21 research outputs found
Multi-Rater Calibration Error Estimation
No full textPublisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.Calibration, the property of producing predicted probabilities that reflect true likelihoods of outcomes, is a relevant attribute of medical image computing models and a key requirement in clinical decision-making. However, empirical Calibration Error (CE) estimates suffer from instability in data-scarce scenarios. Here, for any existing CE we propose a Multi-Rater version of it (MR-CE), a wrapper over conventional calibration metrics, which provides a new strategy for estimating a CE that effectively addresses this limitation in situations where there are multiple annotations per sample. MR-CEs offer more consistent estimates of calibration errors by leveraging the consensus and disagreement among multiple annotators to generate virtually extended test datasets, more robust to typical binning artifacts. We evaluate a MR version of the popular Expected Calibration Error (ECE), and also of the more recent Kernel Density Estimation-ECE (kdeECE), in a comprehensive set of classification and segmentation problems, demonstrating improved stability compared to their single-rater CE counterparts. Specifically, we show that MR-CEs achieve a reduced variability as the test set size decreases across all analysed datasets. Our findings emphasize the critical role of modelling inter-rater variability not only for training but also for evaluating medical image analysis models, in particular when studying the calibration of modern neural networks.Peer reviewe
Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge
No full textGliomas are the most common primary brain malignancies, with different degrees of aggressiveness, variable prognosis and various heterogeneous histologic sub-regions, i.e., peritumoral edematous/invaded tissue, necrotic core, active and non-enhancing core. This intrinsic heterogeneity is also portrayed in their radio-phenotype, as their sub-regions are depicted by varying intensity profiles disseminated across multi-parametric magnetic resonance imaging (mpMRI) scans, reflecting varying biological properties. Their heterogeneous shape, extent, and location are some of the factors that make these tumors difficult to resect, and in some cases inoperable. The amount of resected tumor is a factor also considered in longitudinal scans, when evaluating the apparent tumor for potential diagnosis of progression. Furthermore, there is mounting evidence that accurate segmentation of the various tumor sub-regions can offer the basis for quantitative image analysis towards predictio n of patient overall survival. This study assesses the state-of-the-art machine learning (ML) methods used for brain tumor image analysis in mpMRI scans, during the last seven instances of the International Brain Tumor Segmentation (BraTS) challenge, i.e., 2012-2018. Specifically, we focus on i) evaluating segmentations of the various glioma sub-regions in pre-operative mpMRI scans, ii) assessing potential tumor progression by virtue of longitudinal growth of tumor sub-regions, beyond use of the RECIST/RANO criteria, and iii) predicting the overall survival from pre-operative mpMRI scans of patients that underwent gross total resection. Finally, we investigate the challenge of identifying the best ML algorithms for each of these tasks, considering that apart from being diverse on each instance of the challenge, the multi-institutional mpMRI BraTS dataset has also been a continuously evolving/growing dataset
Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge
No full textGliomas are the most common primary brain malignancies, with different degrees of aggressiveness, variable prognosis and various heterogeneous histologic sub-regions, i.e., peritumoral edematous/invaded tissue, necrotic core, active and non-enhancing core. This intrinsic heterogeneity is also portrayed in their radio-phenotype, as their sub-regions are depicted by varying intensity profiles disseminated across multi-parametric magnetic resonance imaging (mpMRI) scans, reflecting varying biological properties. Their heterogeneous shape, extent, and location are some of the factors that make these tumors difficult to resect, and in some cases inoperable. The amount of resected tumor is a factor also considered in longitudinal scans, when evaluating the apparent tumor for potential diagnosis of progression. Furthermore, there is mounting evidence that accurate segmentation of the various tumor sub-regions can offer the basis for quantitative image analysis towards predictio n of patient overall survival. This study assesses the state-of-the-art machine learning (ML) methods used for brain tumor image analysis in mpMRI scans, during the last seven instances of the International Brain Tumor Segmentation (BraTS) challenge, i.e., 2012-2018. Specifically, we focus on i) evaluating segmentations of the various glioma sub-regions in pre-operative mpMRI scans, ii) assessing potential tumor progression by virtue of longitudinal growth of tumor sub-regions, beyond use of the RECIST/RANO criteria, and iii) predicting the overall survival from pre-operative mpMRI scans of patients that underwent gross total resection. Finally, we investigate the challenge of identifying the best ML algorithms for each of these tasks, considering that apart from being diverse on each instance of the challenge, the multi-institutional mpMRI BraTS dataset has also been a continuously evolving/growing dataset
Evaluation of Uncertainty-Aware Multi-software Ensembles for Hippocampal Segmentation
No full textAccurate hippocampal segmentation can be a useful tool for diagnosing and monitoring neurological conditions such as Alzheimer’s disease and epilepsy. While numerous automated segmentation methods exist, their clinical adoption remains limited. Reliable uncertainty assessment can enhance trust and facilitate clinical translation. This study evaluates five heterogeneous hippocampal segmentation methods InnerEye, ASHS, FastSurfer, HippoSeg, and FreeSurfer—across two dementia datasets and one epilepsy dataset. The sub-ensemble containing InnerEye, FastSurfer, and HippoSeg emerged as both accurate and efficient, highlighting the feasibility of balancing computational cost and performance. Additionally, ensemble-derived uncertainty quantification with sample variance, mutual information, and predictive entropy is shown to reduce inaccurate segmentations by flagging low-confidence cases, potentially providing a mechanism for automatically escalating ambiguous cases for expert assessment
Ecos pirandellianos en el pensamiento de Unamuno
No full textIn this paper I focus on the influence of Luigi Pirandello on Miguel de Unamuno. The first similarities between these two writers appear on a Pirandellian short story titled A character’s tragedy and Unamuno's novel, Niebla. Furthermore, I analyze two dramas of the Spanish author, El hermano Juan and El otro and I remark on its commonalities with the philosophy of Luigi Pirandello. Finally, I realize that the Pirandello’s topics such as madness versus sanity, to be versus to seem, the coexistence of reality and fiction and the art of play within a play appear in both texts of the Spanish author unanimously and consistently.En este trabajo nos centramos en la influencia de Luigi Pirandello sobre Miguel de Unamuno. Y pretendemos analizar las semejanzas entre el cuento pirandelliano titulado La tragedia de un personaje y la novela de Unamuno, Niebla. Además, comentamos las analogías conceptuales de dos dramas del autor español, es decir, El otro y El hermano Juan, con la filosofía de Luigi Pirandello. Finalmente y a modo de conclusión, nos damos cuenta de que los tópicos pirandellianos como la locura frente a la cordura, el ser versus el parecer, la coexistencia entre la realidad y la ficción y la técnica del teatro dentro del teatro aparecen en ambos textos del autor español de forma unánime y coherente
Socio-economic factors explain differences in public health-related variables among women in Bangladesh: A cross-sectional study
No full textKhan MH, Krämer A. Socio-economic factors explain differences in public health-related variables among women in Bangladesh: A cross-sectional study. BMC Public Health. 2008;8(1): 254.Background: Worldwide one billion people are living in slum communities and experts projected that this number would double by 2030. Slum populations, which are increasing at an alarming rate in Bangladesh mainly due to rural-urban migration, are often neglected and characterized by poverty, poor housing, overcrowding, poor environment, and high prevalence of communicable diseases. Unfortunately, comparisons between women living in slums and those not living in slums are very limited in Bangladesh. The objectives of the study were to examine the association of living in slums (dichotomized as slum versus non-slum) with selected public health-related variables among women, first without adjusting for the influence of other factors and then in the presence of socio-economic variables. Methods: Secondary data was used in this study. 120 women living in slums (as cases) and 480 age-matched women living in other areas (as controls) were extracted from the Bangladesh Demographic and Health Survey 2004. Many socio-economic and demographic variables were analysed. SPSS was used to perform simple as well as multiple analyses. P-values based on t-test and Wald test were also reported to show the significance level. Results: Unadjusted results indicated that a significantly higher percent of women living in slums came from country side, had a poorer status by household characteristics, had less access to mass media, and had less education than women not living in slums. Mean BMI, knowledge of AIDS indicated by ever heard about AIDS, knowledge of avoiding AIDS by condom use, receiving adequate antenatal visits (4 or more) during the last pregnancy, and safe delivery practices assisted by skilled sources were significantly lower among women living in slums than those women living in other areas. However, all the unadjusted significant associations with the variable slum were greatly attenuated and became insignificant (expect safe delivery practices) when some socio-economic variables namely childhood place of residence, a composite variable of household characteristics, a composite variable of mass media access, and education were inserted into the multiple regression models. Taken together, childhood place of residence, the composite variable of mass media access, and education were the strongest predictors for the health related outcomes. Conclusion: Reporting unadjusted findings of public health variables in women from slums versus non-slums can be misleading due to confounding factors. Our findings suggest that an association of childhood place of residence, mass media access and public health education should be considered before making any inference based on slum versus non-slum comparisons
Why is unemployment so high in Bulgaria?
No full textThe author seeks to determine the main factors behind poor labor market outcomes in Bulgaria. Unemployment in Bulgaria is high and of long duration. The accumulation of the unemployment stock has been caused by relatively high inflows into unemployment coupled with limited outflows. These features of the Bulgarian labor market are typical of other transition economies in Central Europe and exploring their sources is of broad interest. The author focuses on determinants of and constraints to job creation. He uses data on job creation and job destruction from a survey of employment in all registered firms. He finds that the source of large inflows into unemployment is intensive enterprise restructuring associated with a high pace of job reallocation. However, job creation falls short of job destruction. Three main factors account for the limited job creation and hiring, and thus for low outflows from unemployment: a) The unfriendly business environment, reflected by a low rate of new firm formation, and a relatively small, small and medium enterprise sector. b) Labor market rigidities, including excessive hiring and firing costs. c) Skill and spatial mismatches brought about by enterprise restructuring, as well as low skills and marginalization of the long-term unemployed who cannot successfully compete for new jobs. The author recommends a three pronged strategy to improve labor market performance: (1) removing bureaucratic constraints to entry and expansion of firms; (2) enhancing labor market flexibility through lowering hiring and firing costs; and (3) improving the educational system so as to equip workers with broad and portable skills.Environmental Economics&Policies,Labor Policies,Labor Markets,Public Health Promotion,Health Monitoring&Evaluation,Environmental Economics&Policies,Labor Markets,Health Monitoring&Evaluation,Labor Standards,Banks&Banking Reform
A mile in their shoes: understanding healthcare journeys of refugees and asylum seekers in the UK
No full textThis author accepted manuscript is deposited under a Creative Commons Attribution Non-commercial 4.0 International (CC BY-NC) licence. This means that anyone may distribute, adapt, and build upon the work for non-commercial purposes, subject to full attribution. If you wish to use this manuscript for commercial purposes, please contact [email protected].
This article is published in its final form in the International Journal of Migration, Health and Social Care as: Talks, I., Al Mobarak, B., Katona, C., Hunt, J., Winters, N., & Geniets, A. (2024). A mile in their shoes: understanding health-care journeys of refugees and asylum seekers in the UK. International Journal of Migration, Health and Social Care. https://doi.org/10.1108/IJMHSC-06-2023-0060Purpose
Refugees and asylum seekers worldwide face numerous barriers in accessing health systems. The evidence base regarding who and what helps refugees and asylum seekers facilitate access to and the navigation of the health system in the UK is small. This study aims to address this gap by analysing 14 semi-structured, in-depth interviews with refugees and asylum seekers of different countries of origin in the UK to identify where, when and how they came into contact with the health-care system and what the outcome of these interactions was.
Design/methodology/approach
Semi-structured, in-depth interviews were chosen as the key method for this study. In total, 14 individual interviews were conducted. A trauma-informed research approach was applied to reduce the risk of re-traumatising participants.
Findings
The paper identifies key obstacles as well as “facilitators” of refugees’ and asylum seekers’ health-care experience in the UK and suggests that host families, friends and third-party organisations all play an important role in ensuring refugees and asylum seekers receive the healthcare they need.
Originality/value
To the best of the authors’ knowledge, this is the first qualitative study in the UK that looks at comprehensive health journeys of refugees from their first encounter with health services through to secondary care, highlighting the important role along the way of facilitators such as host families, friends and third-party organisations.This project was funded by the John Fell Fun
Controllable illumination invariant GAN for diverse temporally-consistent surgical video synthesis
No full textSurgical video synthesis offers a cost-effective way to expand training data and enhance the performance of machine learning models in computer-assisted surgery. However, existing video translation methods often produce video sequences with large illumination changes across different views, disrupting the temporal consistency of the videos. Additionally, these methods typically synthesize videos with a monotonous style, whereas diverse synthetic data is desired to improve the generalization ability of downstream machine learning models. To address these challenges, we propose a novel Controllable Illumination Invariant Generative Adversarial Network (CIIGAN) for generating diverse, illumination-consistent video sequences. CIIGAN fuses multi-scale illumination-invariant features from a novel controllable illumination-invariant (CII) image space with multi-scale texture-invariant features from self-constructed 3D scenes. The CII image space, along with the 3D scenes, allows CIIGAN to produce diverse and temporally-consistent video or image translations. Extensive experiments demonstrate that CIIGAN achieves more realistic and illumination-consistent translations compared to previous state-of-the-art baselines. Furthermore, the segmentation networks trained on our diverse synthetic data outperform those trained on monotonous synthetic data
SurgicalGS: dynamic 3D Gaussian splatting for accurate robotic-assisted surgical scene reconstruction
No full textAccurate 3D reconstruction of dynamic surgical scenes from endoscopic video is essential for robotic-assisted surgery. While recent 3D Gaussian Splatting methods have shown promise in achieving high-quality reconstructions with fast rendering speeds, their use of inverse depth loss functions compresses depth variations. This can lead to a loss of fine geometric details, limiting their ability to capture precise 3D geometry and effectiveness in intraoperative applications. To address the limitations of existing methods, we developed SurgicalGS, a dynamic 3D Gaussian Splatting framework specifically designed for improved geometric accuracy in surgical scene reconstruction. Our approach integrates a temporally coherent multi-frame depth fusion and an adaptive motion mask for Gaussian initialisation. Besides, we represent dynamic scenes using the Flexible Deformation Model and introduce a novel normalized depth regularization loss and an unsupervised depth smoothness constraint to ensure high geometric accuracy in the reconstruction. Extensive experiments on two real surgical datasets demonstrate that SurgicalGS achieves state-of-the-art reconstruction quality, especially in precise geometry, advancing the usability of 3D Gaussian Splatting in robotic-assisted surgery. Our code is available at https://github.com/neneyork/SurgicalGS
