arXiv subjects

Richard Brown

Publications and source records attributed to Richard Brown.

8 recordsLinked to original sources

arXiv2023

DeepEdit: Deep Editable Learning for Interactive Segmentation of 3D Medical Images

Automatic segmentation of medical images is a key step for diagnostic and interventional tasks. However, achieving this requires large amounts of annotated volumes, which can be tedious and time-consuming task for expert annotators. In this paper, we introduce DeepEdit, a deep learning-based method for volumetric medical image annotation, that allows automatic and semi-automatic segmentation, and click-based refinement. DeepEdit combines the power of two methods: a non-interactive (i.e. automatic segmentation using nnU-Net, UNET or UNETR) and an interactive segmentation method (i.e. DeepGrow), into a single deep learning model. It allows easy integration of uncertainty-based ranking strategies (i.e. aleatoric and epistemic uncertainty computation) and active learning. We propose and implement a method for training DeepEdit by using standard training combined with user interaction simulation. Once trained, DeepEdit allows clinicians to quickly segment their datasets by using the algorithm in auto segmentation mode or by providing clicks via a user interface (i.e. 3D Slicer, OHIF). We show the value of DeepEdit through evaluation on the PROSTATEx dataset for prostate/prostatic lesions and the Multi-Atlas Labeling Beyond the Cranial Vault (BTCV) dataset for abdominal CT segmentation, using state-of-the-art network architectures as baseline for comparison. DeepEdit could reduce the time and effort annotating 3D medical images compared to DeepGrow alone. Source code is available at https://github.com/Project-MONAI/MONAILabel

eess.IV↗

arXiv2022

MONAI: An open-source framework for deep learning in healthcare

Artificial Intelligence (AI) is having a tremendous impact across most areas of science. Applications of AI in healthcare have the potential to improve our ability to detect, diagnose, prognose, and intervene on human disease. For AI models to be used clinically, they need to be made safe, reproducible and robust, and the underlying software framework must be aware of the particularities (e.g. geometry, physiology, physics) of medical data being processed. This work introduces MONAI, a freely available, community-supported, and consortium-led PyTorch-based framework for deep learning in healthcare. MONAI extends PyTorch to support medical data, with a particular focus on imaging, and provide purpose-specific AI model architectures, transformations and utilities that streamline the development and deployment of medical AI models. MONAI follows best practices for software-development, providing an easy-to-use, robust, well-documented, and well-tested software framework. MONAI preserves the simple, additive, and compositional approach of its underlying PyTorch libraries. MONAI is being used by and receiving contributions from research, clinical and industrial teams from around the world, who are pursuing applications spanning nearly every aspect of healthcare.

cs.LG↗

arXiv2021

Metadata-Based Detection of Child Sexual Abuse Material

Child Sexual Abuse Media (CSAM) is any visual record of a sexually-explicit activity involving minors. CSAM impacts victims differently from the actual abuse because the distribution never ends, and images are permanent. Machine learning-based solutions can help law enforcement quickly identify CSAM and block digital distribution. However, collecting CSAM imagery to train machine learning models has many ethical and legal constraints, creating a barrier to research development. With such restrictions in place, the development of CSAM machine learning detection systems based on file metadata uncovers several opportunities. Metadata is not a record of a crime, and it does not have legal restrictions. Therefore, investing in detection systems based on metadata can increase the rate of discovery of CSAM and help thousands of victims. We propose a framework for training and evaluating deployment-ready machine learning models for CSAM identification. Our framework provides guidelines to evaluate CSAM detection models against intelligent adversaries and models' performance with open data. We apply the proposed framework to the problem of CSAM detection based on file paths. In our experiments, the best-performing model is based on convolutional neural networks and achieves an accuracy of 0.97. Our evaluation shows that the CNN model is robust against offenders actively trying to evade detection by evaluating the model against adversarially modified data. Experiments with open datasets confirm that the model generalizes well and is deployment-ready.

cs.LG↗

arXiv2021

A Novel Multimodal Biometric Authentication System using Machine Learning and Blockchain

Traditional authentication systems that rely on simple passwords, PIN numbers or tokens have many security issues, like easily guessed passwords, PIN numbers written on the back of cards, etc. Thus, biometric authentication methods that rely on physical and behavioural characteristics have been proposed as an alternative for those systems. In real-world applications, authentication systems that involve a single biometric faced many issues, especially lack of accuracy and noisy data, which boost the research community to create multibiometric systems that involve a variety of biometrics. Those systems provide better performance and higher accuracy compared to other authentication methods. However, most of them are inconvenient and requires complex interactions from the user. Thus, in this paper, we introduce a novel multimodal authentication system that relies on machine learning and blockchain, with the aim of providing a more secure, transparent, and convenient authentication mechanism. The proposed system combines four important biometrics, fingerprint, face, age, and gender. The supervised learning algorithm Decision Tree has been used to combine the results of the biometrics verification process and produce a confidence level related to the user. The initial experimental results show the efficiency and robustness of the proposed multimodal systems.

cs.CR↗

arXiv2021

Tracking areas with increased likelihood of surface particle aggregation in the Gulf of Finland: A first look at persistent Lagrangian Coherent Structures (LCS)

We explore the possibility to identify areas of intense patch formation from floating items due to systematic convergence of surface velocity fields by means of a visual comparison of Lagrangian Coherent Structures (LCS) and estimates of areas prone to patch formation using the concept of Finite-Time Compressibility (FTC, a generalisation of the notion of time series of divergence). The LCSs are evaluated using the Finite Time Lyapunov Exponent (FTLE) method. The test area is the Gulf of Finland (GoF) in the Baltic Sea. A basin-wide spatial average of backward FTLE is calculated for the GoF for the first time. This measure of the mixing strength displays a clear seasonal pattern. The evaluated backward FTLE features are linked with potential patch formation regions with high FTC levels. It is shown that areas hosting frequent upwelling or downwelling have consistently stronger than average mixing intensity. The combination of both methods, FTC and LCS, has the potential of being a powerful tool to identify the formation of patches of pollution at the sea surface.

physics.ao-ph↗

arXiv2015

Bayesian data assimilation provides rapid decision support for vector-borne diseases

Predicting the spread of vector-borne diseases in response to incursions requires knowledge of both host and vector demographics in advance of an outbreak. Whereas host population data is typically available, for novel disease introductions there is a high chance of the pathogen utilising a vector for which data is unavailable. This presents a barrier to estimating the parameters of dynamical models representing host-vector-pathogen interaction, and hence limits their ability to provide quantitative risk forecasts. The Theileria orientalis (Ikeda) outbreak in New Zealand cattle demonstrates this problem: even though the vector has received extensive laboratory study, a high degree of uncertainty persists over its national demographic distribution. Addressing this, we develop a Bayesian data assimilation approach whereby indirect observations of vector activity inform a seasonal spatio-temporal risk surface within a stochastic epidemic model. We provide quantitative predictions for the future spread of the epidemic, quantifying uncertainty in the model parameters, case infection times, and the disease status of undetected infections. Importantly, we demonstrate how our model learns sequentially as the epidemic unfolds, and provides evidence for changing epidemic dynamics through time. Our approach therefore provides a significant advance in rapid decision support for novel vector-borne disease outbreaks.

q-bio.PE↗

arXiv2003

Automorphisms of the Fricke characters of groups

In this note, we embed the set of all Fricke characters of a free group F -- the set of all characters of representations of F into SL(2,C) -- as an irreducible affine variety V in complex affine space of dimension 2^n-1. Using the Horowitz generating set as the indeterminates, we show that the ideal I of all polynomials in these indeterminates which vanish on V is finitely generated by the Magnus relation for arbitrary octets of elements in SL(2,C). Using this relation, we produce a basis for I, and show that it is prime. We then show that the natural action of automorphisms of F on V extends to polynomial automorphisms on all of the ambient affine space which, up to sign, preserve a complex volume form. This construction provides an algebraic model for the analysis of the dynamics of the measure preserving action of Out(F) on V.

math.DG↗

arXiv1998

Dependence of the Fundamental Plane Scatter on Galaxy Age

The fundamental plane (FP) has an intrinsic scatter that can not be explained purely by observational errors. Using recently available age estimates for nearby early type galaxies, we show that a galaxy's position relative to the FP depends on its age. In particular, the mean FP corresponds to ellipticals with an age of ~10 Gyr. Younger galaxies are systematically brighter with higher surface brightness relative to the mean relation. Old ellipticals form an `upper envelope' to the FP. For our sample of mostly non-cluster galaxies, age can account for almost half of the scatter in the B band FP. Distance determinations based on the FP may have a systematic bias, if the mean age of the sample varies with redshift. We also show that fundamental plane residuals, B-V colors and Mg_2 line strength are consistent with an ageing central burst superposed on an old stellar population. This reinforces the view that these age estimates are tracing the last major episode of star formation induced by a gaseous merger event. We briefly discuss the empirical `evolutionary tracks' of merger-remnants and young ellipticals in terms of their key observational parameters.

astro-ph↗