competitive performance across most VQA benchmarks and outperformed GPT-4o by margins of 10% to 40% in mammography tasks requiring fine-grained lesion characterization. However, performance remained moderate in neuroradiology (44% macro-average accuracy) and lagged substantially behind domain-specific models in mammography, where specialized systems exceed 80% accuracy compared with GPT-5's 52% to 64%. Conclusions These findings indicate that although GPT-5 represents a meaningful advance toward general, multimodal clinical inference, generalist models are still not yet substitutes for purpose-built systems in highly specialized, perception-critical tasks. We delineate the current VQA landscape, positioning GPT-5 as a powerful adjunct capable of holistic inference while underscoring that domain adaptation, rigorous validation, and guarantees of reasoning transparency remain essential prerequisites for clinical deployment.
PurposeInterstitial fibrosis is a critical histopathologic feature used in the diagnosis and prognosis of kidney diseases. However, conventional assessment using trichrome-stained slides is limited by interobserver variability and inconsistent stain reproducibility. This study evaluates whether dual-mode emission and transmission (DUET) microscopy, which generates virtual trichrome images from hematoxylin and eosin (H&E) slides, can provide a practical, reproducible, and quantitative alternative for assessing interstitial fibrosis in kidney transplant biopsies.ApproachThirty-two kidney transplant biopsies were imaged using DUET to acquire pixel-registered brightfield and fluorescence whole-slide images from H&E-stained slides. A deep learning pipeline was trained on DUET-acquired images to extract collagen-positive signal from both the brightfield and fluorescence channels, which was used to generate virtual trichrome images and to compute quantitative fibrosis scores. Virtual trichrome images were compared with corresponding conventional trichrome slides. Four renal pathologists independently estimated interstitial fibrosis on both modalities to assess interobserver agreement, and computational fibrosis scores were correlated with pathologists’ estimates.ResultsPathologist assessments of fibrosis on DUET-derived virtual trichrome images demonstrated interobserver agreement comparable to traditional trichrome slides, trending higher for virtual trichrome (ICC 0.536 versus 0.477, p=0.376). Agreement of each pathologist with the other three readers was higher for virtual trichrome in three of the four pathologists and significantly higher in one. Fibrosis estimates were systematically lower on virtual trichrome than on real trichrome, by a mean of 7.6 percentage points. The deep learning model for collagen extraction achieved high segmentation performance (AUC 0.994, Dice 0.903). Computational fibrosis quantitation showed moderate correlation with pathologist estimates and demonstrated stronger association with virtual trichrome than real trichrome (r 0.67 versus 0.54, R2 0.44 versus 0.29).ConclusionsDUET microscopy enables the generation of pixel-registered virtual trichrome images from standard H&E slides with interobserver agreement comparable to conventional trichrome. By extracting collagen signal from pixel-registered fluorescence data on the same tissue section, DUET supports reproducible visual assessment and quantitative fibrosis measurement. This approach provides a scalable alternative to conventional trichrome staining and a foundation for integrated digital pathology workflows.
PurposeThe Bayesian ideal observer (IO) defines an upper bound on task-based observer performance and is widely used as a theoretical benchmark for optimizing medical imaging systems and data-acquisition designs. However, except in special cases, the exact IO test statistic is analytically intractable. Recent deep neural network (DNN)-based approaches show promise for approximating IO performance while providing limited analytical insight and no guarantee that the learned test statistic corresponds to a likelihood ratio. In addition, their empirical performance can degrade in data-limited regimes, motivating methods that incorporate stronger statistical structure and regularization. To address this, we propose a data-driven framework based on the variational information bottleneck (VIB) to approximate the IO test statistic for binary signal-detection tasks.ApproachThe proposed method utilizes the variational bottleneck principle to encode high-dimensional image data into a low-dimensional latent space optimized to preserve task-relevant information while suppressing nuisance variability. An analytically motivated Gaussian IO loss encourages the class-conditional latent distributions to follow multivariate Gaussian models. Under the adopted Gaussian model for the class-conditional latent distributions, a closed-form Gaussian likelihood-ratio statistic can be evaluated. Unlike end-to-end convolutional neural network (CNN)-based IO approximation methods, this formulation yields an explicit likelihood-ratio test statistic and provides greater analytical interpretability.ResultsNumerical studies demonstrate that the proposed VIB-based IO method achieves detection performance comparable to conventional CNN approaches on large datasets, and superior performance in data-limited regimes. Furthermore, the learned bottleneck representations exhibit stability under the domain shifts examined, supporting robust deployment across varying imaging conditions.ConclusionsThe proposed VIB-IO framework provides a theoretically grounded and relatively data-efficient alternative to conventional CNN-based IO approximations while retaining greater analytical transparency through an explicit latent-space Gaussian observer model. By combining the representational capacity of DNNs with an analytically tractable latent-space observer model, the proposed framework provides a principled approach for objective, task-based assessment of medical imaging systems.
