Search

Accurate lung nodule detection for computed tomography (CT) scan imagery is challenging in real-world settings due to the sparse occurrence of nodules and similarity to other anatomical structures. In a typical positive case, nodules may appear in as few as 3% of CT slices, complicating detection. To address this, we reframe the problem as an anomaly detection task, targeting rare nodule occurrences in a predominantly normal dataset. We introduce a novel solution leveraging custom data preprocessing and Deformable Detection Transformer (Deformable- DETR). A 7.5mm Maximum Intensity Projection (MIP) is utilized to combine adjacent lung slices into single images, reducing the slice count and decreasing nodule sparsity. This enhances spatial context, allowing for better differentiation between nodules and other structures such as complex vascular structures and bronchioles. Deformable-DETR is employed to detect nodules, with a custom focal loss function to better handle the imbalanced dataset. Our model achieves state-of-the-art performance on the LUNA16 dataset with an F1 score of 94.2% (95.2% recall, 93.3% precision) on a dataset sparsely populated with lung nodules that is reflective of real-world clinical data.

We present Polynomial Attention Drop-in Replacement (PADRe), a novel and unifying framework designed to replace the conventional self-attention mechanism in transformer models. Notably, several recent alternative attention mechanisms, including Hyena, Mamba, SimA, Conv2Former, and Castling-ViT, can be viewed as specific instances of our PADRe framework. PADRe leverages polynomial functions and draws upon established results from approximation theory, enhancing computational efficiency without compromising accuracy. PADRe's key components include multiplicative nonlinearities, which we implement using straightforward, hardware-friendly operations such as Hadamard products, incurring only linear computational and memory costs. PADRe further avoids the need for using complex functions such as Softmax, yet it maintains comparable or superior accuracy compared to traditional self-attention. We assess the effectiveness of PADRe as a drop-in replacement for self-attention across diverse computer vision tasks. These tasks include image classification, image-based 2D object detection, and 3D point cloud object detection. Empirical results demonstrate that PADRe runs significantly faster than the conventional self-attention (11x ~ 43x faster on server GPU and mobile NPU) while maintaining similar accuracy when substituting self-attention in the transformer models.

The Circle of Willis (CoW) is an important network of arteries connecting major circulations of the brain. Its vascular architecture is believed to affect the risk, severity, and clinical outcome of serious neuro-vascular diseases. However, characterizing the highly variable CoW anatomy is still a manual and time-consuming expert task. The CoW is usually imaged by two angiographic imaging modalities, magnetic resonance angiography (MRA) and computed tomography angiography (CTA), but there exist limited public datasets with annotations on CoW anatomy, especially for CTA. Therefore we organized the TopCoW Challenge in 2023 with the release of an annotated CoW dataset. The TopCoW dataset was the first public dataset with voxel-level annotations for thirteen possible CoW vessel components, enabled by virtual-reality (VR) technology. It was also the first large dataset with paired MRA and CTA from the same patients. TopCoW challenge formalized the CoW characterization problem as a multiclass anatomical segmentation task with an emphasis on topological metrics. We invited submissions worldwide for the CoW segmentation task, which attracted over 140 registered participants from four continents. The top performing teams managed to segment many CoW components to Dice scores around 90%, but with lower scores for communicating arteries and rare variants. There were also topological mistakes for predictions with high Dice scores. Additional topological analysis revealed further areas for improvement in detecting certain CoW components and matching CoW variant topology accurately. TopCoW represented a first attempt at benchmarking the CoW anatomical segmentation task for MRA and CTA, both morphologically and topologically., Comment: 24 pages, 11 figures, 9 tables. Summary Paper for the MICCAI TopCoW 2023 Challenge

We present an efficient framework for solving algebraically-constrained global non-convex polynomial optimization problems over subsets of the hypercube. We prove the existence of an equivalent nonlinear reformulation of such problems that possesses essentially no spurious local minima. Through numerical experiments on previously intractable global constrained polynomial optimization problems in high dimension, we show that polynomial scaling in dimension and degree is achievable when computing the optimal value and location.

We present the Fast Chebyshev Transform (FCT), a fast, randomized algorithm to compute a Chebyshev approximation of functions in high-dimensions from the knowledge of the location of its nonzero Chebyshev coefficients. Rather than sampling a full-resolution Chebyshev grid in each dimension, we randomly sample several grids with varied resolutions and solve a least-squares problem in coefficient space in order to compute a polynomial approximating the function of interest across all grids simultaneously. We theoretically and empirically show that the FCT exhibits quasi-linear scaling and high numerical accuracy on challenging and complex high-dimensional problems. We demonstrate the effectiveness of our approach compared to alternative Chebyshev approximation schemes. In particular, we highlight our algorithm's effectiveness in high dimensions, demonstrating significant speedups over commonly-used alternative techniques.

Present and next generation of long-baseline accelerator experiments are bringing the measurement of neutrino oscillations into the precision era with ever-increasing statistics. One of the most challenging aspects of achieving such measurements is developing relevant systematic uncertainties in the modeling of nuclear effects in neutrino-nucleus interactions. To address this problem, state-of-the-art detectors are being developed to extract detailed information about all particles produced in neutrino interactions. To fully profit from these experimental advancements, it is essential to have reliable models of propagation of the outgoing hadrons through nuclear matter able to predict how the energy is distributed between all the final-state observed particles. In this article, we investigate the role of nuclear de-excitation in neutrino-nucleus scattering using two Monte Carlo cascade models: NuWro and INCL coupled with the de-excitation code ABLA. The ablation model ABLA is used here for the first time to model de-excitation in neutrino interactions. As input to ABLA, we develop a consistent simulation of nuclear excitation energy tuned to electron-scattering data. The paper includes the characterization of the leading proton kinematics and of the nuclear cluster production during cascade and de-excitation. The observability of nuclear clusters as vertex activity and their role in a precise neutrino energy reconstruction is quantified., Comment: 14 pages, 13 figures

We present a novel efficient theoretical and numerical framework for solving global non-convex polynomial optimization problems. We analytically demonstrate that such problems can be efficiently reformulated using a non-linear objective over a convex set; further, these reformulated problems possess no spurious local minima (i.e., every local minimum is a global minimum). We introduce an algorithm for solving these resulting problems using the augmented Lagrangian and the method of Burer and Monteiro. We show through numerical experiments that polynomial scaling in dimension and degree is achievable for computing the optimal value and location of previously intractable global polynomial optimization problems in high dimension.

Positional and structural encodings (PSE) enable better identifiability of nodes within a graph, rendering them essential tools for empowering modern GNNs, and in particular graph Transformers. However, designing PSEs that work optimally for all graph prediction tasks is a challenging and unsolved problem. Here, we present the Graph Positional and Structural Encoder (GPSE), the first-ever graph encoder designed to capture rich PSE representations for augmenting any GNN. GPSE learns an efficient common latent representation for multiple PSEs, and is highly transferable: The encoder trained on a particular graph dataset can be used effectively on datasets drawn from markedly different distributions and modalities. We show that across a wide range of benchmarks, GPSE-enhanced models can significantly outperform those that employ explicitly computed PSEs, and at least match their performance in others. Our results pave the way for the development of foundational pre-trained graph encoders for extracting positional and structural information, and highlight their potential as a more powerful and efficient alternative to explicitly computed PSEs and existing self-supervised pre-training approaches. Our framework and pre-trained models are publicly available at https://github.com/G-Taxonomy-Workgroup/GPSE. For convenience, GPSE has also been integrated into the PyG library to facilitate downstream applications., Comment: Accepted at ICML 2024; 34 pages, 6 figures

The translation of AI-generated brain metastases (BM) segmentation into clinical practice relies heavily on diverse, high-quality annotated medical imaging datasets. The BraTS-METS 2023 challenge has gained momentum for testing and benchmarking algorithms using rigorously annotated internationally compiled real-world datasets. This study presents the results of the segmentation challenge and characterizes the challenging cases that impacted the performance of the winning algorithms. Untreated brain metastases on standard anatomic MRI sequences (T1, T2, FLAIR, T1PG) from eight contributed international datasets were annotated in stepwise method: published UNET algorithms, student, neuroradiologist, final approver neuroradiologist. Segmentations were ranked based on lesion-wise Dice and Hausdorff distance (HD95) scores. False positives (FP) and false negatives (FN) were rigorously penalized, receiving a score of 0 for Dice and a fixed penalty of 374 for HD95. Eight datasets comprising 1303 studies were annotated, with 402 studies (3076 lesions) released on Synapse as publicly available datasets to challenge competitors. Additionally, 31 studies (139 lesions) were held out for validation, and 59 studies (218 lesions) were used for testing. Segmentation accuracy was measured as rank across subjects, with the winning team achieving a LesionWise mean score of 7.9. Common errors among the leading teams included false negatives for small lesions and misregistration of masks in space.The BraTS-METS 2023 challenge successfully curated well-annotated, diverse datasets and identified common errors, facilitating the translation of BM segmentation across varied clinical environments and providing personalized volumetric reports to patients undergoing BM treatment.

The emission of $\gamma$-rays after a neutron capture in a cryogenic detector can generate mono-energetic nuclear recoils in the sub-keV regime, of direct interest for the calibration of Dark Matter and Coherent Elastic Neutrino Nucleus Scattering experiments. Here we show that accurate predictions of the nuclear recoil spectra induced by neutron captures require taking into account the interplay between the development in time of the de-excitation $\gamma$-cascade of the target nucleus and that of the associated atomic collisions in matter. We present detailed simulations coupling the FIFRELIN code for the description of the $\gamma$-cascades and the IRADINA code for the modelling of the fast atomic movements in matter. Nuclear recoil spectra are predicted, and made available to the community, for concrete cases of Al$_2$O$_3$, Si, Ge and CaWO$_4$ crystals exposed to a low intensity beam of thermal neutrons. We find that timing effects cause new calibration peaks to emerge in the recoil spectra and also impact the shape of the continuous recoil distribution. We discuss how they could give access to a rich physics program, spanning the accurate study of the response of cryogenic detectors in the sub-keV range, tests of solid state physics simulations and tests of nuclear models.

The summation method for the calculation of reactor $\bar{\nu}_e$ fluxes and spectra is methodically revised and improved. For the first time, a complete uncertainty budget accounting for all known effects likely to impact these calculations is proposed. Uncertainties of a few percents at low energies and ranging up to 20% at high energies are obtained on the calculation of a typical reactor $\bar{\nu}_e$ spectrum. Although huge improvements have been achieved over the past decade, the quality and incompleteness of the present day evaluated nuclear decay data still limit the accuracy of the calculations and therefore dominate by far these uncertainties. Pushing the $\beta$-decay modeling of the thousands of branches making a reactor $\bar{\nu}_e$ spectrum to a high level of details comparatively brings modest changes. In particular, including nuclear structure calculations in the evaluation of the non-unique forbidden transitions gives a smaller impact than anticipated in past studies. Finally, this new modeling is challenged against state-of-the-art predictions and measurements. While a good agreement is observed with the most recent Inverse Beta Decay measurements of reactor $\bar{\nu}_e$ fluxes and spectra, it is unable to properly describe the reference aggregate $\beta$ spectra measured at the Institut Laue-Langevin High-Flux reactor in the 80s. This result adds to recent suspicions $\beta$ the reliability of these data and preferentially points toward a misprediction of the $^{235}$U $\bar{\nu}_e$ spectrum., Comment: Reformatted figures. Added supplementary materials

This paper applies a quantum machine learning technique to predict mobile users' trajectories in mobile wireless networks using an approach called quantum reservoir computing (QRC). Mobile users' trajectories prediction belongs to the task of temporal information processing and it is a mobility management problem that is essential for self-organizing and autonomous 6G networks. Our aim is to accurately predict the future positions of mobile users in wireless networks using QRC. To do so, we use a real-world time series dataset to model mobile users' trajectories. The QRC approach has two components: reservoir computing (RC) and quantum computing (QC). In RC, the training is more computational-efficient than the training of simple recurrent neural networks (RNN) since, in RC, only the weights of the output layer are trainable. The internal part of RC is what is called the reservoir. For the RC to perform well, the weights of the reservoir should be chosen carefully to create highly complex and nonlinear dynamics. The QC is used to create such dynamical reservoir that maps the input time series into higher dimensional computational space composed of dynamical states. After obtaining the high-dimensional dynamical states, a simple linear regression is performed to train the output weights and thus the prediction of the mobile users' trajectories can be performed efficiently. In this paper, we apply a QRC approach based on the Hamiltonian time evolution of a quantum system. We simulate the time evolution using IBM gate-based quantum computers and we show in the experimental results that the use of QRC to predict the mobile users' trajectories with only a few qubits is efficient and is able to outperform the classical approaches such as the long short-term memory (LSTM) approach and the echo-state networks (ESN) approach., Comment: 10 pages, 12 figures, 1 table. This paper is a preprint of a paper submitted to IET Quantum Communication. If accepted, the copy of record will be available at the IET Digital Library

Coherent elastic neutrino-nucleus scattering and low-mass Dark Matter detectors rely crucially on the understanding of their response to nuclear recoils. We report the first observation of a nuclear recoil peak at around 112 eV induced by neutron capture. The measurement was performed with a CaWO$_4$ cryogenic detector from the NUCLEUS experiment exposed to a $^{252}$Cf source placed in a compact moderator. The measured spectrum is found in agreement with simulations and the expected peak structure from the single-$\gamma$ de-excitation of $^{183}$W is identified with 3 $\sigma$ significance. This result demonstrates a new method for precise, in-situ, and non-intrusive calibration of low-threshold experiments.

Anomalies in past neutrino measurements have led to the discovery that these particles have non-zero mass and oscillate between their three flavors when they propagate. In the 2010's, similar anomalies observed in the antineutrino spectra emitted by nuclear reactors have triggered the hypothesis of the existence of a supplementary neutrino state that would be sterile i.e. not interacting via the weak interaction. The STEREO experiment was designed to study this scientific case that would potentially extend the Standard Model of Particle Physics. Here we present a complete study based on our full set of data with significantly improved sensitivity. Installed at the ILL (Institut Laue Langevin) research reactor, STEREO has accurately measured the antineutrino energy spectrum associated to the fission of 235U. This measurement confirms the anomalies whereas, thanks to the segmentation of the STEREO detector and its very short mean distance to the core (10~m), the same data reject the hypothesis of a light sterile neutrino. Such a direct measurement of the antineutrino energy spectrum suggests instead that biases in the nuclear experimental data used for the predictions are at the origin of the anomalies. Our result supports the neutrino content of the Standard Model and establishes a new reference for the 235U antineutrino energy spectrum. We anticipate that this result will allow to progress towards finer tests of the fundamental properties of neutrinos but also to benchmark models and nuclear data of interest for reactor physics and for observations of astrophysical or geo-neutrinos., Comment: 34 pages, 13 figures. Final version submitted to Nature

The precise modeling of the de-excitation of Gd isotopes is of great interest for experimental studies of neutrinos using Gd-loaded organic liquid scintillators. The FIFRELIN code was recently used within the purposes of the STEREO experiment for the modeling of the Gd de-excitation after neutron capture in order to achieve a good control of the detection efficiency. In this work, we report on the recent additions in the FIFRELIN de-excitation model with the purpose of enhancing further the de-excitation description. Experimental transition intensities from EGAF database are now included in the FIFRELIN cascades, in order to improve the description of the higher energy part of the spectrum. Furthermore, the angular correlations between {\gamma} rays are now implemented in FIFRELIN, to account for the relative anisotropies between them. In addition, conversion electrons are now treated more precisely in the whole spectrum range, while the subsequent emission of X rays is also accounted for. The impact of the aforementioned improvements in FIFRELIN is tested by simulating neutron captures in various positions inside the STEREO detector. A repository of up-to-date FIFRELIN simulations of the Gd isotopes is made available for the community, with the possibility of expanding for other isotopes which can be suitable for different applications., Comment: Corrected typos on author names on arXiv metadata

We investigate the possible origins of the norm and shape reactor antineutrino anomalies in the framework of a summation model (SM) where $\beta^{-}$ transitions are simulated by a phenomenological Gamow-Teller $\beta$-decay strength model. The general trends of the discrepancies to the Huber-Mueller model on the antineutrino side can be reproduced both in norm and shape. From the exact electron-antineutrino correspondence of the SM model, we predict similar distortions in the electron spectra, suggesting that biases on the reference fission-electron spectra could be at the origin of the anomalies.

It has been postulated that the delivery of radiotherapy at ultra-high dose rates ("FLASH") reduces normal tissue toxicities by depleting them of oxygen. The fraction of normal tissue and cancer cells surviving radiotherapy depends on dose and oxygen levels in an exponential manner and even a very small fraction of tissue at low oxygen levels can determine radiotherapy response. The effect of FLASH on radiation-induced normal and tumour tissue cell killing was studied by simulating oxygen diffusion, metabolism, and radiolytic oxygen depletion over domains with simulated capillary architectures. Two architectural models were used: 1.) randomly distributed capillaries and 2.) capillaries forming a regular square lattice array. The resulting oxygen partial pressure distribution histograms were used to simulate normal and tumour tissue cell survival using the linear quadratic model of cell survival, modified to incorporate oxygen-enhancement ratio (OER) effects. Tumour cell survival was found to be increased by FLASH as compared to conventional radiotherapy, with a 0-1 order of magnitude increase for expected levels of tumour hypoxia, depending on the relative magnitudes of radiolytic oxygen depletion and tissue oxygen metabolism. Interestingly, for the random capillary model, the impact of FLASH on well-oxygenated (normal) tissues was found to be much greater, with an estimated increase in cell survival by up to 10 orders of magnitude, even though reductions in mean tissue partial pressure were modest, less than 7 mmHg for the parameter values studied. The presence of very small nearly hypoxic regions in otherwise well-perfused normal tissues with high mean oxygen levels resulted in a greater proportional sparing of normal tissue than tumour cells during FLASH irradiation, possibly explaining empirical normal tissue sparing and iso-tumour control results., Comment: Published version

The modeling of neutrino-nucleus interactions constitutes a challenging source of systematic uncertainty for the extraction of precise values of neutrino oscillation parameters in long-baseline accelerator neutrino experiments. To improve such modeling and minimize the corresponding uncertainties, a new generation of detectors is being developed, which aim to measure the complete final state of particles resulting from neutrino interactions. In order to fully benefit from the improved detector capabilities, precise simulations of the nuclear effects on the final-state nucleons are needed. This article presents the study of the in-medium propagation of knocked-out protons, i.e., final-state interactions (FSI), comparing the NuWro and INCL cascade models. The INCL model is used here for the first time to predict exclusive final states of neutrino interactions. This study of INCL in the framework of neutrino interactions features various novelties, including the production of nuclear clusters (e.g., deuterons, $\alpha$ particles) in the final state. The paper includes a complete characterization of the final state after FSI, comparisons to available measurements of single transverse variables, and an assessment of the observability of nuclear clusters., Comment: 19 pages, 19 figures

This article describes the setup and performance of the near and far detectors in the Double Chooz experiment. The electron antineutrinos of the Chooz nuclear power plant were measured in two identically designed detectors with different average baselines of about 400 m and 1050 m from the two reactor cores. Over many years of data taking the neutrino signals were extracted from interactions in the detectors with the goal of measuring a fundamental parameter in the context of neutrino oscillation, the mixing angle {\theta}13. The central part of the Double Chooz detectors was a main detector comprising four cylindrical volumes filled with organic liquids. From the inside towards the outside there were volumes containing gadolinium-loaded scintillator, gadolinium-free scintillator, a buffer oil and, optically separated, another liquid scintillator acting as veto system. Above this main detector an additional outer veto system using plastic scintillator strips was installed. The technologies developed in Double Chooz were inspiration for several other antineutrino detectors in the field. The detector design allowed implementation of efficient background rejection techniques including use of pulse shape information provided by the data acquisition system. The Double Chooz detectors featured remarkable stability, in particular for the detected photons, as well as high radiopurity of the detector components., Comment: 49 pages, 29 figures