Sergios Gatidis

Scott L. Fleming, Alejandro Lozano, William J. Haberkorn et al. · stanford

The ability of large language models (LLMs) to follow natural language instructions with human-level fluency suggests many opportunities in healthcare to reduce administrative burden and improve quality of care. However, evaluating LLMs on realistic text generation tasks for healthcare remains challenging. Existing question answering datasets for electronic health record (EHR) data fail to capture the complexity of information needs and documentation burdens experienced by clinicians. To address these challenges, we introduce MedAlign, a benchmark dataset of 983 natural language instructions for EHR data. MedAlign is curated by 15 clinicians (7 specialities), includes clinician-written reference responses for 303 instructions, and provides 276 longitudinal EHRs for grounding instruction-response pairs. We used MedAlign to evaluate 6 general domain LLMs, having clinicians rank the accuracy and quality of each LLM response. We found high error rates, ranging from 35% (GPT-4) to 68% (MPT-7B-Instruct), and an 8.3% drop in accuracy moving from 32k to 2k context lengths for GPT-4. Finally, we report correlations between clinician rankings and automated natural language generation metrics as a way to rank LLMs without human review. We make MedAlign available under a research data use agreement to enable LLM evaluations on tasks aligned with clinician needs and preferences.

Jianning Li, Zongwei Zhou, Jiancheng Yang et al.

Prior to the deep learning era, shape was commonly used to describe the objects. Nowadays, state-of-the-art (SOTA) algorithms in medical imaging are predominantly diverging from computer vision, where voxel grids, meshes, point clouds, and implicit surface models are used. This is seen from numerous shape-related publications in premier vision conferences as well as the growing popularity of ShapeNet (about 51,300 models) and Princeton ModelNet (127,915 models). For the medical domain, we present a large collection of anatomical shapes (e.g., bones, organs, vessels) and 3D models of surgical instrument, called MedShapeNet, created to facilitate the translation of data-driven vision algorithms to medical applications and to adapt SOTA vision algorithms to medical problems. As a unique feature, we directly model the majority of shapes on the imaging data of real patients. As of today, MedShapeNet includes 23 dataset with more than 100,000 shapes that are paired with annotations (ground truth). Our data is freely accessible via a web interface and a Python application programming interface (API) and can be used for discriminative, reconstructive, and variational benchmarks as well as various applications in virtual, augmented, or mixed reality, and 3D printing. Exemplary, we present use cases in the fields of classification of brain tumors, facial and skull reconstructions, multi-class anatomy completion, education, and 3D printing. In future, we will extend the data and improve the interfaces. The project pages are: https://medshapenet.ikim.nrw/ and https://github.com/Jianningli/medshapenet-feedback

Justinas Zaliaduonis, Patrick Putzky, Till Richter et al.

Contrastive learning has become a leading paradigm for self-supervised representation learning, yet the conditions under which it recovers meaningful latent geometry remain incompletely understood. We develop a measure-theoretic framework formalizing the diversity condition, a support requirement on positive-pair sampling that is necessary for isometric latent recovery. We show that the standard full-support von Mises-Fisher setting implies the satisfaction of the diversity condition and as a consequence global contrastive loss minimizers recover latent geometry up to orthogonal transformation, while restricted conditionals can make non-orthogonal maps attain strictly lower asymptotic contrastive loss. We introduce a support-corrected Information Noise Contrastive Estimation (InfoNCE) variant as a theoretical fix: this correction makes orthogonal latent space recovery achievable but does not uniquely select it. Experiments on synthetic benchmarks validate the identifiability predictions, and CIFAR-10 experiments are consistent with the qualitative prediction that architectural inductive bias becomes more important when sampling diversity is limited. Together, our results clarify how sampling mechanisms and encoder inductive bias interact in contrastive representation learning.

Dave Van Veen, Cara Van Uden, Louis Blankemeier et al.

Analyzing vast textual data and summarizing key information from electronic health records imposes a substantial burden on how clinicians allocate their time. Although large language models (LLMs) have shown promise in natural language processing (NLP), their effectiveness on a diverse range of clinical summarization tasks remains unproven. In this study, we apply adaptation methods to eight LLMs, spanning four distinct clinical summarization tasks: radiology reports, patient questions, progress notes, and doctor-patient dialogue. Quantitative assessments with syntactic, semantic, and conceptual NLP metrics reveal trade-offs between models and adaptation methods. A clinical reader study with ten physicians evaluates summary completeness, correctness, and conciseness; in a majority of cases, summaries from our best adapted LLMs are either equivalent (45%) or superior (36%) compared to summaries from medical experts. The ensuing safety analysis highlights challenges faced by both LLMs and medical experts, as we connect errors to potential medical harm and categorize types of fabricated information. Our research provides evidence of LLMs outperforming medical experts in clinical text summarization across multiple tasks. This suggests that integrating LLMs into clinical workflows could alleviate documentation burden, allowing clinicians to focus more on patient care.

Jan Nikolas Morshuis, Sergios Gatidis, Matthias Hein et al.

Deep Learning (DL) methods have shown promising results for solving ill-posed inverse problems such as MR image reconstruction from undersampled $k$-space data. However, these approaches currently have no guarantees for reconstruction quality and the reliability of such algorithms is only poorly understood. Adversarial attacks offer a valuable tool to understand possible failure modes and worst case performance of DL-based reconstruction algorithms. In this paper we describe adversarial attacks on multi-coil $k$-space measurements and evaluate them on the recently proposed E2E-VarNet and a simpler UNet-based model. In contrast to prior work, the attacks are targeted to specifically alter diagnostically relevant regions. Using two realistic attack models (adversarial $k$-space noise and adversarial rotations) we are able to show that current state-of-the-art DL-based reconstruction algorithms are indeed sensitive to such perturbations to a degree where relevant diagnostic information may be lost. Surprisingly, in our experiments the UNet and the more sophisticated E2E-VarNet were similarly sensitive to such attacks. Our findings add further to the evidence that caution must be exercised as DL-based methods move closer to clinical practice.

Yan-Ran, Wang, Liangqiong Qu et al.

Despite its tremendous value for the diagnosis, treatment monitoring and surveillance of children with cancer, whole body staging with positron emission tomography (PET) is time consuming and associated with considerable radiation exposure. 100x (1% of the standard clinical dosage) ultra-low-dose/ultra-fast whole-body PET reconstruction has the potential for cancer imaging with unprecedented speed and improved safety, but it cannot be achieved by the naive use of machine learning techniques. In this study, we utilize the global similarity between baseline and follow-up PET and magnetic resonance (MR) images to develop Masked-LMCTrans, a longitudinal multi-modality co-attentional CNN-Transformer that provides interaction and joint reasoning between serial PET/MRs of the same patient. We mask the tumor area in the referenced baseline PET and reconstruct the follow-up PET scans. In this manner, Masked-LMCTrans reconstructs 100x almost-zero radio-exposure whole-body PET that was not possible before. The technique also opens a new pathway for longitudinal radiology imaging reconstruction, a significantly under-explored area to date. Our model was trained and tested with Stanford PET/MRI scans of pediatric lymphoma patients and evaluated externally on PET/MRI images from Tübingen University. The high image quality of the reconstructed 100x whole-body PET images resulting from the application of Masked-LMCTrans will substantially advance the development of safer imaging approaches and shorter exam-durations for pediatric patients, as well as expand the possibilities for frequent longitudinal monitoring of these patients by PET.

Philipp Wesp, Robbie Holland, Vasiliki Sideri-Lampretsa et al.

Vision foundation models (FMs) achieve state-of-the-art performance in medical imaging. However, they encode information in abstract latent representations that clinicians cannot interrogate or verify. The goal of this study is to investigate Sparse Autoencoders (SAEs) for replacing opaque FM image representations with human-interpretable, sparse features. We train SAEs on embeddings from BiomedParse (biomedical) and DINOv3 (general-purpose) using 909,873 CT and MRI 2D image slices from the TotalSegmentator dataset. We find that learned sparse features: (a) reconstruct original embeddings with high fidelity (R2 up to 0.941) and recover up to 87.8% of downstream performance using only 10 features (99.4% dimensionality reduction), (b) preserve semantic fidelity in image retrieval tasks, (c) correspond to specific concepts that can be expressed in language using large language model (LLM)-based auto-interpretation. (d) bridge clinical language and abstract latent representations in zero-shot language-driven image retrieval. Our work indicates SAEs are a promising pathway towards interpretable, concept-driven medical vision systems. Code repository: https://github.com/pwesp/sail.

Siying Xu, Kerstin Hammernik, Andreas Lingg et al.

Cardiac Cine Magnetic Resonance Imaging (MRI) provides an accurate assessment of heart morphology and function in clinical practice. However, MRI requires long acquisition times, with recent deep learning-based methods showing great promise to accelerate imaging and enhance reconstruction quality. Existing networks exhibit some common limitations that constrain further acceleration possibilities, including single-domain learning, reliance on a single regularization term, and equal feature contribution. To address these limitations, we propose to embed information from multiple domains, including low-rank, image, and k-space, in a novel deep learning network for MRI reconstruction, which we denote as A-LIKNet. A-LIKNet adopts a parallel-branch structure, enabling independent learning in the k-space and image domain. Coupled information sharing layers realize the information exchange between domains. Furthermore, we introduce attention mechanisms into the network to assign greater weights to more critical coils or important temporal frames. Training and testing were conducted on an in-house dataset, including 91 cardiovascular patients and 38 healthy subjects scanned with 2D cardiac Cine using retrospective undersampling. Additionally, we evaluated A-LIKNet on the real-time 8x prospectively undersampled data from the OCMR dataset. The results demonstrate that our proposed A-LIKNet outperforms existing methods and provides high-quality reconstructions. The network can effectively reconstruct highly retrospectively undersampled dynamic MR images up to 24x accelerations, indicating its potential for single breath-hold imaging.

Yabin Zhang, Chong Wang, Yunhe Gao et al.

Chest X-rays (CXRs) are among the most frequently performed imaging examinations worldwide, yet rising imaging volumes increase radiologist workload and the risk of diagnostic errors. Although artificial intelligence (AI) systems have shown promise for CXR interpretation, most generate only final predictions, without making explicit how visual evidence is translated into radiographic findings and diagnostic predictions. We present CheXOne, a reasoning-enabled vision-language model for CXR interpretation. CheXOne jointly generates diagnostic predictions and explicit, clinically grounded reasoning traces that connect visual evidence, radiographic findings, and these predictions. The model is trained on 14.7 million instruction and reasoning samples curated from 30 public datasets spanning 36 CXR interpretation tasks, using a two-stage framework that combines instruction tuning with reinforcement learning to improve reasoning quality. We evaluate CheXOne in zero-shot settings across visual question answering, report generation, visual grounding and reasoning assessment, covering 17 evaluation settings. CheXOne outperforms existing medical and general-domain foundation models and achieves strong performance on independent public benchmarks. A clinical reader study demonstrates that CheXOne-drafted reports are comparable to or better than resident-written reports in 55% of cases, while effectively addressing clinical indications and enhancing both report writing and CXR interpretation efficiency. Further analyses involving radiologists reveal that the generated reasoning traces show high clinical factuality and provide causal support for the final predictions, offering a plausible explanation for the performance gains. These results suggest that explicit reasoning can improve model performance, interpretability and clinical utility in AI-assisted CXR interpretation.

Sarah Müller, Louisa Fay, Lisa M. Koch et al.

Medical imaging cohorts are often confounded by factors such as acquisition devices, hospital sites, patient backgrounds, and many more. As a result, deep learning models tend to learn spurious correlations instead of causally related features, limiting their generalizability to new and unseen data. This problem can be addressed by minimizing dependence measures between intermediate representations of task-related and non-task-related variables. These measures include mutual information, distance correlation, and the performance of adversarial classifiers. Here, we benchmark such dependence measures for the task of preventing shortcut learning. We study a simplified setting using Morpho-MNIST and a medical imaging task with CheXpert chest radiographs. Our results provide insights into how to mitigate confounding factors in medical imaging.

Chong Wang, Yabin Zhang, Yunhe Gao et al.

Foundation models for medical imaging are typically pretrained on increasingly large datasets, following a "scale-at-all-costs" paradigm. However, this strategy faces two critical challenges: large-scale medical datasets often contain substantial redundancy and severe class imbalance that bias representation learning toward over-represented patterns, and indiscriminate training regardless of heterogeneity in data quality incurs considerable computational inefficiency. Here we demonstrate that active, principled data curation during pretraining can serve as a viable, cost-effective alternative to brute-force dataset enlargement. We introduce CheXficient, a chest X-ray (CXR) foundation model that selectively prioritizes informative training samples. CheXficient is pretrained on only 22.7% of 1,235,004 paired CXR images and reports while consuming under 27.3% of the total compute budget, yet achieving comparable or superior performance to its full-data counterpart and other large-scale pretrained models. We assess CheXficient across 20 individual benchmarks spanning 5 task types, including non-adapted off-the-shelf evaluations (zero-shot findings classification and crossmodal retrieval) and adapted downstream tasks (disease prediction, semantic segmentation, and radiology report generation). Further analyses show that CheXficient systematically prioritizes under-represented training samples, improving generalizability on long-tailed or rare conditions. Overall, our work offers practical insights into the data and computation demands for efficient pretraining and downstream adaptation of medical vision-language foundation models.

Jakob Dexl, Katharina Jeblick, Andreas Mittermeier et al.

We report the design and results of the third autoPET challenge (MICCAI 2024), which benchmarked automated lesion segmentation in whole-body PET/CT under a compositional generalization setting. Training data comprised 1,014 [18F]-FDG PET/CT studies from the University Hospital Tübingen and 597 [18F]/[68Ga]-PSMA PET/CT studies from the LMU University Hospital Munich, constituting the largest publicly available annotated PSMA PET/CT dataset to date. The held-out test set of 200 studies covered four tracer-center combinations, two of which represented unseen compositional pairings. A complementary data-centric award category isolated the contribution of data handling strategies by restricting participants to a fixed baseline model. Seventeen teams submitted 27 algorithms, predominantly nnU-Net-based 3D networks with PET/CT channel concatenation. The top-ranked algorithm achieved a mean DSC of 0.66, FNV of 3.18 mL, and FPV of 2.78 mL across all four test conditions, improving DSC by 8% and reducing the false-negative volume by 5 mL relative to the provided baseline. Ranking was stable across bootstrap resampling and alternative ranking schemes for the top tier. Beyond the benchmark, we provide an in-depth analysis of segmentation performance at the patient and lesion level. Three main conclusions can be drawn: (1) in-domain multitracer PET/CT segmentation is sufficient and probably approaching reader agreement; (2) compositional generalization to unseen tracer-center combinations remains an open problem mainly driven by systematic volume overestimation; (3) heterogeneity and case difficulty drive performance variation substantially more than the choice of algorithm among top-ranked teams.

Qi Wang, Veronika Ecker, Marcel Früh et al.

Physiological motion can affect the diagnostic quality of magnetic resonance imaging (MRI). While various retrospective motion correction methods exist, many struggle to generalize across different motion types and body regions. In particular, machine learning (ML)-based corrections are often tailored to specific applications and datasets. We hypothesize that motion artifacts, though diverse, share underlying patterns that can be disentangled and exploited. To address this, we propose a hierarchical vector-quantized (VQ) variational auto-encoder that learns a disentangled embedding of motion-to-clean image features. A codebook is deployed to capture finite collection of motion patterns at multiple resolutions, enabling coarse-to-fine correction. An auto-regressive model is trained to learn the prior distribution of motion-free images and is used at inference to guide the correction process. Unlike conventional approaches, our method does not require artifact-specific training and can generalize to unseen motion patterns. We demonstrate the approach on simulated whole-body motion artifacts and observe robust correction across varying motion severity. Our results suggest that the model effectively disentangled physical motion of the simulated motion-effective scans, therefore, improving the generalizability of the ML-based MRI motion correction. Our work of disentangling the motion features shed a light on its potential application across anatomical regions and motion types.

Asad Aali, Dave Van Veen, Yamin Ishraq Arefeen et al.

Brief hospital course (BHC) summaries are clinical documents that summarize a patient's hospital stay. While large language models (LLMs) depict remarkable capabilities in automating real-world tasks, their capabilities for healthcare applications such as synthesizing BHCs from clinical notes have not been shown. We introduce a novel pre-processed dataset, the MIMIC-IV-BHC, encapsulating clinical note and brief hospital course (BHC) pairs to adapt LLMs for BHC synthesis. Furthermore, we introduce a benchmark of the summarization performance of two general-purpose LLMs and three healthcare-adapted LLMs. Using clinical notes as input, we apply prompting-based (using in-context learning) and fine-tuning-based adaptation strategies to three open-source LLMs (Clinical-T5-Large, Llama2-13B, FLAN-UL2) and two proprietary LLMs (GPT-3.5, GPT-4). We evaluate these LLMs across multiple context-length inputs using natural language similarity metrics. We further conduct a clinical study with five clinicians, comparing clinician-written and LLM-generated BHCs across 30 samples, focusing on their potential to enhance clinical decision-making through improved summary quality. We observe that the Llama2-13B fine-tuned LLM outperforms other domain-adapted models given quantitative evaluation metrics of BLEU and BERT-Score. GPT-4 with in-context learning shows more robustness to increasing context lengths of clinical note inputs than fine-tuned Llama2-13B. Despite comparable quantitative metrics, the reader study depicts a significant preference for summaries generated by GPT-4 with in-context learning compared to both Llama2-13B fine-tuned summaries and the original summaries, highlighting the need for qualitative clinical evaluation.

Aya Ghoul, Jiazhen Pan, Andreas Lingg et al.

Accurate motion estimation at high acceleration factors enables rapid motion-compensated reconstruction in Magnetic Resonance Imaging (MRI) without compromising the diagnostic image quality. In this work, we introduce an attention-aware deep learning-based framework that can perform non-rigid pairwise registration for fully sampled and accelerated MRI. We extract local visual representations to build similarity maps between the registered image pairs at multiple resolution levels and additionally leverage long-range contextual information using a transformer-based module to alleviate ambiguities in the presence of artifacts caused by undersampling. We combine local and global dependencies to perform simultaneous coarse and fine motion estimation. The proposed method was evaluated on in-house acquired fully sampled and accelerated data of 101 patients and 62 healthy subjects undergoing cardiac and thoracic MRI. The impact of motion estimation accuracy on the downstream task of motion-compensated reconstruction was analyzed. We demonstrate that our model derives reliable and consistent motion fields across different sampling trajectories (Cartesian and radial) and acceleration factors of up to 16x for cardiac motion and 30x for respiratory motion and achieves superior image quality in motion-compensated reconstruction qualitatively and quantitatively compared to conventional and recent deep learning-based approaches. The code is publicly available at https://github.com/lab-midas/GMARAFT.

Jiaming Liu, Felix Petersen, Yunhe Gao et al.

Adversarial diffusion and diffusion-inversion methods have advanced unpaired image-to-image translation, but each faces key limitations. Adversarial approaches require target-domain adversarial loss during training, which can limit generalization to unseen data, while diffusion-inversion methods often produce low-fidelity translations due to imperfect inversion into noise-latent representations. In this work, we propose the Self-Supervised Semantic Bridge (SSB), a versatile framework that integrates external semantic priors into diffusion bridge models to enable spatially faithful translation without cross-domain supervision. Our key idea is to leverage self-supervised visual encoders to learn representations that are invariant to appearance changes but capture geometric structure, forming a shared latent space that conditions the diffusion bridges. Extensive experiments show that SSB outperforms strong prior methods for challenging medical image synthesis in both in-domain and out-of-domain settings, and extends easily to high-quality text-guided editing.

Asad Aali, Vasiliki Bikia, Maya Varma et al. · stanford

With the growing use of language models (LMs) in clinical environments, there is an immediate need to evaluate the accuracy and safety of LM-generated medical text. Currently, such evaluation relies solely on manual physician review. However, detecting errors in LM-generated text is challenging because 1) manual review is costly and 2) expert-composed reference outputs are often unavailable in real-world settings. While the "LM-as-judge" paradigm (a LM evaluating another LM) offers scalable evaluation, even frontier LMs can miss subtle but clinically significant errors. To address these challenges, we propose MedVAL, a novel, self-supervised, data-efficient distillation method that leverages synthetic data to train evaluator LMs to assess whether LM-generated medical outputs are factually consistent with inputs, without requiring physician labels or reference outputs. To evaluate LM performance, we introduce MedVAL-Bench, a dataset of 840 physician-annotated outputs across 6 diverse medical tasks capturing real-world challenges. Across 10 state-of-the-art LMs spanning open-source and proprietary models, MedVAL distillation significantly improves (p < 0.001) alignment with physicians across seen and unseen tasks, increasing average F1 scores from 66% to 83%. Despite strong baseline performance, MedVAL improves the best-performing proprietary LM (GPT-4o) by 8% without training on physician-labeled data, demonstrating a performance statistically non-inferior to a single human expert (p < 0.001). To support a scalable, risk-aware pathway towards clinical integration, we open-source: 1) Codebase (https://github.com/StanfordMIMI/MedVAL), 2) MedVAL-Bench (https://huggingface.co/datasets/stanfordmimi/MedVAL-Bench), 3) MedVAL-4B (https://huggingface.co/stanfordmimi/MedVAL-4B). Our benchmark provides evidence of LMs approaching expert-level ability in validating AI-generated medical text.

Johannes Moll, Louisa Fay, Asfandyar Azhar et al.

Radiology reports are critical for clinical decision-making but often lack a standardized format, limiting both human interpretability and machine learning (ML) applications. While large language models (LLMs) have shown strong capabilities in reformatting clinical text, their high computational requirements, lack of transparency, and data privacy concerns hinder practical deployment. To address these challenges, we explore lightweight encoder-decoder models (<300M parameters)-specifically T5 and BERT2BERT-for structuring radiology reports from the MIMIC-CXR and CheXpert Plus datasets. We benchmark these models against eight open-source LLMs (1B-70B), adapted using prefix prompting, in-context learning (ICL), and low-rank adaptation (LoRA) finetuning. Our best-performing lightweight model outperforms all LLMs adapted using prompt-based techniques on a human-annotated test set. While some LoRA-finetuned LLMs achieve modest gains over the lightweight model on the Findings section (BLEU 6.4%, ROUGE-L 4.8%, BERTScore 3.6%, F1-RadGraph 1.1%, GREEN 3.6%, and F1-SRR-BERT 4.3%), these improvements come at the cost of substantially greater computational resources. For example, LLaMA-3-70B incurred more than 400 times the inference time, cost, and carbon emissions compared to the lightweight model. These results underscore the potential of lightweight, task-specific models as sustainable and privacy-preserving solutions for structuring clinical text in resource-constrained healthcare settings.

Louis Blankemeier, Ashwin Kumar, Joseph Paul Cohen et al.

The large volume of abdominal computed tomography (CT) scans coupled with the shortage of radiologists have intensified the need for automated medical image analysis tools. Previous state-of-the-art approaches for automated analysis leverage vision-language models (VLMs) that jointly model images and radiology reports. However, current medical VLMs are generally limited to 2D images and short reports. Here to overcome these shortcomings for abdominal CT interpretation, we introduce Merlin, a 3D VLM that learns from volumetric CT scans, electronic health record data and radiology reports. This approach is enabled by a multistage pretraining framework that does not require additional manual annotations. We trained Merlin using a high-quality clinical dataset of paired CT scans (>6 million images from 15,331 CT scans), diagnosis codes (>1.8 million codes) and radiology reports (>6 million tokens). We comprehensively evaluated Merlin on 6 task types and 752 individual tasks that covered diagnostic, prognostic and quality-related tasks. The non-adapted (off-the-shelf) tasks included zero-shot classification of findings (30 findings), phenotype classification (692 phenotypes) and zero-shot cross-modal retrieval (image-to-findings and image-to-impression). The model-adapted tasks included 5-year chronic disease prediction (6 diseases), radiology report generation and 3D semantic segmentation (20 organs). We validated Merlin at scale, with internal testing on 5,137 CT scans and external testing on 44,098 CT scans from 3 independent sites and 2 public datasets. The results demonstrated high generalization across institutions and anatomies. Merlin outperformed 2D VLMs, CT foundation models and off-the-shelf radiology models. We also release our trained models, code, and dataset, available at: https://github.com/StanfordMIMI/Merlin.

Uddeshya Upadhyay, Yanbei Chen, Tobias Hepp et al.

Image-to-image translation plays a vital role in tackling various medical imaging tasks such as attenuation correction, motion correction, undersampled reconstruction, and denoising. Generative adversarial networks have been shown to achieve the state-of-the-art in generating high fidelity images for these tasks. However, the state-of-the-art GAN-based frameworks do not estimate the uncertainty in the predictions made by the network that is essential for making informed medical decisions and subsequent revision by medical experts and has recently been shown to improve the performance and interpretability of the model. In this work, we propose an uncertainty-guided progressive learning scheme for image-to-image translation. By incorporating aleatoric uncertainty as attention maps for GANs trained in a progressive manner, we generate images of increasing fidelity progressively. We demonstrate the efficacy of our model on three challenging medical image translation tasks, including PET to CT translation, undersampled MRI reconstruction, and MRI motion artefact correction. Our model generalizes well in three different tasks and improves performance over state of the art under full-supervision and weak-supervision with limited data. Code is released here: https://github.com/ExplainableML/UncerGuidedI2I

August DuMont Schütte, Jürgen Hetzel, Sergios Gatidis et al.

Privacy concerns around sharing personally identifiable information are a major practical barrier to data sharing in medical research. However, in many cases, researchers have no interest in a particular individual's information but rather aim to derive insights at the level of cohorts. Here, we utilize Generative Adversarial Networks (GANs) to create derived medical imaging datasets consisting entirely of synthetic patient data. The synthetic images ideally have, in aggregate, similar statistical properties to those of a source dataset but do not contain sensitive personal information. We assess the quality of synthetic data generated by two GAN models for chest radiographs with 14 different radiology findings and brain computed tomography (CT) scans with six types of intracranial hemorrhages. We measure the synthetic image quality by the performance difference of predictive models trained on either the synthetic or the real dataset. We find that synthetic data performance disproportionately benefits from a reduced number of unique label combinations. Our open-source benchmark also indicates that at low number of samples per class, label overfitting effects start to dominate GAN training. We additionally conducted a reader study in which trained radiologists do not perform better than random on discriminating between synthetic and real medical images for intermediate levels of resolutions. In accordance with our benchmark results, the classification accuracy of radiologists increases at higher spatial resolution levels. Our study offers valuable guidelines and outlines practical conditions under which insights derived from synthetic medical images are similar to those that would have been derived from real imaging data. Our results indicate that synthetic data sharing may be an attractive and privacy-preserving alternative to sharing real patient-level data in the right settings.

Zhihong Chen, Maya Varma, Justin Xu et al. · mila, oxford

Over 1.4 billion chest X-rays (CXRs) are performed annually due to their cost-effectiveness as an initial diagnostic test. This scale of radiological studies provides a significant opportunity to streamline CXR interpretation and documentation. While foundation models are a promising solution, the lack of publicly available large-scale datasets and benchmarks inhibits their iterative development and real-world evaluation. To overcome these challenges, we constructed a large-scale dataset (CheXinstruct), which we utilized to train a vision-language foundation model (CheXagent). We systematically demonstrated competitive performance across eight distinct task types on our novel evaluation benchmark (CheXbench). Beyond technical validation, we assessed the real-world utility of CheXagent in directly drafting radiology reports. Our clinical assessment with eight radiologists revealed a 36% time saving for residents using CheXagent-drafted reports, while attending radiologists showed no significant time difference editing resident-drafted or CheXagent-drafted reports. The CheXagent-drafted reports improved the writing efficiency of both radiology residents and attending radiologists in 81% and 61% of cases, respectively, without loss of quality. Overall, we demonstrate that CheXagent can effectively perform a variety of CXR interpretation tasks and holds potential to assist radiologists in routine clinical workflows.

Magdalini Paschali, Zhihong Chen, Louis Blankemeier et al. · stanford

Recent advances in artificial intelligence have witnessed the emergence of large-scale deep learning models capable of interpreting and generating both textual and imaging data. Such models, typically referred to as foundation models, are trained on extensive corpora of unlabeled data and demonstrate high performance across various tasks. Foundation models have recently received extensive attention from academic, industry, and regulatory bodies. Given the potentially transformative impact that foundation models can have on the field of radiology, this review aims to establish a standardized terminology concerning foundation models, with a specific focus on the requirements of training data, model training paradigms, model capabilities, and evaluation strategies. We further outline potential pathways to facilitate the training of radiology-specific foundation models, with a critical emphasis on elucidating both the benefits and challenges associated with such models. Overall, we envision that this review can unify technical advances and clinical needs in the training of foundation models for radiology in a safe and responsible manner, for ultimately benefiting patients, providers, and radiologists.

Christian Bluethgen, Dave Van Veen, Cyril Zakka et al.

At the heart of radiological practice is the challenge of integrating complex imaging data with clinical information to produce actionable insights. Nuanced application of language is key for various activities, including managing requests, describing and interpreting imaging findings in the context of clinical data, and concisely documenting and communicating the outcomes. The emergence of large language models (LLMs) offers an opportunity to improve the management and interpretation of the vast data in radiology. Despite being primarily general-purpose, these advanced computational models demonstrate impressive capabilities in specialized language-related tasks, even without specific training. Unlocking the potential of LLMs for radiology requires basic understanding of their foundations and a strategic approach to navigate their idiosyncrasies. This review, drawing from practical radiology and machine learning expertise and recent literature, provides readers insight into the potential of LLMs in radiology. It examines best practices that have so far stood the test of time in the rapidly evolving landscape of LLMs. This includes practical advice for optimizing LLM characteristics for radiology practices along with limitations, effective prompting, and fine-tuning strategies.

Chih-Ying Liu, Jeya Maria Jose Valanarasu, Camila Gonzalez et al.

Most deep learning models in medical imaging are trained on adult data with unclear performance on pediatric images. In this work, we aim to address this challenge in the context of automated anatomy segmentation in whole-body Computed Tomography (CT). We evaluate the performance of CT organ segmentation algorithms trained on adult data when applied to pediatric CT volumes and identify substantial age-dependent underperformance. We subsequently propose and evaluate strategies, including data augmentation and continual learning approaches, to achieve good segmentation accuracy across all age groups. Our best-performing model, trained using continual learning, achieves high segmentation accuracy on both adult and pediatric data (Dice scores of 0.90 and 0.84 respectively).

Eva Prakash, Jeya Maria Jose Valanarasu, Zhihong Chen et al.

Purpose: To explore best-practice approaches for generating synthetic chest X-ray images and augmenting medical imaging datasets to optimize the performance of deep learning models in downstream tasks like classification and segmentation. Materials and Methods: We utilized a latent diffusion model to condition the generation of synthetic chest X-rays on text prompts and/or segmentation masks. We explored methods like using a proxy model and using radiologist feedback to improve the quality of synthetic data. These synthetic images were then generated from relevant disease information or geometrically transformed segmentation masks and added to ground truth training set images from the CheXpert, CANDID-PTX, SIIM, and RSNA Pneumonia datasets to measure improvements in classification and segmentation model performance on the test sets. F1 and Dice scores were used to evaluate classification and segmentation respectively. One-tailed t-tests with Bonferroni correction assessed the statistical significance of performance improvements with synthetic data. Results: Across all experiments, the synthetic data we generated resulted in a maximum mean classification F1 score improvement of 0.150453 (CI: 0.099108-0.201798; P=0.0031) compared to using only real data. For segmentation, the maximum Dice score improvement was 0.14575 (CI: 0.108267-0.183233; P=0.0064). Conclusion: Best practices for generating synthetic chest X-ray images for downstream tasks include conditioning on single-disease labels or geometrically transformed segmentation masks, as well as potentially using proxy modeling for fine-tuning such generations.

Aya Ghoul, Kerstin Hammernik, Andreas Lingg et al.

In Magnetic Resonance Imaging (MRI), high temporal-resolved motion can be useful for image acquisition and reconstruction, MR-guided radiotherapy, dynamic contrast-enhancement, flow and perfusion imaging, and functional assessment of motion patterns in cardiovascular, abdominal, peristaltic, fetal, or musculoskeletal imaging. Conventionally, these motion estimates are derived through image-based registration, a particularly challenging task for complex motion patterns and high dynamic resolution. The accelerated scans in such applications result in imaging artifacts that compromise the motion estimation. In this work, we propose a novel self-supervised deep learning-based framework, dubbed the Local-All Pass Attention Network (LAPANet), for non-rigid motion estimation directly from the acquired accelerated Fourier space, i.e. k-space. The proposed approach models non-rigid motion as the cumulative sum of local translational displacements, following the Local All-Pass (LAP) registration technique. LAPANet was evaluated on cardiac motion estimation across various sampling trajectories and acceleration rates. Our results demonstrate superior accuracy compared to prior conventional and deep learning-based registration methods, accommodating as few as 2 lines/frame in a Cartesian trajectory and 3 spokes/frame in a non-Cartesian trajectory. The achieved high temporal resolution (less than 5 ms) for non-rigid motion opens new avenues for motion detection, tracking and correction in dynamic and real-time MRI applications.

Louisa Fay, Hajer Reguigui, Bin Yang et al.

Deep learning models can excel on medical tasks, yet often experience spurious correlations, known as shortcut learning, leading to poor generalization in new environments. Particularly in medical imaging, where multiple spurious correlations can coexist, misclassifications can have severe consequences. We propose MIMM-X, a framework that disentangles causal features from multiple spurious correlations by minimizing their mutual information. It enables predictions based on true underlying causal relationships rather than dataset-specific shortcuts. We evaluate MIMM-X on three datasets (UK Biobank, NAKO, CheXpert) across two imaging modalities (MRI and X-ray). Results demonstrate that MIMM-X effectively mitigates shortcut learning of multiple spurious correlations.

Frederike Lübeck, Jonas Wildberger, Frederik Träuble et al.

Cardiovascular disease (CVD) risk prediction models are essential for identifying high-risk individuals and guiding preventive actions. However, existing models struggle with the challenges of real-world clinical practice as they oversimplify patient profiles, rely on rigid input schemas, and are sensitive to distribution shifts. We developed AdaCVD, an adaptable CVD risk prediction framework built on large language models extensively fine-tuned on over half a million participants from the UK Biobank. In benchmark comparisons, AdaCVD surpasses established risk scores and standard machine learning approaches, achieving state-of-the-art performance. Crucially, for the first time, it addresses key clinical challenges across three dimensions: it flexibly incorporates comprehensive yet variable patient information; it seamlessly integrates both structured data and unstructured text; and it rapidly adapts to new patient populations using minimal additional data. In stratified analyses, it demonstrates robust performance across demographic, socioeconomic, and clinical subgroups, including underrepresented cohorts. AdaCVD offers a promising path toward more flexible, AI-driven clinical decision support tools suited to the realities of heterogeneous and dynamic healthcare environments.

Thomas Küstner, Jiazhen Pan, Haikun Qi et al.

Physiological motion, such as cardiac and respiratory motion, during Magnetic Resonance (MR) image acquisition can cause image artifacts. Motion correction techniques have been proposed to compensate for these types of motion during thoracic scans, relying on accurate motion estimation from undersampled motion-resolved reconstruction. A particular interest and challenge lie in the derivation of reliable non-rigid motion fields from the undersampled motion-resolved data. Motion estimation is usually formulated in image space via diffusion, parametric-spline, or optical flow methods. However, image-based registration can be impaired by remaining aliasing artifacts due to the undersampled motion-resolved reconstruction. In this work, we describe a formalism to perform non-rigid registration directly in the sampled Fourier space, i.e. k-space. We propose a deep-learning based approach to perform fast and accurate non-rigid registration from the undersampled k-space data. The basic working principle originates from the Local All-Pass (LAP) technique, a recently introduced optical flow-based registration. The proposed LAPNet is compared against traditional and deep learning image-based registrations and tested on fully-sampled and highly-accelerated (with two undersampling strategies) 3D respiratory motion-resolved MR images in a cohort of 40 patients with suspected liver or lung metastases and 25 healthy subjects. The proposed LAPNet provided consistent and superior performance to image-based approaches throughout different sampling trajectories and acceleration factors.

Karim Armanious, Sherif Abdulatif, Wenbin Shi et al.

Age is an essential factor in modern diagnostic procedures. However, assessment of the true biological age (BA) remains a daunting task due to the lack of reference ground-truth labels. Current BA estimation approaches are either restricted to skeletal images or rely on non-imaging modalities that yield a whole-body BA assessment. However, various organ systems may exhibit different aging characteristics due to lifestyle and genetic factors. In this initial study, we propose a new framework for organ-specific BA estimation utilizing 3D magnetic resonance image (MRI) scans. As a first step, this framework predicts the chronological age (CA) together with the corresponding patient-dependent aleatoric uncertainty. An iterative training algorithm is then utilized to segregate atypical aging patients from the given population based on the predicted uncertainty scores. In this manner, we hypothesize that training a new model on the remaining population should approximate the true BA behavior. We apply the proposed methodology on a brain MRI dataset containing healthy individuals as well as Alzheimer's patients. We demonstrate the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer's patients.

Karim Armanious, Sherif Abdulatif, Wenbin Shi et al.

The concept of biological age (BA), although important in clinical practice, is hard to grasp mainly due to the lack of a clearly defined reference standard. For specific applications, especially in pediatrics, medical image data are used for BA estimation in a routine clinical context. Beyond this young age group, BA estimation is mostly restricted to whole-body assessment using non-imaging indicators such as blood biomarkers, genetic and cellular data. However, various organ systems may exhibit different aging characteristics due to lifestyle and genetic factors. Thus, a whole-body assessment of the BA does not reflect the deviations of aging behavior between organs. To this end, we propose a new imaging-based framework for organ-specific BA estimation. In this initial study, we focus mainly on brain MRI. As a first step, we introduce a chronological age (CA) estimation framework using deep convolutional neural networks (Age-Net). We quantitatively assess the performance of this framework in comparison to existing state-of-the-art CA estimation approaches. Furthermore, we expand upon Age-Net with a novel iterative data-cleaning algorithm to segregate atypical-aging patients (BA $\not \approx$ CA) from the given population. We hypothesize that the remaining population should approximate the true BA behavior. We apply the proposed methodology on a brain magnetic resonance image (MRI) dataset containing healthy individuals as well as Alzheimer's patients with different dementia ratings. We demonstrate the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer's patients. A statistical and visualization-based analysis has provided evidence regarding the potential and current challenges of the proposed methodology.

Thomas Küstner, Tobias Hepp, Marc Fischer et al.

Purpose: To enable fast and reliable assessment of subcutaneous and visceral adipose tissue compartments derived from whole-body MRI. Methods: Quantification and localization of different adipose tissue compartments from whole-body MR images is of high interest to examine metabolic conditions. For correct identification and phenotyping of individuals at increased risk for metabolic diseases, a reliable automatic segmentation of adipose tissue into subcutaneous and visceral adipose tissue is required. In this work we propose a 3D convolutional neural network (DCNet) to provide a robust and objective segmentation. In this retrospective study, we collected 1000 cases (66$\pm$ 13 years; 523 women) from the Tuebingen Family Study and from the German Center for Diabetes research (TUEF/DZD), as well as 300 cases (53$\pm$ 11 years; 152 women) from the German National Cohort (NAKO) database for model training, validation, and testing with a transfer learning between the cohorts. These datasets had variable imaging sequences, imaging contrasts, receiver coil arrangements, scanners and imaging field strengths. The proposed DCNet was compared against a comparable 3D UNet segmentation in terms of sensitivity, specificity, precision, accuracy, and Dice overlap. Results: Fast (5-7seconds) and reliable adipose tissue segmentation can be obtained with high Dice overlap (0.94), sensitivity (96.6%), specificity (95.1%), precision (92.1%) and accuracy (98.4%) from 3D whole-body MR datasets (field of view coverage 450x450x2000mm${}^3$). Segmentation masks and adipose tissue profiles are automatically reported back to the referring physician. Conclusion: Automatic adipose tissue segmentation is feasible in 3D whole-body MR data sets and is generalizable to different epidemiological cohort studies with the proposed DCNet.

Karim Armanious, Vijeth Kumar, Sherif Abdulatif et al.

Local deformations in medical modalities are common phenomena due to a multitude of factors such as metallic implants or limited field of views in magnetic resonance imaging (MRI). Completion of the missing or distorted regions is of special interest for automatic image analysis frameworks to enhance post-processing tasks such as segmentation or classification. In this work, we propose a new generative framework for medical image inpainting, titled ipA-MedGAN. It bypasses the limitations of previous frameworks by enabling inpainting of arbitrary shaped regions without a prior localization of the regions of interest. Thorough qualitative and quantitative comparisons with other inpainting and translational approaches have illustrated the superior performance of the proposed framework for the task of brain MR inpainting.

Karim Armanious, Sherif Abdulatif, Anish Rao Bhaktharaguttu et al.

Individuals age differently depending on a multitude of different factors such as lifestyle, medical history and genetics. Often, the global chronological age is not indicative of the true ageing process. An organ-based age estimation would yield a more accurate health state assessment. In this work, we propose a new deep learning architecture for organ-based age estimation based on magnetic resonance images (MRI). The proposed network is a 3D convolutional neural network (CNN) with increased depth and width made possible by the hybrid utilization of inception and fire modules. We apply the proposed framework for the tasks of brain and knee age estimation. Quantitative comparisons against concurrent MR-based regression networks and different 2D and 3D data feeding strategies illustrated the superior performance of the proposed work.

Karim Armanious, Aastha Tanwar, Sherif Abdulatif et al.

Motion is one of the main sources for artifacts in magnetic resonance (MR) images. It can have significant consequences on the diagnostic quality of the resultant scans. Previously, supervised adversarial approaches have been suggested for the correction of MR motion artifacts. However, these approaches suffer from the limitation of required paired co-registered datasets for training which are often hard or impossible to acquire. Building upon our previous work, we introduce a new adversarial framework with a new generator architecture and loss function for the unsupervised correction of severe rigid motion artifacts in the brain region. Quantitative and qualitative comparisons with other supervised and unsupervised translation approaches showcase the enhanced performance of the introduced framework.

Karim Armanious, Chenming Jiang, Sherif Abdulatif et al.

Image-to-image translation is a new field in computer vision with multiple potential applications in the medical domain. However, for supervised image translation frameworks, co-registered datasets, paired in a pixel-wise sense, are required. This is often difficult to acquire in realistic medical scenarios. On the other hand, unsupervised translation frameworks often result in blurred translated images with unrealistic details. In this work, we propose a new unsupervised translation framework which is titled Cycle-MedGAN. The proposed framework utilizes new non-adversarial cycle losses which direct the framework to minimize the textural and perceptual discrepancies in the translated images. Qualitative and quantitative comparisons against other unsupervised translation approaches demonstrate the performance of the proposed framework for PET-CT translation and MR motion correction.

Karim Armanious, Youssef Mecky, Sergios Gatidis et al.

Numerous factors could lead to partial deteriorations of medical images. For example, metallic implants will lead to localized perturbations in MRI scans. This will affect further post-processing tasks such as attenuation correction in PET/MRI or radiation therapy planning. In this work, we propose the inpainting of medical images via Generative Adversarial Networks (GANs). The proposed framework incorporates two patch-based discriminator networks with additional style and perceptual losses for the inpainting of missing information in realistically detailed and contextually consistent manner. The proposed framework outperformed other natural image inpainting techniques both qualitatively and quantitatively on two different medical modalities.

Karim Armanious, Sergios Gatidis, Konstantin Nikolaou et al.

Motion artifacts are a primary source of magnetic resonance (MR) image quality deterioration with strong repercussions on diagnostic performance. Currently, MR motion correction is carried out either prospectively, with the help of motion tracking systems, or retrospectively by mainly utilizing computationally expensive iterative algorithms. In this paper, we utilize a new adversarial framework, titled MedGAN, for the joint retrospective correction of rigid and non-rigid motion artifacts in different body regions and without the need for a reference image. MedGAN utilizes a unique combination of non-adversarial losses and a new generator architecture to capture the textures and fine-detailed structures of the desired artifact-free MR images. Quantitative and qualitative comparisons with other adversarial techniques have illustrated the proposed model performance.

Benjamin Gutierrez-Becker, Sergios Gatidis, Daniel Gutmann et al.

Morphological analysis of organs based on images is a key task in medical imaging computing. Several approaches have been proposed for the quantitative assessment of morphological changes, and they have been widely used for the analysis of the effects of aging, disease and other factors in organ morphology. In this work, we propose a deep neural network for predicting diabetes on abdominal shapes. The network directly operates on raw point clouds without requiring mesh processing or shape alignment. Instead of relying on hand-crafted shape descriptors, an optimal representation is learned in the end-to-end training stage of the network. For comparison, we extend the state-of-the-art shape descriptor BrainPrint to the AbdomenPrint. Our results demonstrate that the network learns shape representations that better separates healthy and diabetic individuals than traditional representations.

Thomas Küstner, Sergios Gatidis, Annika Liebgott et al.

Magnetic resonance (MR) imaging offers a wide variety of imaging techniques. A large amount of data is created per examination which needs to be checked for sufficient quality in order to derive a meaningful diagnosis. This is a manual process and therefore time- and cost-intensive. Any imaging artifacts originating from scanner hardware, signal processing or induced by the patient may reduce the image quality and complicate the diagnosis or any image post-processing. Therefore, the assessment or the ensurance of sufficient image quality in an automated manner is of high interest. Usually no reference image is available or difficult to define. Therefore, classical reference-based approaches are not applicable. Model observers mimicking the human observers (HO) can assist in this task. Thus, we propose a new machine-learning-based reference-free MR image quality assessment framework which is trained on HO-derived labels to assess MR image quality immediately after each acquisition. We include the concept of active learning and present an efficient blinded reading platform to reduce the effort in the HO labeling procedure. Derived image features and the applied classifiers (support-vector-machine, deep neural network) are investigated for a cohort of 250 patients. The MR image quality assessment framework can achieve a high test accuracy of 93.7$\%$ for estimating quality classes on a 5-point Likert-scale. The proposed MR image quality assessment framework is able to provide an accurate and efficient quality estimation which can be used as a prospective quality assurance including automatic acquisition adaptation or guided MR scanner operation, and/or as a retrospective quality assessment including support of diagnostic decisions or quality control in cohort studies.

Karim Armanious, Chenming Jiang, Marc Fischer et al.

Image-to-image translation is considered a new frontier in the field of medical image analysis, with numerous potential applications. However, a large portion of recent approaches offers individualized solutions based on specialized task-specific architectures or require refinement through non-end-to-end training. In this paper, we propose a new framework, named MedGAN, for medical image-to-image translation which operates on the image level in an end-to-end manner. MedGAN builds upon recent advances in the field of generative adversarial networks (GANs) by merging the adversarial framework with a new combination of non-adversarial losses. We utilize a discriminator network as a trainable feature extractor which penalizes the discrepancy between the translated medical images and the desired modalities. Moreover, style-transfer losses are utilized to match the textures and fine-structures of the desired target images to the translated images. Additionally, we present a new generator architecture, titled CasNet, which enhances the sharpness of the translated medical outputs through progressive refinement via encoder-decoder pairs. Without any application-specific modifications, we apply MedGAN on three different tasks: PET-CT translation, correction of MR motion artefacts and PET image denoising. Perceptual analysis by radiologists and quantitative evaluations illustrate that the MedGAN outperforms other existing translation approaches.