Vajira Thambawita

Debesh Jha, Vanshali Sharma, Debapriya Banik et al. · oxford

Automatic analysis of colonoscopy images has been an active field of research motivated by the importance of early detection of precancerous polyps. However, detecting polyps during the live examination can be challenging due to various factors such as variation of skills and experience among the endoscopists, lack of attentiveness, and fatigue leading to a high polyp miss-rate. Deep learning has emerged as a promising solution to this challenge as it can assist endoscopists in detecting and classifying overlooked polyps and abnormalities in real time. In addition to the algorithm's accuracy, transparency and interpretability are crucial to explaining the whys and hows of the algorithm's prediction. Further, most algorithms are developed in private data, closed source, or proprietary software, and methods lack reproducibility. Therefore, to promote the development of efficient and transparent methods, we have organized the "Medico automatic polyp segmentation (Medico 2020)" and "MedAI: Transparency in Medical Image Segmentation (MedAI 2021)" competitions. We present a comprehensive summary and analyze each contribution, highlight the strength of the best-performing methods, and discuss the possibility of clinical translations of such methods into the clinic. For the transparency task, a multi-disciplinary team, including expert gastroenterologists, accessed each submission and evaluated the team based on open-source practices, failure case analysis, ablation studies, usability and understandability of evaluations to gain a deeper understanding of the models' credibility for clinical deployment. Through the comprehensive analysis of the challenge, we not only highlight the advancements in polyp and surgical instrument segmentation but also encourage qualitative evaluation for building more transparent and understandable AI-based colonoscopy systems.

Matthias Eisenmann, Annika Reinke, Vivienn Weru et al. · utoronto

The number of international benchmarking competitions is steadily increasing in various fields of machine learning (ML) research and practice. So far, however, little is known about the common practice as well as bottlenecks faced by the community in tackling the research questions posed. To shed light on the status quo of algorithm development in the specific field of biomedical imaging analysis, we designed an international survey that was issued to all participants of challenges conducted in conjunction with the IEEE ISBI 2021 and MICCAI 2021 conferences (80 competitions in total). The survey covered participants' expertise and working environments, their chosen strategies, as well as algorithm characteristics. A median of 72% challenge participants took part in the survey. According to our results, knowledge exchange was the primary incentive (70%) for participation, while the reception of prize money played only a minor role (16%). While a median of 80 working hours was spent on method development, a large portion of participants stated that they did not have enough time for method development (32%). 25% perceived the infrastructure to be a bottleneck. Overall, 94% of all solutions were deep learning-based. Of these, 84% were based on standard architectures. 43% of the respondents reported that the data samples (e.g., images) were too large to be processed at once. This was most commonly addressed by patch-based training (69%), downsampling (37%), and solving 3D analysis tasks as a series of 2D tasks. K-fold cross-validation on the training set was performed by only 37% of the participants and only 50% of the participants performed ensembling based on multiple identical models (61%) or heterogeneous models (39%). 48% of the respondents applied postprocessing steps.

Vajira Thambawita, Steven A. Hicks, Andrea M. Storås et al.

A manual assessment of sperm motility requires microscopy observation, which is challenging due to the fast-moving spermatozoa in the field of view. To obtain correct results, manual evaluation requires extensive training. Therefore, computer-assisted sperm analysis (CASA) has become increasingly used in clinics. Despite this, more data is needed to train supervised machine learning approaches in order to improve accuracy and reliability in the assessment of sperm motility and kinematics. In this regard, we provide a dataset called VISEM-Tracking with 20 video recordings of 30 seconds (comprising 29,196 frames) of wet sperm preparations with manually annotated bounding-box coordinates and a set of sperm characteristics analyzed by experts in the domain. In addition to the annotated data, we provide unlabeled video clips for easy-to-use access and analysis of the data via methods such as self- or unsupervised learning. As part of this paper, we present baseline sperm detection performances using the YOLOv5 deep learning (DL) model trained on the VISEM-Tracking dataset. As a result, we show that the dataset can be used to train complex DL models to analyze spermatozoa.

Roman Macháček, Leila Mozaffari, Zahra Sepasdar et al.

In order to take advantage of AI solutions in endoscopy diagnostics, we must overcome the issue of limited annotations. These limitations are caused by the high privacy concerns in the medical field and the requirement of getting aid from experts for the time-consuming and costly medical data annotation process. In computer vision, image synthesis has made a significant contribution in recent years as a result of the progress of generative adversarial networks (GANs) and diffusion probabilistic models (DPM). Novel DPMs have outperformed GANs in text, image, and video generation tasks. Therefore, this study proposes a conditional DPM framework to generate synthetic GI polyp images conditioned on given generated segmentation masks. Our experimental results show that our system can generate an unlimited number of high-fidelity synthetic polyp images with the corresponding ground truth masks of polyps. To test the usefulness of the generated data, we trained binary image segmentation models to study the effect of using synthetic data. Results show that the best micro-imagewise IOU of 0.7751 was achieved from DeepLabv3+ when the training data consists of both real data and synthetic data. However, the results reflect that achieving good segmentation performance with synthetic data heavily depends on model architectures.

Jan Andre Fagereng, Vajira Thambawita, Andrea M. Storås et al.

Early identification of a polyp in the lower gastrointestinal (GI) tract can lead to prevention of life-threatening colorectal cancer. Developing computer-aided diagnosis (CAD) systems to detect polyps can improve detection accuracy and efficiency and save the time of the domain experts called endoscopists. Lack of annotated data is a common challenge when building CAD systems. Generating synthetic medical data is an active research area to overcome the problem of having relatively few true positive cases in the medical domain. To be able to efficiently train machine learning (ML) models, which are the core of CAD systems, a considerable amount of data should be used. In this respect, we propose the PolypConnect pipeline, which can convert non-polyp images into polyp images to increase the size of training datasets for training. We present the whole pipeline with quantitative and qualitative evaluations involving endoscopists. The polyp segmentation model trained using synthetic data, and real data shows a 5.1% improvement of mean intersection over union (mIOU), compared to the model trained only using real data. The codes of all the experiments are available on GitHub to reproduce the results.

Hugo L. Hammer, Vajira Thambawita, Pål Halvorsen

A narrated e-book combines synchronized audio with digital text, highlighting the currently spoken word or sentence during playback. This format supports early literacy and assists individuals with reading challenges, while also allowing general readers to seamlessly switch between reading and listening. With the emergence of natural-sounding neural Text-to-Speech (TTS) technology, several commercial services have been developed to leverage these technology for converting standard text e-books into high-quality narrated e-books. However, no open-source solutions currently exist to perform this task. In this paper, we present Calliope, an open-source framework designed to fill this gap. Our method leverages state-of-the-art open-source TTS to convert a text e-book into a narrated e-book in the EPUB 3 Media Overlay format. The method offers several innovative steps: audio timestamps are captured directly during TTS, ensuring exact synchronization between narration and text highlighting; the publisher's original typography, styling, and embedded media are strictly preserved; and the entire pipeline operates offline. This offline capability eliminates recurring API costs, mitigates privacy concerns, and avoids copyright compliance issues associated with cloud-based services. The framework currently supports the state-of-the-art open-source TTS systems XTTS-v2 and Chatterbox. A potential alternative approach involves first generating narration via TTS and subsequently synchronizing it with the text using forced alignment. However, while our method ensures exact synchronization, our experiments show that forced alignment introduces drift between the audio and text highlighting significant enough to degrade the reading experience. Source code and usage instructions are available at https://github.com/hugohammer/TTS-Narrated-Ebook-Creator.git.

Sushant Gautam, Cise Midoglu, Vajira Thambawita et al.

Hallucinations in video-capable vision-language models (Video-VLMs) remain frequent and high-confidence, while existing uncertainty metrics often fail to align with correctness. We introduce VideoHEDGE, a modular framework for hallucination detection in video question answering that extends entropy-based reliability estimation from images to temporally structured inputs. Given a video-question pair, VideoHEDGE draws a baseline answer and multiple high-temperature generations from both clean clips and photometrically and spatiotemporally perturbed variants, then clusters the resulting textual outputs into semantic hypotheses using either Natural Language Inference (NLI)-based or embedding-based methods. Cluster-level probability masses yield three reliability scores: Semantic Entropy (SE), RadFlag, and Vision-Amplified Semantic Entropy (VASE). We evaluate VideoHEDGE on the SoccerChat benchmark using an LLM-as-a-judge to obtain binary hallucination labels. Across three 7B Video-VLMs (Qwen2-VL, Qwen2.5-VL, and a SoccerChat-finetuned model), VASE consistently achieves the highest ROC-AUC, especially at larger distortion budgets, while SE and RadFlag often operate near chance. We further show that embedding-based clustering matches NLI-based clustering in detection performance at substantially lower computational cost, and that domain fine-tuning reduces hallucination frequency but yields only modest improvements in calibration. The hedge-bench PyPI library enables reproducible and extensible benchmarking, with full code and experimental resources available at https://github.com/Simula/HEDGE#videohedge .

Birk Torpmann-Hagen, Vajira Thambawita, Kyrre Glette et al.

Generalizability is seen as one of the major challenges in deep learning, in particular in the domain of medical imaging, where a change of hospital or in imaging routines can lead to a complete failure of a model. To tackle this, we introduce Consistency Training, a training procedure and alternative to data augmentation based on maximizing models' prediction consistency across augmented and unaugmented data in order to facilitate better out-of-distribution generalization. To this end, we develop a novel region-based segmentation loss function called Segmentation Inconsistency Loss (SIL), which considers the differences between pairs of augmented and unaugmented predictions and labels. We demonstrate that Consistency Training outperforms conventional data augmentation on several out-of-distribution datasets on polyp segmentation, a popular medical task.

Vajira Thambawita, Andrea M. Storås, Steven A. Hicks et al.

Head and neck cancers are the fifth most common cancer worldwide, and recently, analysis of Positron Emission Tomography (PET) and Computed Tomography (CT) images has been proposed to identify patients with a prognosis. Even though the results look promising, more research is needed to further validate and improve the results. This paper presents the work done by team MLC for the 2022 version of the HECKTOR grand challenge held at MICCAI 2022. For Task 1, the automatic segmentation task, our approach was, in contrast to earlier solutions using 3D segmentation, to keep it as simple as possible using a 2D model, analyzing every slice as a standalone image. In addition, we were interested in understanding how different modalities influence the results. We proposed two approaches; one using only the CT scans to make predictions and another using a combination of the CT and PET scans. For Task 2, the prediction of recurrence-free survival, we first proposed two approaches, one where we only use patient data and one where we combined the patient data with segmentations from the image model. For the prediction of the first two approaches, we used Random Forest. In our third approach, we combined patient data and image data using XGBoost. Low kidney function might worsen cancer prognosis. In this approach, we therefore estimated the kidney function of the patients and included it as a feature. Overall, we conclude that our simple methods were not able to compete with the highest-ranking submissions, but we still obtained reasonably good scores. We also got interesting insights into how the combination of different modalities can influence the segmentation and predictions.

Vladimir Monakhov, Vajira Thambawita, Pål Halvorsen et al.

The interest for video anomaly detection systems has gained traction for the past few years. The current approaches use deep learning to perform anomaly detection in videos, but this approach has multiple problems. For starters, deep learning in general has issues with noise, concept drift, explainability, and training data volumes. Additionally, anomaly detection in itself is a complex task and faces challenges such as unknowness, heterogeneity, and class imbalance. Anomaly detection using deep learning is therefore mainly constrained to generative models such as generative adversarial networks and autoencoders due to their unsupervised nature, but even they suffer from general deep learning issues and are hard to train properly. In this paper, we explore the capabilities of the Hierarchical Temporal Memory (HTM) algorithm to perform anomaly detection in videos, as it has favorable properties such as noise tolerance and online learning which combats concept drift. We introduce a novel version of HTM, namely, Grid HTM, which is an HTM-based architecture specifically for anomaly detection in complex videos such as surveillance footage.

Annika Reinke, Evangelia Christodoulou, Sthuthi Sadananda et al.

Benchmarking competitions are central to the development of artificial intelligence (AI) in medical imaging, defining performance standards and shaping methodological progress. However, it remains unclear whether these benchmarks provide data that are sufficiently representative, accessible, and reusable to support clinically meaningful AI. In this work, we assess fairness along two complementary dimensions: (1) whether challenge datasets are representative of real-world clinical diversity, and (2) whether they are accessible and legally reusable in line with the FAIR principles. To address this question, we conducted a large-scale systematic study of 241 biomedical image analysis challenges comprising 458 tasks across 19 imaging modalities. Our findings show substantial biases in dataset composition, including geographic location, modality-, and problem type-related biases, indicating that current benchmarks do not adequately reflect real-world clinical diversity. Despite their widespread influence, challenge datasets were frequently constrained by restrictive or ambiguous access conditions, inconsistent or non-compliant licensing practices, and incomplete documentation, limiting reproducibility and long-term reuse. Together, these shortcomings expose foundational fairness limitations in our benchmarking ecosystem and highlight a disconnect between leaderboard success and clinical relevance.

Sushant Gautam, Andrea Storås, Cise Midoglu et al.

We introduce Kvasir-VQA, an extended dataset derived from the HyperKvasir and Kvasir-Instrument datasets, augmented with question-and-answer annotations to facilitate advanced machine learning tasks in Gastrointestinal (GI) diagnostics. This dataset comprises 6,500 annotated images spanning various GI tract conditions and surgical instruments, and it supports multiple question types including yes/no, choice, location, and numerical count. The dataset is intended for applications such as image captioning, Visual Question Answering (VQA), text-based generation of synthetic medical images, object detection, and classification. Our experiments demonstrate the dataset's effectiveness in training models for three selected tasks, showcasing significant applications in medical image analysis and diagnostics. We also present evaluation metrics for each task, highlighting the usability and versatility of our dataset. The dataset and supporting artifacts are available at https://datasets.simula.no/kvasir-vqa.

Madhura Edirisooriya, Dasuni Kawya, Ishan Kumarasinghe et al.

Deep learning in cardiac MRI (CMR) is fundamentally constrained by both data scarcity and privacy regulations. This study systematically benchmarks three generative architectures: Denoising Diffusion Probabilistic Models (DDPM), Latent Diffusion Models (LDM), and Flow Matching (FM) for synthetic CMR generation. Utilizing a two-stage pipeline where anatomical masks condition image synthesis, we evaluate generated data across three critical axes: fidelity, utility, and privacy. Our results show that diffusion-based models, particularly DDPM, provide the most effective balance between downstream segmentation utility, image fidelity, and privacy preservation under limited-data conditions, while FM demonstrates promising privacy characteristics with slightly lower task-level performance. These findings quantify the trade-offs between cross-domain generalization and patient confidentiality, establishing a framework for safe and effective synthetic data augmentation in medical imaging.

Vajira Thambawita, Jonas L. Isaksen, Jørgen K. Kanters et al.

Deep learning has achieved expert-level performance in automated electrocardiogram (ECG) diagnosis, yet the "black-box" nature of these models hinders their clinical deployment. Trust in medical AI requires not just high accuracy but also transparency regarding the specific physiological features driving predictions. Existing explainability methods for ECGs typically rely on post-hoc approximations (e.g., Grad-CAM and SHAP), which can be unstable, computationally expensive, and unfaithful to the model's actual decision-making process. In this work, we propose the ECG-IMN, an Interpretable Mesomorphic Neural Network tailored for high-resolution 12-lead ECG classification. Unlike standard classifiers, the ECG-IMN functions as a hypernetwork: a deep convolutional backbone generates the parameters of a strictly linear model specific to each input sample. This architecture enforces intrinsic interpretability, as the decision logic is mathematically transparent and the generated weights (W) serve as exact, high-resolution feature attribution maps. We introduce a transition decoder that effectively maps latent features to sample-wise weights, enabling precise localization of pathological evidence (e.g., ST-elevation, T-wave inversion) in both time and lead dimensions. We evaluate our approach on the PTB-XL dataset for classification tasks, demonstrating that the ECG-IMN achieves competitive predictive performance (AUROC comparable to black-box baselines) while providing faithful, instance-specific explanations. By explicitly decoupling parameter generation from prediction execution, our framework bridges the gap between deep learning capability and clinical trustworthiness, offering a principled path toward "white-box" cardiac diagnostics.

Arjan Mahmuod, Adrian Rod Hammerstad, Muzaffar Yousef et al.

Deep learning models for atrial fibrillation (AF) detection are increasingly trained on heterogeneous electrocardiogram (ECG) datasets with varying sampling frequencies, yet the specific consequences of these discrepancies on model performance, calibration, and robustness remain insufficiently characterized. To address this, we conducted a systematic benchmark using 12-lead, 10-second recordings from the PTB-XL dataset, resampled to target frequencies of 62, 100, 250, and 500 Hz, to evaluate a standard 1-D Convolutional Neural Network (CNN) and a hybrid CNN-Long Short-Term Memory (LSTM) architecture under a rigorous patient-safe cross-validation framework. Our analysis reveals that sampling frequency significantly impacts detection metrics in an architecture-dependent manner; the hybrid CNN-LSTM model demonstrated optimal performance and consistent calibration at intermediate frequencies (100-250 Hz), whereas the 1-D CNN baseline exhibited marked degradation in accuracy and sensitivity at 500 Hz, suggesting increased susceptibility to high-frequency noise. We conclude that ECG sampling frequency is a critical, underappreciated factor in arrhythmia detection, and future foundation models must explicitly control for temporal resolution to ensure clinical reliability and reproducibility.

Adam Jakobsen, Sushant Gautam, Hugo Lewi Hammer et al.

AI systems in healthcare research have shown potential to increase patient throughput and assist clinicians, yet progress is constrained by limited access to real patient data. To address this issue, we present a zero-shot, knowledge-guided framework for psychiatric tabular data in which large language models (LLMs) are steered via Retrieval-Augmented Generation using the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and the International Classification of Diseases (ICD-10). We conducted experiments using different combinations of knowledge bases to generate privacy-preserving synthetic data. The resulting models were benchmarked against two state-of-the-art deep learning models for synthetic tabular data generation, namely CTGAN and TVAE, both of which rely on real data and therefore entail potential privacy risks. Evaluation was performed on six anxiety-related disorders: specific phobia, social anxiety disorder, agoraphobia, generalized anxiety disorder, separation anxiety disorder, and panic disorder. CTGAN typically achieves the best marginals and multivariate structure, while the knowledge-augmented LLM is competitive on pairwise structure and attains the lowest pairwise error in separation anxiety and social anxiety. An ablation study shows that clinical retrieval reliably improves univariate and pairwise fidelity over a no-retrieval LLM. Privacy analyses indicate that the real data-free LLM yields modest overlaps and a low average linkage risk comparable to CTGAN, whereas TVAE exhibits extensive duplication despite a low k-map score. Overall, grounding an LLM in clinical knowledge enables high-quality, privacy-preserving synthetic psychiatric data when real datasets are unavailable or cannot be shared.

Ishan Kumarasinghe, Dasuni Kawya, Madhura Edirisooriya et al.

Synthetic cardiac MRI (CMRI) generation has emerged as a promising strategy to overcome the scarcity of annotated medical imaging data. Recent advances in GANs, VAEs, diffusion probabilistic models, and flow-matching techniques aim to generate anatomically accurate images while addressing challenges such as limited labeled datasets, vendor variability, and risks of privacy leakage through model memorization. Maskconditioned generation improves structural fidelity by guiding synthesis with segmentation maps, while diffusion and flowmatching models offer strong boundary preservation and efficient deterministic transformations. Cross-domain generalization is further supported through vendor-style conditioning and preprocessing steps like intensity normalization. To ensure privacy, studies increasingly incorporate membership inference attacks, nearest-neighbor analyses, and differential privacy mechanisms. Utility evaluations commonly measure downstream segmentation performance, with evidence showing that anatomically constrained synthetic data can enhance accuracy and robustness across multi-vendor settings. This review aims to compare existing CMRI generation approaches through the lenses of fidelity, utility, and privacy, highlighting current limitations and the need for integrated, evaluation-driven frameworks for reliable clinical workflows.

Pegah Salehi, Sajad Amouei Sheshkal, Vajira Thambawita et al.

This paper examines the integration of real-time talking-head generation for interviewer training, focusing on overcoming challenges in Audio Feature Extraction (AFE), which often introduces latency and limits responsiveness in real-time applications. To address these issues, we propose and implement a fully integrated system that replaces conventional AFE models with Open AI's Whisper, leveraging its encoder to optimize processing and improve overall system efficiency. Our evaluation of two open-source real-time models across three different datasets shows that Whisper not only accelerates processing but also improves specific aspects of rendering quality, resulting in more realistic and responsive talking-head interactions. These advancements make the system a more effective tool for immersive, interactive training applications, expanding the potential of AI-driven avatars in interviewer training.

Sushant Gautam, Vajira Thambawita, Michael Riegler et al.

The Medico 2025 challenge addresses Visual Question Answering (VQA) for Gastrointestinal (GI) imaging, organized as part of the MediaEval task series. The challenge focuses on developing Explainable Artificial Intelligence (XAI) models that answer clinically relevant questions based on GI endoscopy images while providing interpretable justifications aligned with medical reasoning. It introduces two subtasks: (1) answering diverse types of visual questions using the Kvasir-VQA-x1 dataset, and (2) generating multimodal explanations to support clinical decision-making. The Kvasir-VQA-x1 dataset, created from 6,500 images and 159,549 complex question-answer (QA) pairs, serves as the benchmark for the challenge. By combining quantitative performance metrics and expert-reviewed explainability assessments, this task aims to advance trustworthy Artificial Intelligence (AI) in medical image analysis. Instructions, data access, and an updated guide for participation are available in the official competition repository: https://github.com/simula/MediaEval-Medico-2025

Sushant Gautam, Cise Midoglu, Vajira Thambawita et al.

The integration of artificial intelligence in sports analytics has transformed soccer video understanding, enabling real-time, automated insights into complex game dynamics. Traditional approaches rely on isolated data streams, limiting their effectiveness in capturing the full context of a match. To address this, we introduce SoccerChat, a multimodal conversational AI framework that integrates visual and textual data for enhanced soccer video comprehension. Leveraging the extensive SoccerNet dataset, enriched with jersey color annotations and automatic speech recognition (ASR) transcripts, SoccerChat is fine-tuned on a structured video instruction dataset to facilitate accurate game understanding, event classification, and referee decision making. We benchmark SoccerChat on action classification and referee decision-making tasks, demonstrating its performance in general soccer event comprehension while maintaining competitive accuracy in referee decision making. Our findings highlight the importance of multimodal integration in advancing soccer analytics, paving the way for more interactive and explainable AI-driven sports analysis. https://github.com/simula/SoccerChat

Vajira Thambawita, Pegah Salehi, Sajad Amouei Sheshkal et al.

Analyzing medical data to find abnormalities is a time-consuming and costly task, particularly for rare abnormalities, requiring tremendous efforts from medical experts. Artificial intelligence has become a popular tool for the automatic processing of medical data, acting as a supportive tool for doctors. However, the machine learning models used to build these tools are highly dependent on the data used to train them. Large amounts of data can be difficult to obtain in medicine due to privacy, expensive and time-consuming annotations, and a general lack of data samples for infrequent lesions. Here, we present a novel synthetic data generation pipeline, called SinGAN-Seg, to produce synthetic medical images with corresponding masks using a single training image. Our method is different from the traditional GANs because our model needs only a single image and the corresponding ground truth to train. Our method produces alternative artificial segmentation datasets with ground truth masks when real datasets are not allowed to share. The pipeline is evaluated using qualitative and quantitative comparisons between real and synthetic data to show that the style transfer technique used in our pipeline significantly improves the quality of the generated data and our method is better than other state-of-the-art GANs to prepare synthetic images when the size of training datasets are limited. By training UNet++ using both real and the synthetic data generated from the SinGAN-Seg pipeline, we show that models trained with synthetic data have very close performances to those trained on real data when the datasets have a considerable amount of data. In contrast, Synthetic data generated from the SinGAN-Seg pipeline can improve the performance of segmentation models when training datasets do not have a considerable amount of data. The code is available on GitHub.

Sushant Gautam, Mehdi Houshmand Sarkhoosh, Jan Held et al.

The application of Automatic Speech Recognition (ASR) technology in soccer offers numerous opportunities for sports analytics. Specifically, extracting audio commentaries with ASR provides valuable insights into the events of the game, and opens the door to several downstream applications such as automatic highlight generation. This paper presents SoccerNet-Echoes, an augmentation of the SoccerNet dataset with automatically generated transcriptions of audio commentaries from soccer game broadcasts, enhancing video content with rich layers of textual information derived from the game audio using ASR. These textual commentaries, generated using the Whisper model and translated with Google Translate, extend the usefulness of the SoccerNet dataset in diverse applications such as enhanced action spotting, automatic caption generation, and game summarization. By incorporating textual data alongside visual and auditory content, SoccerNet-Echoes aims to serve as a comprehensive resource for the development of algorithms specialized in capturing the dynamics of soccer games. We detail the methods involved in the curation of this dataset and the integration of ASR. We also highlight the implications of a multimodal approach in sports analytics, and how the enriched dataset can support diverse applications, thus broadening the scope of research and development in the field of sports analytics.

Oriana Presacan, Alexandru Dorobantiu, Vajira Thambawita et al.

Accurate embryo morphology assessment is essential in assisted reproductive technology for selecting the most viable embryo. Artificial intelligence has the potential to enhance this process. However, the limited availability of embryo data presents challenges for training deep learning models. To address this, we trained two generative models using two datasets-one we created and made publicly available, and one existing public dataset-to generate synthetic embryo images at various cell stages, including 2-cell, 4-cell, 8-cell, morula, and blastocyst. These were combined with real images to train classification models for embryo cell stage prediction. Our results demonstrate that incorporating synthetic images alongside real data improved classification performance, with the model achieving 97% accuracy compared to 94.5% when trained solely on real data. This trend remained consistent when tested on an external Blastocyst dataset from a different clinic. Notably, even when trained exclusively on synthetic data and tested on real data, the model achieved a high accuracy of 92%. Furthermore, combining synthetic data from both generative models yielded better classification results than using data from a single generative model. Four embryologists evaluated the fidelity of the synthetic images through a Turing test, during which they annotated inaccuracies and offered feedback. The analysis showed the diffusion model outperformed the generative adversarial network, deceiving embryologists 66.6% versus 25.3% and achieving lower Frechet inception distance scores.

Lucia Borrego, Vajira Thambawita, Marco Ciuffreda et al.

The scarcity of high-quality segmentation masks remains a major bottleneck for medical image analysis, particularly in non-contrast CT (NCCT) neuroimaging, where manual annotation is costly and variable. To address this limitation, we propose an anatomy-preserving generative framework for the unconditional synthesis of multi-class brain segmentation masks, including ischemic infarcts. The proposed approach combines a variational autoencoder trained exclusively on segmentation masks to learn an anatomical latent representation, with a diffusion model operating in this latent space to generate new samples from pure noise. At inference, synthetic masks are obtained by decoding denoised latent vectors through the frozen VAE decoder, with optional coarse control over lesion presence via a binary prompt. Qualitative results show that the generated masks preserve global brain anatomy, discrete tissue semantics, and realistic variability, while avoiding the structural artifacts commonly observed in pixel-space generative models. Overall, the proposed framework offers a simple and scalable solution for anatomy-aware mask generation in data-scarce medical imaging scenarios.

Oriana Presacan, Alireza Nik, Vajira Thambawita et al.

Large Language Models (LLMs) rely on various decoding strategies to generate text, and these choices can significantly affect output quality. In healthcare, where accuracy is critical, the impact of decoding strategies remains underexplored. We investigate this effect in five open-ended medical tasks, including translation, summarization, question answering, dialogue, and image captioning, evaluating 11 decoding strategies with medically specialized and general-purpose LLMs of different sizes. Our results show that deterministic strategies generally outperform stochastic ones: beam search achieves the highest scores, while η and top-k sampling perform worst. Slower decoding methods tend to yield better quality. Larger models achieve higher scores overall but have longer inference times and are no more robust to decoding. Surprisingly, while medical LLMs outperform general ones in two of the five tasks, statistical analysis shows no overall performance advantage and reveals greater sensitivity to decoding choice. We further compare multiple evaluation metrics and find that correlations vary by task, with MAUVE showing weak agreement with BERTScore and ROUGE, as well as greater sensitivity to the decoding strategy. These results highlight the need for careful selection of decoding methods in medical applications, as their influence can sometimes exceed that of model choice.

J. M. I. H. Jayakody, A. M. H. H. Alahakoon, C. R. M. Perera et al.

The paradigm of electrocardiogram (ECG) analysis has evolved into real-time digital analysis, facilitated by artificial intelligence (AI) and machine learning (ML), which has improved the diagnostic precision and predictive capacity of cardiac diseases. This work proposes a novel deep learning (DL) architecture called the attention-integrated convolutional residual network (AICRN) to regress key ECG parameters such as the PR interval, the QT interval, the QRS duration, the heart rate, the peak amplitude of the R wave, and the amplitude of the T wave for interpretable ECG analysis. Our architecture is specially designed with spatial and channel attention-related mechanisms to address the type and spatial location of the ECG features for regression. The models employ a convolutional residual network to address vanishing and exploding gradient problems. The designed system addresses traditional analysis challenges, such as loss of focus due to human errors, and facilitates the fast and easy detection of cardiac events, thereby reducing the manual efforts required to solve analysis tasks. AICRN models outperform existing models in parameter regression with higher precision. This work demonstrates that DL can play a crucial role in the interpretability and precision of ECG analysis, opening up new clinical applications for cardiac monitoring and management.

Ridma Jayasundara, Ishan Fernando, Adeepa Fernando et al.

Cardiovascular diseases (CVDs) are the leading cause of mortality worldwide, accounting for over 30% of global deaths according to the World Health Organization (WHO). Importantly, one-third of these deaths are preventable with timely and accurate diagnosis. The electrocardiogram (ECG), a non-invasive method for recording the electrical activity of the heart, is crucial for diagnosing CVDs. However, privacy concerns surrounding the use of patient ECG data in research have spurred interest in synthetic data, which preserves the statistical properties of real data without compromising patient confidentiality. This study explores the potential of synthetic ECG data for training deep learning models from regression to classification tasks and evaluates the feasibility of transfer learning to enhance classification performance on real ECG data. We experimented with popular deep learning models to predict four key cardiac parameters, namely, Heart Rate (HR), PR interval, QT interval, and QRS complex-using separate regression models. Subsequently, we leveraged these regression models for transfer learning to perform 5-class ECG signal classification. Our experiments systematically investigate whether transfer learning from regression to classification is viable, enabling better utilization of diverse open-access and synthetic ECG datasets. Our findings demonstrate that transfer learning from regression to classification improves classification performance, highlighting its potential to maximize the utility of available data and advance deep learning applications in this domain.

Michael A. Riegler, Kristoffer Herland Hellton, Vajira Thambawita et al.

Selecting prior distributions in Bayesian statistics is challenging, resource-intensive, and subjective. We analyze using large-language models (LLMs) to suggest suitable, knowledge-based informative priors. We developed an extensive prompt asking LLMs not only to suggest priors but also to verify and reflect on their choices. We evaluated Claude Opus, Gemini 2.5 Pro, and ChatGPT-4o-mini on two real datasets: heart disease risk and concrete strength. All LLMs correctly identified the direction for all associations (e.g., that heart disease risk is higher for males). The quality of suggested priors was measured by their Kullback-Leibler divergence from the maximum likelihood estimator's distribution. The LLMs suggested both moderately and weakly informative priors. The moderate priors were often overconfident, resulting in distributions misaligned with the data. In our experiments, Claude and Gemini provided better priors than ChatGPT. For weakly informative priors, a key performance difference emerged: ChatGPT and Gemini defaulted to an "unnecessarily vague" mean of 0, while Claude did not, demonstrating a significant advantage. The ability of LLMs to identify correct associations shows their great potential as an efficient, objective method for developing informative priors. However, the primary challenge remains in calibrating the width of these priors to avoid over- and under-confidence.

Chathura Wimalasiri, Piumal Rathnayake, Shamod Wijerathne et al.

Periodontitis, a chronic inflammatory disease causing alveolar bone loss, significantly affects oral health and quality of life. Accurate assessment of bone loss severity and pattern is critical for diagnosis and treatment planning. In this study, we propose a novel AI-based deep learning framework to automatically detect and quantify alveolar bone loss and its patterns using intraoral periapical (IOPA) radiographs. Our method combines YOLOv8 for tooth detection with Keypoint R-CNN models to identify anatomical landmarks, enabling precise calculation of bone loss severity. Additionally, YOLOv8x-seg models segment bone levels and tooth masks to determine bone loss patterns (horizontal vs. angular) via geometric analysis. Evaluated on a large, expertly annotated dataset of 1000 radiographs, our approach achieved high accuracy in detecting bone loss severity (intra-class correlation coefficient up to 0.80) and bone loss pattern classification (accuracy 87%). This automated system offers a rapid, objective, and reproducible tool for periodontal assessment, reducing reliance on subjective manual evaluation. By integrating AI into dental radiographic analysis, our framework has the potential to improve early diagnosis and personalized treatment planning for periodontitis, ultimately enhancing patient care and clinical outcomes.

Pegah Salehi, Sajad Amouei Sheshkal, Vajira Thambawita et al.

Dynamic facial emotion is essential for believable AI-generated avatars, yet most systems remain visually static, limiting their use in simulations like virtual training for investigative interviews with abused children. We present a real-time architecture combining Unreal Engine 5 MetaHuman rendering with NVIDIA Omniverse Audio2Face to generate facial expressions from vocal prosody in photorealistic child avatars. Due to limited TTS options, both avatars were voiced using young adult female models from two systems to better fit character profiles, introducing a voice-age mismatch. This confound may affect audiovisual alignment. We used a two-PC setup to decouple speech generation from GPU-intensive rendering, enabling low-latency interaction in desktop and VR. A between-subjects study (N=70) compared audio+visual vs. visual-only conditions as participants rated emotional clarity, facial realism, and empathy for avatars expressing joy, sadness, and anger. While emotions were generally recognized - especially sadness and joy - anger was harder to detect without audio, highlighting the role of voice in high-arousal expressions. Interestingly, silencing clips improved perceived realism by removing mismatches between voice and animation, especially when tone or age felt incongruent. These results emphasize the importance of audiovisual congruence: mismatched voice undermines expression, while a good match can enhance weaker visuals - posing challenges for emotionally coherent avatars in sensitive contexts.

Matthias Boeker, Vajira Thambawita, Michael Riegler et al.

Understanding sleep and activity patterns plays a crucial role in physical and mental health. This study introduces a novel approach for sleep detection using weakly supervised learning for scenarios where reliable ground truth labels are unavailable. The proposed method relies on a set of weak labels, derived from the predictions generated by conventional sleep detection algorithms. Introducing a novel approach, we suggest a novel generalised non-linear statistical model in which the number of weak sleep labels is modelled as outcome of a binomial distribution. The probability of sleep in the binomial distribution is linked to the outcomes of neural networks trained to detect sleep based on actigraphy. We show that maximizing the likelihood function of the model, is equivalent to minimizing the soft cross-entropy loss. Additionally, we explored the use of the Brier score as a loss function for weak labels. The efficacy of the suggested modelling framework was demonstrated using the Multi-Ethnic Study of Atherosclerosis dataset. A \gls{lstm} trained on the soft cross-entropy outperformed conventional sleep detection algorithms, other neural network architectures and loss functions in accuracy and model calibration. This research not only advances sleep detection techniques in scenarios where ground truth data is scarce but also contributes to the broader field of weakly supervised learning by introducing innovative approach in modelling sets of weak labels.

Sharib Ali, Noha Ghatwary, Debesh Jha et al.

Polyps are well-known cancer precursors identified by colonoscopy. However, variability in their size, location, and surface largely affect identification, localisation, and characterisation. Moreover, colonoscopic surveillance and removal of polyps (referred to as polypectomy ) are highly operator-dependent procedures. There exist a high missed detection rate and incomplete removal of colonic polyps due to their variable nature, the difficulties to delineate the abnormality, the high recurrence rates, and the anatomical topography of the colon. There have been several developments in realising automated methods for both detection and segmentation of these polyps using machine learning. However, the major drawback in most of these methods is their ability to generalise to out-of-sample unseen datasets that come from different centres, modalities and acquisition systems. To test this hypothesis rigorously we curated a multi-centre and multi-population dataset acquired from multiple colonoscopy systems and challenged teams comprising machine learning experts to develop robust automated detection and segmentation methods as part of our crowd-sourcing Endoscopic computer vision challenge (EndoCV) 2021. In this paper, we analyse the detection results of the four top (among seven) teams and the segmentation results of the five top teams (among 16). Our analyses demonstrate that the top-ranking teams concentrated on accuracy (i.e., accuracy > 80% on overall Dice score on different validation sets) over real-time performance required for clinical applicability. We further dissect the methods and provide an experiment-based hypothesis that reveals the need for improved generalisability to tackle diversity present in multi-centre datasets.

Cise Midoglu, Steven A. Hicks, Vajira Thambawita et al.

Soccer has a considerable market share of the global sports industry, and the interest in viewing videos from soccer games continues to grow. In this respect, it is important to provide game summaries and highlights of the main game events. However, annotating and producing events and summaries often require expensive equipment and a lot of tedious, cumbersome, manual labor. Therefore, automating the video production pipeline providing fast game highlights at a much lower cost is seen as the "holy grail". In this context, recent developments in Artificial Intelligence (AI) technology have shown great potential. Still, state-of-the-art approaches are far from being adequate for practical scenarios that have demanding real-time requirements, as well as strict performance criteria (where at least the detection of official events such as goals and cards must be 100% accurate). In addition, event detection should be thoroughly enhanced by annotation and classification, proper clipping, generating short descriptions, selecting appropriate thumbnails for highlight clips, and finally, combining the event highlights into an overall game summary, similar to what is commonly aired during sports news. Even though the event tagging operation has by far received the most attention, an end-to-end video production pipeline also includes various other operations which serve the overall purpose of automated soccer analysis. This challenge aims to assist the automation of such a production pipeline using AI. In particular, we focus on the enhancement operations that take place after an event has been detected, namely event clipping (Task 1), thumbnail selection (Task 2), and game summarization (Task 3). Challenge website: https://mmsys2022.ie/authors/grand-challenge.

Vajira Thambawita, Steven A. Hicks, Pål Halvorsen et al.

Detection of colon polyps has become a trending topic in the intersecting fields of machine learning and gastrointestinal endoscopy. The focus has mainly been on per-frame classification. More recently, polyp segmentation has gained attention in the medical community. Segmentation has the advantage of being more accurate than per-frame classification or object detection as it can show the affected area in greater detail. For our contribution to the EndoCV 2021 segmentation challenge, we propose two separate approaches. First, a segmentation model named TriUNet composed of three separate UNet models. Second, we combine TriUNet with an ensemble of well-known segmentation models, namely UNet++, FPN, DeepLabv3, and DeepLabv3+, into a model called DivergentNets to produce more generalizable medical image segmentation masks. In addition, we propose a modified Dice loss that calculates loss only for a single class when performing multiclass segmentation, forcing the model to focus on what is most important. Overall, the proposed methods achieved the best average scores for each respective round in the challenge, with TriUNet being the winning model in Round I and DivergentNets being the winning model in Round II of the segmentation generalization challenge at EndoCV 2021. The implementation of our approach is made publicly available on GitHub.

Rabindra Khadga, Debesh Jha, Steven Hicks et al.

Widely used traditional supervised deep learning methods require a large number of training samples but often fail to generalize on unseen datasets. Therefore, a more general application of any trained model is quite limited for medical imaging for clinical practice. Using separately trained models for each unique lesion category or a unique patient population will require sufficiently large curated datasets, which is not practical to use in a real-world clinical set-up. Few-shot learning approaches can not only minimize the need for an enormous number of reliable ground truth labels that are labour-intensive and expensive but can also be used to model on a dataset coming from a new population. To this end, we propose to exploit an optimization-based implicit model agnostic meta-learning (iMAML) algorithm under few-shot settings for medical image segmentation. Our approach can leverage the learned weights from diverse but small training samples to perform analysis on unseen datasets with high accuracy. We show that, unlike classical few-shot learning approaches, our method improves generalization capability. To our knowledge, this is the first work that exploits iMAML for medical image segmentation and explores the strength of the model on scenarios such as meta-training on unique and mixed instances of lesion datasets. Our quantitative results on publicly available skin and polyp datasets show that the proposed method outperforms the naive supervised baseline model and two recent few-shot segmentation approaches by large margins. In addition, our iMAML approach shows an improvement of 2%-4% in dice score compared to its counterpart MAML for most experiments.

Vajira Thambawita, Steven Hicks, Pål Halvorsen et al.

Segmentation of findings in the gastrointestinal tract is a challenging but also an important task which is an important building stone for sufficient automatic decision support systems. In this work, we present our solution for the Medico 2020 task, which focused on the problem of colon polyp segmentation. We present our simple but efficient idea of using an augmentation method that uses grids in a pyramid-like manner (large to small) for segmentation. Our results show that the proposed methods work as indented and can also lead to comparable results when competing with other methods.

Vajira Thambawita, Debesh Jha, Hugo Lewi Hammer et al.

Precise and efficient automated identification of Gastrointestinal (GI) tract diseases can help doctors treat more patients and improve the rate of disease detection and identification. Currently, automatic analysis of diseases in the GI tract is a hot topic in both computer science and medical-related journals. Nevertheless, the evaluation of such an automatic analysis is often incomplete or simply wrong. Algorithms are often only tested on small and biased datasets, and cross-dataset evaluations are rarely performed. A clear understanding of evaluation metrics and machine learning models with cross datasets is crucial to bring research in the field to a new quality level. Towards this goal, we present comprehensive evaluations of five distinct machine learning models using Global Features and Deep Neural Networks that can classify 16 different key types of GI tract conditions, including pathological findings, anatomical landmarks, polyp removal conditions, and normal findings from images captured by common GI tract examination instruments. In our evaluation, we introduce performance hexagons using six performance metrics such as recall, precision, specificity, accuracy, F1-score, and Matthews Correlation Coefficient to demonstrate how to determine the real capabilities of models rather than evaluating them shallowly. Furthermore, we perform cross-dataset evaluations using different datasets for training and testing. With these cross-dataset evaluations, we demonstrate the challenge of actually building a generalizable model that could be used across different hospitals. Our experiments clearly show that more sophisticated performance metrics and evaluation methods need to be applied to get reliable models rather than depending on evaluations of the splits of the same dataset, i.e., the performance metrics should always be interpreted together rather than relying on a single metric.

Vajira Thambawita, Pål Halvorsen, Hugo Hammer et al.

In this paper, we present a two-step deep learning method that is used to predict sperm motility and morphology-based on video recordings of human spermatozoa. First, we use an autoencoder to extract temporal features from a given semen video and plot these into image-space, which we call feature-images. Second, these feature-images are used to perform transfer learning to predict the motility and morphology values of human sperm. The presented method shows it's capability to extract temporal information into spatial domain feature-images which can be used with traditional convolutional neural networks. Furthermore, the accuracy of the predicted motility of a given semen sample shows that a deep learning-based model can capture the temporal information of microscopic recordings of human semen.

Vajira Thambawita, Pål Halvorsen, Hugo Hammer et al.

In this paper, we analyse two deep learning methods to predict sperm motility and sperm morphology from sperm videos. We use two different inputs: stacked pure frames of videos and dense optical flows of video frames. To solve this regression task of predicting motility and morphology, stacked dense optical flows and extracted original frames from sperm videos were used with the modified state of the art convolution neural networks. For modifications of the selected models, we have introduced an additional multi-layer perceptron to overcome the problem of over-fitting. The method which had an additional multi-layer perceptron with dropout layers, shows the best results when the inputs consist of both dense optical flows and an original frame of videos.

Steven A. Hicks, Jorunn M. Andersen, Oliwia Witczak et al.

Methods for automatic analysis of clinical data are usually targeted towards a specific modality and do not make use of all relevant data available. In the field of male human reproduction, clinical and biological data are not used to its fullest potential. Manual evaluation of a semen sample using a microscope is time-consuming and requires extensive training. Furthermore, the validity of manual semen analysis has been questioned due to limited reproducibility, and often high inter-personnel variation. The existing computer-aided sperm analyzer systems are not recommended for routine clinical use due to methodological challenges caused by the consistency of the semen sample. Thus, there is a need for an improved methodology. We use modern and classical machine learning techniques together with a dataset consisting of 85 videos of human semen samples and related participant data to automatically predict sperm motility. Used techniques include simple linear regression and more sophisticated methods using convolutional neural networks. Our results indicate that sperm motility prediction based on deep learning using sperm motility videos is rapid to perform and consistent. The algorithms performed worse when participant data was added. In conclusion, machine learning-based automatic analysis may become a valuable tool in male infertility investigation and research.

Vajira Thambawita, Debesh Jha, Michael Riegler et al.

In this paper, we present our approach for the 2018 Medico Task classifying diseases in the gastrointestinal tract. We have proposed a system based on global features and deep neural networks. The best approach combines two neural networks, and the reproducible experimental results signify the efficiency of the proposed model with an accuracy rate of 95.80%, a precision of 95.87%, and an F1-score of 95.80%.