Although there are relatively few diverse, high-quality medical imaging datasets on which to train computer vision artificial intelligence models, even fewer datasets contain expertly classified observations that can be repurposed to train or test such models. The traditional annotation process is laborious and time-consuming. Repurposing annotations and consolidating similar types of annotations from disparate sources has never been practical. Until recently, the use of natural language processing to convert a clinical radiology report into labels required custom training of a language model for each use case. Newer technologies such as large language models have made it possible to generate accurate and normalized labels at scale, using only clinical reports and specific prompt engineering. The combination of automatically generated labels extracted and normalized from reports in conjunction with foundational image models provides a means to create labels for model training. This article provides a short history and review of the annotation and labeling process of medical images, from the traditional manual methods to the newest semiautomated methods that provide a more scalable solution for creating useful models more efficiently. <b>Keywords:</b> Feature Detection, Diagnosis, Semi-supervised Learning © RSNA, 2025.

Accurate liver lesion segmentation in ultrasound is a challenging task due to high speckle noise, ambiguous lesion boundaries, and inhomogeneous intensity distribution inside the lesion regions. This work first collected and annotated a dataset for liver lesion segmentation in ultrasound. In this paper, we propose a novel convolutional neural network to learn dual self-attentive transformer features for boosting liver lesion segmentation by leveraging the complementary information among non-local features encoded at different layers of the transformer architecture. To do so, we devise a dual self-attention refinement (DSR) module to synergistically utilize self-attention and reverse self-attention mechanisms to extract complementary lesion characteristics between cascaded multi-layer feature maps, assisting the model to produce more accurate segmentation results. Moreover, we propose a False-Positive-Negative loss to enable our network to further suppress the non-liver-lesion noise at shallow transformer layers and enhance more target liver lesion details into CNN features at deep transformer layers. Experimental results show that our network outperforms state-of-the-art methods quantitatively and qualitatively.

Medical imaging is a vital diagnostic tool that provides detailed insights into human anatomy but faces challenges affecting its accuracy and efficiency. Advanced generative AI models offer promising solutions. Unlike previous reviews with a narrow focus, a comprehensive evaluation across techniques and modalities is necessary. This systematic review integrates the three state-of-the-art leading approaches, GANs, Diffusion Models, and Transformers, examining their applicability, methodologies, and clinical implications in improving medical image quality. Using the PRISMA framework, 63 studies from 989 were selected via Google Scholar and PubMed, focusing on GANs, Transformers, and Diffusion Models. Articles from ACM, IEEE Xplore, and Springer were analyzed. Generative AI techniques show promise in improving image resolution, reducing noise, and enhancing fidelity. GANs generate high-quality images, Transformers utilize global context, and Diffusion Models are effective in denoising and reconstruction. Challenges include high computational costs, limited dataset diversity, and issues with generalizability, with a focus on quantitative metrics over clinical applicability. This review highlights the transformative impact of GANs, Transformers, and Diffusion Models in advancing medical imaging. Future research must address computational and generalization challenges, emphasize open science, and validate these techniques in diverse clinical settings to unlock their full potential. These efforts could enhance diagnostic accuracy, lower costs, and improve patient outcome.

Brain lesion segmentation is challenging in medical image analysis, aiming to delineate lesion regions precisely. Deep learning (DL) techniques have recently demonstrated promising results across various computer vision tasks, including semantic segmentation, object detection, and image classification. This paper offers an overview of recent DL algorithms for brain tumor and stroke segmentation, drawing on literature from 2021 to 2024. It highlights the strengths, limitations, current research challenges, and unexplored areas in imaging-based brain lesion classification based on insights from over 250 recent review papers. Techniques addressing difficulties like class imbalance and multi-modalities are presented. Optimization methods for improving performance regarding computational and structural complexity and processing speed are discussed. These include lightweight neural networks, multilayer architectures, and computationally efficient, highly accurate network designs. The paper also reviews generic and latest frameworks of different brain lesion detection techniques and highlights publicly available benchmark datasets and their issues. Furthermore, open research areas, application prospects, and future directions for DL-based brain lesion classification are discussed. Future directions include integrating neural architecture search methods with domain knowledge, predicting patient survival levels, and learning to separate brain lesions using patient statistics. To ensure patient privacy, future research is anticipated to explore privacy-preserving learning frameworks. Overall, the presented suggestions serve as a guideline for researchers and system designers involved in brain lesion detection and stroke segmentation tasks.

In recent years, machine learning and deep learning have shown potential for improving Parkinson's disease (PD) diagnosis, one of the most common neurodegenerative diseases. This comprehensive analysis examines machine learning and deep learning-based Parkinson's disease diagnosis using MRI, speech, and handwriting datasets. To thoroughly analyze PD, this study collected data from scientific literature, experimental investigations, publicly accessible datasets, and global health reports. This study examines the worldwide historical setting of Parkinson's disease, focusing on its increasing prevalence and inequities in treatment access across various regions. A comprehensive summary consolidates essential findings from clinical investigations and pertinent datasets related to Parkinson's disease management. The worldwide context, prospective treatments, therapies, and drugs for Parkinson's disease have been thoroughly examined. This analysis identifies significant research deficiencies and suggests future methods, emphasizing the necessity for more extensive and diverse datasets and improved model accessibility. The current study proposes the Meta-Park model for diagnosing Parkinson's disease, achieving training, testing, and validation accuracy of 97.67 %, 95 %, and 94.04 %. This method provides a dependable and scalable way to improve clinical decision-making in managing Parkinson's disease. This research seeks to provide innovative, data-driven decisions for early diagnosis and effective treatment by merging the proposed method with a thorough examination of existing interventions, providing renewed hope to patients and the medical community.

Deformable registration is essential in medical image analysis, especially for handling various multi- and mono-modal registration tasks in neuroimaging. Existing studies lack exploration of brain MR-CT registration, and face challenges in both accuracy and efficiency improvements of learning-based methods. To enlarge the practice of multi-modal registration in brain, we present SR-Reg, a new benchmark dataset comprising 180 volumetric paired MR-CT images and annotated anatomical regions. Building on this foundation, we introduce MambaMorph, a novel deformable registration network based on an efficient state space model Mamba for global feature learning, with a fine-grained feature extractor for low-level embedding. Experimental results demonstrate that MambaMorph surpasses advanced ConvNet-based and Transformer-based networks across several multi- and mono-modal tasks, showcasing impressive enhancements of efficacy and efficiency. Code and dataset are available at https://github.com/mileswyn/MambaMorph.

Current virtual imaging phantoms primarily emphasize geometric&#xD;accuracy of anatomical structures. However, to enhance realism, it is also important&#xD;to incorporate intra-organ detail. Because biological tissues are heterogeneous in&#xD;composition, virtual phantoms should reflect this by including realistic intra-organ&#xD;texture and material variation.&#xD;We propose training two 3D Double U-Net conditional generative adversarial&#xD;networks (3D DUC-GAN) to generate sixteen unique textures that encompass organs&#xD;found within the torso. The model was trained on 378 CT image-segmentation&#xD;pairs taken from a publicly available dataset with 18 additional pairs reserved for&#xD;testing. Textured phantoms were generated and imaged using DukeSim, a virtual CT&#xD;simulation platform.&#xD;Results showed that the deep learning model was able to synthesize realistic&#xD;heterogeneous phantoms from a set of homogeneous phantoms. These phantoms were&#xD;compared with original CT scans and had a mean absolute difference of 46.15 ± 1.06&#xD;HU. The structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR)&#xD;were 0.86 ± 0.004 and 28.62 ± 0.14, respectively. The maximum mean discrepancy&#xD;between the generated and actual distribution was 0.0016. These metrics marked&#xD;an improvement of 27%, 5.9%, 6.2%, and 28% respectively, compared to current&#xD;homogeneous texture methods. The generated phantoms that underwent a virtual&#xD;CT scan had a closer visual resemblance to the true CT scan compared to the previous&#xD;method.&#xD;The resulting heterogeneous phantoms offer a significant step toward more realistic&#xD;in silico trials, enabling enhanced simulation of imaging procedures with greater fidelity&#xD;to true anatomical variation.

Ultrasound (US) is widely accessible and radiation-free but has a steep learning curve due to its dynamic nature and non-standard imaging planes. Additionally, the constant need to shift focus between the US screen and the patient poses a challenge. To address these issues, we integrate deep learning (DL)-based semantic segmentation for real-time (RT) automated kidney volumetric measurements, which are essential for clinical assessment but are traditionally time-consuming and prone to fatigue. This automation allows clinicians to concentrate on image interpretation rather than manual measurements. Complementing DL, augmented reality (AR) enhances the usability of US by projecting the display directly into the clinician's field of view, improving ergonomics and reducing the cognitive load associated with screen-to-patient transitions. Two AR-DL-assisted US pipelines on HoloLens-2 are proposed: one streams directly via the application programming interface for a wireless setup, while the other supports any US device with video output for broader accessibility. We evaluate RT feasibility and accuracy using the Open Kidney Dataset and open-source segmentation models (nnU-Net, Segmenter, YOLO with MedSAM and LiteMedSAM). Our open-source GitHub pipeline includes model implementations, measurement algorithms, and a Wi-Fi-based streaming solution, enhancing US training and diagnostics, especially in point-of-care settings.

Automated radiology report generation (RRG) aims to produce detailed textual reports from clinical imaging, such as computed tomography (CT) scans, to improve the accuracy and efficiency of diagnosis and provision of management advice. RRG is complicated by two key challenges: (1) inherent complexity in extracting relevant information from imaging data under resource constraints, and (2) difficulty in objectively evaluating discrepancies between model-generated and expert-written reports. To address these challenges, we propose $\mu^2$LLM, a $\underline{\textbf{mu}}$ltiscale $\underline{\textbf{mu}}$ltimodal large language models for RRG tasks. The novel ${\mu}^2$Tokenizer, as an intermediate layer, integrates multi-modal features from the multiscale visual tokenizer and the text tokenizer, then enhances report generation quality through direct preference optimization (DPO), guided by GREEN-RedLlama. Experimental results on four large CT image-report medical datasets demonstrate that our method outperforms existing approaches, highlighting the potential of our fine-tuned $\mu^2$LLMs on limited data for RRG tasks. At the same time, for prompt engineering, we introduce a five-stage, LLM-driven pipeline that converts routine CT reports into paired visual-question-answer triples and citation-linked reasoning narratives, creating a scalable, high-quality supervisory corpus for explainable multimodal radiology LLM. All code, datasets, and models will be publicly available in our official repository. https://github.com/Siyou-Li/u2Tokenizer