Explore the vast range of titles available.

Recent advancements in visual foundation models open new avenues in the field of surgical instrument segmentation in medical images. Segmentation foundation models provide high segmentation accuracy for objects of interest that are selected via prompts in the form of points, bounding boxes, or text. However, the choice of suitable prompts either requires manual interaction or relies on two-stage pipelines based on supervised, typically domain-specific models. This limits their applicability for domain-agnostic surgical instrument segmentation. We propose a method for surgical instrument segmentation that leverages the power of the segmentation foundation model SAM2 while eliminating the need for a user-defined input prompt or domain-specific annotated datasets. We achieve this by utilizing an anomaly detector generated from non-instrument images to identify instruments as unseen regions and in this way, define a SAM2 input prompt based solely on image-level annotations. For three datasets for surgical instrument segmentation from diverse domains (EndoVis2017, CaDIS, and PASO-SIS), we achieve mean Normalized Surface Distances ranging from . This demonstrates the competitiveness of our method compared to alternatives, while its training- and mask-free nature makes it well-suited for surgical workflow integration. By simplifying surgical instrument segmentation, we advance the field of computer-assisted surgery and unlock a wide variety of assistance functions with minimal effort.

Clinicians spend significant time reviewing medical images and transcribing findings. By integrating visual and textual data, foundation models have the potential to reduce workloads and boost efficiency, yet their practical clinical value remains uncertain. In this study, we find that OpenAI’s ChatGPT-4o and two medical vision-language models (VLMs) significantly underperform ophthalmologists in key tasks for age-related macular degeneration (AMD). To address this, we developed a dedicated training curriculum, designed by domain specialists, to optimize VLMs for tasks related to clinical decision making. The resulting model, RetinaVLM-Specialist, significantly outperforms foundation medical VLMs and ChatGPT-4o in AMD disease staging (F1: 0.63 vs. 0.33) and referral (0.67 vs. 0.50), achieving performance comparable to junior ophthalmologists. In a reader study, two senior ophthalmologists confirmed that RetinaVLM’s reports were substantially more accurate than those written by ChatGPT-4o (64.3% vs. 14.3%). Overall, our curriculum-based approach offers a blueprint for adapting foundation models to real-world medical applications.

Age-related macular degeneration (AMD) remains a disease with high morbidity and an incompletely understood pathophysiological mechanism. The ocular blood supply has been implicated in the development of the disease process, of which most research has focused on the role of the choroid and choriocapillaris. Recently, interest has developed into the role of the retinal vasculature in AMD, particularly with the advent of optical coherence tomography angiography (OCTA), which enables non-invasive imaging of the eye’s blood vessels. This review summarises the up-to-date body of work in this field including the proposed links between observed changes in the retinal vessels and the development of AMD and potential future directions for research in this area. The review highlights that the strongest evidence supports the observation that patients with early to intermediate AMD have reduced vessel density in the superficial vascular complex of the retina, but also emphasises the need for caution when interpreting such studies due to their variable methodologies and nomenclature.

Aims Age-related macular degeneration (AMD) is characterised by a progressive loss of central vision. Intermediate AMD is a risk factor for progression to advanced stages categorised as geographic atrophy (GA) and neovascular AMD. However, rates of progression to advanced stages vary between individuals. Recent advances in imaging and computing technologies have enabled deep phenotyping of intermediate AMD. The aim of this project is to utilise machine learning (ML) and advanced statistical modelling as an innovative approach to discover novel features and accurately quantify markers of pathological retinal ageing that can individualise progression to advanced AMD. Methods The PINNACLE study consists of both retrospective and prospective parts. In the retrospective part, more than 400,000 optical coherent tomography (OCT) images collected from four University Teaching Hospitals and the UK Biobank Population Study are being pooled, centrally stored and pre-processed. With this large dataset featuring eyes with AMD at various stages and healthy controls, we aim to identify imaging biomarkers for disease progression for intermediate AMD via supervised and unsupervised ML. The prospective study part will firstly characterise the progression of intermediate AMD in patients followed between one and three years; secondly, it will validate the utility of biomarkers identified in the retrospective cohort as predictors of progression towards late AMD. Patients aged 55–90 years old with intermediate AMD in at least one eye will be recruited across multiple sites in UK, Austria and Switzerland for visual function tests, multimodal retinal imaging and genotyping. Imaging will be repeated every four months to identify early focal signs of deterioration on spectral-domain optical coherence tomography (OCT) by human graders. A focal event triggers more frequent follow-up with visual function and imaging tests. The primary outcome is the sensitivity and specificity of the OCT imaging biomarkers. Secondary outcomes include sensitivity and specificity of novel multimodal imaging characteristics at predicting disease progression, ROC curves, time from development of imaging change to development of these endpoints, structure-function correlations, structure-genotype correlation and predictive risk models. Conclusions This is one of the first studies in intermediate AMD to combine both ML, retrospective and prospective AMD patient data with the goal of identifying biomarkers of progression and to report the natural history of progression of intermediate AMD with multimodal retinal imaging.

Objectives To study the changes in vessel densities (VD) stratified by vessel diameter in the retinal superficial and deep vascular complexes (SVC/DVC) using optical coherence tomography angiography (OCTA) images obtained from people with diabetes and age-matched healthy controls. Methods We quantified the VD based on vessel diameter categorized as <10, 10–20 and >20 μm in the SVC/DVC obtained on 3 × 3 mm 2 OCTA scans using a deep learning-based segmentation and vascular graph extraction tool in people with diabetes and age-matched healthy controls. Results OCTA images obtained from 854 eyes of 854 subjects were divided into 5 groups: healthy controls ( n  = 555); people with diabetes with no diabetic retinopathy (DR, n  = 90), mild and moderate non-proliferative DR (NPDR) ( n  = 96), severe NPDR ( n  = 42) and proliferative DR (PDR) ( n  = 71). Both SVC and DVC showed significant decrease in VD with increasing DR severity ( p  < 0.001). The largest difference was observed in the <10 μm vessels of the SVC between healthy controls and no DR (13.9% lower in no DR, p  < 0.001). Progressive decrease in <10 μm vessels of the SVC and DVC was seen with increasing DR severity ( p  < 0.001). However, 10–20 μm vessels only showed decline in the DVC, but not the SVC ( p  < 0.001) and there was no change observed in the >20 μm vessels in either plexus. Conclusions Our findings suggest that OCTA is able to demonstrate a distinct vulnerability of the smallest retinal vessels in both plexuses that worsens with increasing severity of DR.

With the increasing complexity of radiotherapy treatments typical 1D and 2D quality assurance (QA) detectors may fail to detect out-of-plane dose discrepancies, in particular in the presence of motion. In this work, small samples of the PRESAGE® 3D radiochromic dosimeter were used in combination with a motion phantom to measure real-time multileaf collimator (MLC)-tracked radiotherapy treatments. A different sample of PRESAGE® was irradiated for each of three different irradiation scenarios: (1) static: static sample, without tracking (2) motion: moving sample, without tracking and (3) tracking: moving sample, with tracking. Our in-house software DynaTrack dynamically moves the linac's MLC leafs based on the target position. The doses delivered to the samples were reconstructed based on the recorded positions of the MLC and phantom during the beam delivery. PRESAGE® samples were imaged with an in-house optical-CT scanner. Comparison between simulated and measured 3D dose showed good agreement for all three irradiation scenarios (static: 99.2%; motion: 99.7%; tracking: 99.3% with a 3%, 2 mm and a 10% threshold local gamma criterion), failing only at the edges of the PRESAGE® samples (~ 6 mm). Given that the dose distributions deposited using the DynaTrack system have been independently verified, this experiment demonstrates the ability of PRESAGE to measure 3D doses correctly in a tracking context. We conclude that this methodology could be used in the future to validate the delivery of dynamic MLC-tracked radiotherapy.

Training vision-language models (VLMs) for medical report generation is often hindered by the scarcity of high-quality annotated data. This work evaluates the use of a weighted loss function to improve data efficiency. Compared to standard cross-entropy loss, which treats all token prediction errors equally, the reweighted loss shifts the focus to semantically salient tokens with outsized clinical importance. In experiments on ophthalmological report generation, we show that this simple method improves efficiency across multiple data scales, achieving similar report quality with up to ten times less training data.

Supervised contrastive learning (SupCon) has proven to be a powerful alternative to the standard cross-entropy loss for classification of multi-class balanced datasets. However, it struggles to learn well-conditioned representations of datasets with long-tailed class distributions. This problem is potentially exacerbated for binary imbalanced distributions, which are commonly encountered during many real-world problems such as medical diagnosis. In experiments on seven binary datasets of natural and medical images, we show that the performance of SupCon decreases with increasing class imbalance. To substantiate these findings, we introduce two novel metrics that evaluate the quality of the learned representation space. By measuring the class distribution in local neighborhoods, we are able to uncover structural deficiencies of the representation space that classical metrics cannot detect. Informed by these insights, we propose two new supervised contrastive learning strategies tailored to binary imbalanced datasets that improve the structure of the representation space and increase downstream classification accuracy over standard SupCon by up to 35%. We make our code available.

Recent breakthroughs in natural language processing and computer vision, driven by efficient pre-training on large datasets, have enabled foundation models to excel on a wide range of tasks. However, this potential has not yet been fully realised in time series analysis, as existing methods fail to address the heterogeneity in large time series corpora. Prevalent in domains ranging from medicine to finance, time series vary substantially in characteristics such as variate count, inter-variate relationships, temporal patterns, and sampling frequency. To address this, we introduce a novel pre-training paradigm specifically designed to handle time series heterogeneity. We propose a tokeniser with learnable domain signatures, a dual masking strategy, and a normalised cross-correlation loss, enabling our open model for general time series analysis (OTiS) to efficiently learn from large time series corpora. Extensive benchmarking on diverse tasks, such as classification, regression, and forecasting, demonstrates that OTiS outperforms state-of-the-art baselines. Our code and pre-trained weights are available at https://github.com/oetu/otis.