Optical Coherence Tomography (OCT) is a critical imaging modality for diagnosing ocular and systemic conditions, yet its accessibility is hindered by the need for specialized expertise and high computational demands. To address these challenges, we introduce ChatOCT, an offline-capable, domain-adaptive clinical decision support system (CDSS) that integrates structured expert Q&A generation, OCT-specific knowledge injection, and activation-aware model compression. Unlike existing systems, ChatOCT functions without internet access, making it suitable for low-resource environments. ChatOCT is built upon LLaMA-2-7B, incorporating domain-specific knowledge from PubMed and OCT News through a two-stage training process: (1) knowledge injection for OCT-specific expertise and (2) Q&A instruction tuning for structured, interactive diagnostic reasoning. To ensure feasibility in offline environments, we apply activation-aware weight quantization, reducing GPU memory usage to ~ 4.74 GB, enabling deployment on standard OCT hardware. A novel expert answer generation framework mitigates hallucinations by structuring responses in a multi-step process, ensuring accuracy and interpretability. ChatOCT outperforms state-of-the-art baselines such as LLaMA-2, PMC-LLaMA-13B, and ChatDoctor by 10-15 points in coherence, relevance, and clinical utility, while reducing GPU memory requirements by 79%, while maintaining real-time responsiveness (~ 20 ms inference time). Expert ophthalmologists rated ChatOCT's outputs as clinically actionable and aligned with real-world decision-making needs, confirming its potential to assist frontline healthcare providers. ChatOCT represents an innovative offline clinical decision support system for optical coherence tomography (OCT) that runs entirely on local embedded hardware, enabling real-time analysis in resource-limited settings without internet connectivity. By offering a scalable, generalizable pipeline that integrates knowledge injection, instruction tuning, and model compression, ChatOCT provides a blueprint for next-generation, resource-efficient clinical AI solutions across multiple medical domains.

Prompt engineering plays a crucial role in optimizing artificial intelligence (AI) and large language model (LLM) outputs by refining input structure, a key factor in medical applications where precision and reliability are paramount. This Clinical Perspective provides an overview of prompt engineering techniques and their relevance to interventional radiology (IR). It explores key strategies, including zero-shot, one- or few-shot, chain-of-thought, tree-of-thought, self-consistency, and directional stimulus prompting, demonstrating their application in IR-specific contexts. Practical examples illustrate how these techniques can be effectively structured for workplace and clinical use. Additionally, the article discusses best practices for designing effective prompts and addresses challenges in the clinical use of generative AI, including data privacy and regulatory concerns. It concludes with an outlook on the future of generative AI in IR, highlighting advances including retrieval-augmented generation, domain-specific LLMs, and multimodal models.

Accurate classification of focal liver lesions is crucial for diagnosis and treatment in hepatology. However, traditional supervised deep learning models depend on large-scale annotated datasets, which are often limited in medical imaging. Recently, Vision-Language models (VLMs) such as Contrastive Language-Image Pre-training model (CLIP) has been applied to image classifications. Compared to the conventional convolutional neural network (CNN), which classifiers image based on visual information only, VLM leverages multimodal learning with text and images, allowing it to learn effectively even with a limited amount of labeled data. Inspired by CLIP, we pro-pose a Liver-VLM, a model specifically designed for focal liver lesions (FLLs) classification. First, Liver-VLM incorporates class information into the text encoder without introducing additional inference overhead. Second, by calculating the pairwise cosine similarities between image and text embeddings and optimizing the model with a cross-entropy loss, Liver-VLM ef-fectively aligns image features with class-level text features. Experimental results on MPCT-FLLs dataset demonstrate that the Liver-VLM model out-performs both the standard CLIP and MedCLIP models in terms of accuracy and Area Under the Curve (AUC). Further analysis shows that using a lightweight ResNet18 backbone enhances classification performance, particularly under data-constrained conditions.

Nuclear medicine is rapidly evolving with new molecular imaging targets and advanced computational tools that promise to enhance diagnostic precision and personalized therapy. Recent years have seen a surge in novel PET and SPECT tracers, such as those targeting prostate-specific membrane antigen (PSMA) in prostate cancer, fibroblast activation protein (FAP) in tumor stroma, and tau protein in neurodegenerative disease. These tracers enable more specific visualization of disease processes compared to traditional agents, fitting into a broader shift toward precision imaging in oncology and neurology. In parallel, artificial intelligence (AI) and machine learning techniques are being integrated into tracer development and image analysis. AI-driven methods can accelerate radiopharmaceutical discovery, optimize pharmacokinetic properties, and assist in interpreting complex imaging datasets. This editorial provides an expanded overview of emerging imaging targets and techniques, including theranostic applications that pair diagnosis with radionuclide therapy, and examines how AI is augmenting nuclear medicine. We discuss the implications of these advancements within the field's historical trajectory and address the regulatory, manufacturing, and clinical challenges that must be navigated. Innovations in molecular targeting and AI are poised to transform nuclear medicine practice, enabling more personalized diagnostics and radiotheranostic strategies in the era of precision healthcare.

Cardiac Magnetic Resonance (CMR) imaging is a vital non-invasive tool for diagnosing heart diseases and evaluating cardiac health. However, the limited availability of large-scale, high-quality CMR datasets poses a major challenge to the effective application of artificial intelligence (AI) in this domain. Even the amount of unlabeled data and the health status it covers are difficult to meet the needs of model pretraining, which hinders the performance of AI models on downstream tasks. In this study, we present Cardiac Phenotype-Guided CMR Generation (CPGG), a novel approach for generating diverse CMR data that covers a wide spectrum of cardiac health status. The CPGG framework consists of two stages: in the first stage, a generative model is trained using cardiac phenotypes derived from CMR data; in the second stage, a masked autoregressive diffusion model, conditioned on these phenotypes, generates high-fidelity CMR cine sequences that capture both structural and functional features of the heart in a fine-grained manner. We synthesized a massive amount of CMR to expand the pretraining data. Experimental results show that CPGG generates high-quality synthetic CMR data, significantly improving performance on various downstream tasks, including diagnosis and cardiac phenotypes prediction. These gains are demonstrated across both public and private datasets, highlighting the effectiveness of our approach. Code is availabel at https://anonymous.4open.science/r/CPGG.

Chest X-ray (CXR) is a diagnostic tool for cardiothoracic assessment. They make up 50% of all diagnostic imaging tests. With hundreds of images examined every day, radiologists can suffer from fatigue. This fatigue may reduce diagnostic accuracy and slow down report generation. We describe a prototype computer-assisted diagnosis (CAD) pipeline employing computer vision (CV) and Natural Language Processing (NLP). It was trained and evaluated on the publicly available MIMIC-CXR dataset. We perform image quality assessment, view labelling, and segmentation-based cardiomegaly severity classification. We use the output of the severity classification for large language model-based report generation. Four board-certified radiologists assessed the output accuracy of our CAD pipeline. Across the dataset composed of 377,100 CXR images and 227,827 free-text radiology reports, our system identified 0.18% of cases with mixed-sex mentions, 0.02% of poor quality images (F1 = 0.81), and 0.28% of wrongly labelled views (accuracy 99.4%). We assigned views for 4.18% of images which have unlabelled views. Our binary cardiomegaly classification model has 95.2% accuracy. The inter-radiologist agreement on evaluating the generated report's semantics and correctness for radiologist-MIMIC is 0.62 (strict agreement) and 0.85 (relaxed agreement) similar to the radiologist-CAD agreement of 0.55 (strict) and 0.93 (relaxed). Our work found and corrected several incorrect or missing metadata annotations for the MIMIC-CXR dataset. The performance of our CAD system suggests performance on par with human radiologists. Future improvements revolve around improved text generation and the development of CV tools for other diseases.

Radiology has usually been the field of medicine that has been at the forefront of technological advances, often being the first to wholeheartedly embrace them. Whether it's from digitization to cloud side architecture, radiology has led the way for adopting the latest advances. With the advent of large language models (LLMs), especially with the unprecedented explosion of freely available ChatGPT, time is ripe for radiology and radiologists to find novel ways to use the technology to improve their workflow. Towards this, we believe these LLMs have a key role in the radiology reading room not only to expedite processes, simplify mundane and archaic tasks, but also to increase the radiologist's and radiologist trainee's knowledge base at a far faster pace. In this article, we discuss some of the ways we believe ChatGPT, and the likes can be harnessed in the reading room.

Machine learning (ML) classification of myocardial scarring in cardiac MRI is often hindered by limited explainability, particularly with convolutional neural networks (CNNs). To address this, we developed One Match (OM), an algorithm that builds on template matching to improve on both the explainability and performance of ML myocardial scaring classification. By incorporating OM, we aim to foster trust in AI models for medical diagnostics and demonstrate that improved interpretability does not have to compromise classification accuracy. Using a cardiac MRI dataset from 279 patients, this study evaluates One Match, which classifies myocardial scarring in images by matching each image to a set of labeled template images. It uses the highest correlation score from these matches for classification and is compared to a traditional sequential CNN. Enhancements such as autodidactic enhancement (AE) and patient-level classifications (PLCs) were applied to improve the predictive accuracy of both methods. Results are reported as follows: accuracy, sensitivity, specificity, precision, and F1-score. The highest classification performance was observed with the OM algorithm when enhanced by both AE and PLCs, 95.3% accuracy, 92.3% sensitivity, 96.7% specificity, 92.3% precision, and 92.3% F1-score, marking a significant improvement over the base configurations. AE alone had a positive impact on OM increasing accuracy from 89.0% to 93.2%, but decreased the accuracy of the CNN from 85.3% to 82.9%. In contrast, PLCs improved accuracy for both the CNN and OM, raising the CNN's accuracy by 4.2% and OM's by 7.4%. This study demonstrates the effectiveness of OM in classifying myocardial scars, particularly when enhanced with AE and PLCs. The interpretability of OM also enabled the examination of misclassifications, providing insights that could accelerate development and foster greater trust among clinical stakeholders.

Despite significant progress in applying large language models (LLMs) to the medical domain, several limitations still prevent them from practical applications. Among these are the constraints on model size and the lack of cohort-specific labeled datasets. In this work, we investigated the potential of improving a lightweight LLM, such as Llama 3.1-8B, through fine-tuning with datasets using synthetic labels. Two tasks are jointly trained by combining their respective instruction datasets. When the quality of the task-specific synthetic labels is relatively high (e.g., generated by GPT4-o), Llama 3.1-8B achieves satisfactory performance on the open-ended disease detection task, with a micro F1 score of 0.91. Conversely, when the quality of the task-relevant synthetic labels is relatively low (e.g., from the MIMIC-CXR dataset), fine-tuned Llama 3.1-8B is able to surpass its noisy teacher labels (micro F1 score of 0.67 v.s. 0.63) when calibrated against curated labels, indicating the strong inherent underlying capability of the model. These findings demonstrate the potential offine-tuning LLMs with synthetic labels, offering a promising direction for future research on LLM specialization in the medical domain.

Optimizing resource allocation for Parkinson's disease (PD) motor rehabilitation necessitates identifying biomarkers of responsiveness and dynamic neuroplasticity signatures underlying efficacy. A cohort study of 52 early-stage PD patients undergoing 2-week multidisciplinary intensive rehabilitation therapy (MIRT) was conducted, which stratified participants into responders and nonresponders. A multimodal analysis of resting-state electroencephalography (EEG) microstates and functional magnetic resonance imaging (fMRI) coactivation patterns was performed to characterize MIRT-induced spatiotemporal network reorganization. Responders demonstrated clinically meaningful improvement in motor symptoms, exceeding the minimal clinically important difference threshold of 3.25 on the Unified PD Rating Scale part III, alongside significant reductions in bradykinesia and a significant enhancement in quality-of-life scores at the 3-month follow-up. Resting-state EEG in responders showed a significant attenuation in microstate C and a significant enhancement in microstate D occurrences, along with significantly increased transitions from microstate A/B to D, which significantly correlated with motor function, especially in bradykinesia gains. Concurrently, fMRI analyses identified a prolonged dwell time of the dorsal attention network coactivation/ventral attention network deactivation pattern, which was significantly inversely associated with microstate C occurrence and significantly linked to motor improvement. The identified brain spatiotemporal neural markers were validated using machine learning models to assess the efficacy of MIRT in motor rehabilitation for PD patients, achieving an average accuracy rate of 86%. These findings suggest that MIRT may facilitate a shift in neural networks from sensory processing to higher-order cognitive control, with the dynamic reallocation of attentional resources. This preliminary study validates the necessity of integrating cognitive-motor strategies for the motor rehabilitation of PD and identifies novel neural markers for assessing treatment efficacy.