arXiv subjects

Marianna Fontana

Publications and source records attributed to Marianna Fontana.

8 recordsLinked to original sources

arXiv2026

Real-time lepton identification at LHCb in Run 3 using Lipschitz neural networks

The LHCb physics program in Run 3 relies critically on the efficient real-time selection of events containing muons and electrons, which are key signatures in a wide range of heavy-flavor and exotic decay processes. The LHCb Run 3 detector now operates with a fully software-based trigger that processes the complete detector readout at the LHC bunch-crossing rate, with the first trigger stage executed on GPUs. In this environment, particle-identification algorithms must achieve high efficiency and background rejection while satisfying stringent constraints on throughput and memory footprint. We present algorithms for muon and electron identification in the LHCb Run 3 GPU trigger based on Lipschitz-constrained neural networks. Separate networks are developed for muons and electrons and are trained using simulated events. Their performance is evaluated relative to the previous baseline algorithms, demonstrating improved discrimination across a wide range of kinematic regions while remaining compatible with the requirements of real-time GPU execution.

hep-ex↗

arXiv2026

A converged architecture for processing 32 Tbps of physics data in real-time at the LHCb experiment

The LHCb detector at the Large Hadron Collider has been upgraded to acquire an unprecedented 32 Tbps of particle-collision data to provide new insights in the High Energy Physics domain. The data produced by the detector is filtered in real-time to select interesting collisions. As part of the upgrade, a pre-filtering stage has been removed leading to a factor 40 increase in data rate. To deal with the high throughput demands of LHCb real-time data processing, we present an off-the-shelf network architecture using zero-copy techniques in conjunction with an efficient, fully-GPU-based filter. Our converged architecture is able to process the full 32 Tbps of particle-collision data in real-time, the highest in any physics experiment to date. Our result extends the reach of the LHCb physics programme and sets a new standard for real-time data processing at particle physics experiments.

hep-ex↗

arXiv2024

Inline AI: Open-source Deep Learning Inference for Cardiac MR

Cardiac Magnetic Resonance (CMR) is established as a non-invasive imaging technique for evaluation of heart function, anatomy, and myocardial tissue characterization. Quantitative biomarkers are central for diagnosis and management of heart disease. Deep learning (DL) is playing an ever more important role in extracting these quantitative measures from CMR images. While many researchers have reported promising results in training and evaluating models, model deployment into the imaging workflow is less explored. A new imaging AI framework, the InlineAI, was developed and open-sourced. The main innovation is to enable the model inference inline as a part of imaging computation, instead of as an offline post-processing step and to allow users to plug in their models. We demonstrate the system capability on three applications: long-axis CMR cine landmark detection, short-axis CMR cine analysis of function and anatomy, and quantitative perfusion mapping. The InlineAI allowed models to be deployed into imaging workflow in a streaming manner directly on the scanner. The model was loaded and inference on incoming images were performed while the data acquisition was ongoing, and results were sent back to scanner. Several biomarkers were extracted from model outputs in the demonstrated applications and reported as curves and tabular values. All processes are full automated. the model inference was completed within 6-45s after the end of imaging data acquisition.

eess.IV↗

arXiv2020

Muon identification for LHCb Run 3

Muon identification is of paramount importance for the physics programme of LHCb. In the upgrade phase, starting from Run 3 of the LHC, the trigger of the experiment will be solely based on software. The luminosity increase to $2\times10^{33}$ cm$^{-2}$s$^{-1}$ will require an improvement of the muon identification criteria, aiming at performances equal or better than those of Run 2, but in a much more challenging environment. In this paper, two new muon identification algorithms developed in view of the LHCb upgrade are presented, and their performance in terms of signal efficiency versus background reduction is shown.

hep-ex↗

arXiv2020

Landmark detection in Cardiac Magnetic Resonance Imaging Using A Convolutional Neural Network

Purpose: To develop a convolutional neural network (CNN) solution for robust landmark detection in cardiac MR images. Methods: This retrospective study included cine, LGE and T1 mapping scans from two hospitals. The training set included 2,329 patients and 34,019 images. A hold-out test set included 531 patients and 7,723 images. CNN models were developed to detect two mitral valve plane and apical points on long-axis (LAX) images. On short-axis (SAX) images, anterior and posterior RV insertion points and LV center were detected. Model outputs were compared to manual labels by two operators for accuracy with a t-test for statistical significance. The trained model was deployed to MR scanners. Results: For the LAX images, success detection was 99.8% for cine, 99.4% for LGE. For the SAX, success rate was 96.6%, 97.6% and 98.9% for cine, LGE and T1-mapping. The L2 distances between model and manual labels were 2 to 3.5 mm, indicating close agreement between model landmarks to manual labels. No significant differences were found for the anterior RV insertion angle and LV length by the models and operators for all views and imaging sequences. Model inference on MR scanner took 610ms/5.6s on GPU/CPU, respectively, for a typical cardiac cine series. Conclusions: This study developed, validated and deployed a CNN solution for robust landmark detection in both long and short-axis CMR images for cine, LGE and T1 mapping sequences, with the accuracy comparable to the inter-operator variation.

eess.IV↗

arXiv2020

Automated Inline Analysis of Myocardial Perfusion MRI with Deep Learning

Recent development of quantitative myocardial blood flow (MBF) mapping allows direct evaluation of absolute myocardial perfusion, by computing pixel-wise flow maps. Clinical studies suggest quantitative evaluation would be more desirable for objectivity and efficiency. Objective assessment can be further facilitated by segmenting the myocardium and automatically generating reports following the AHA model. This will free user interaction for analysis and lead to a 'one-click' solution to improve workflow. This paper proposes a deep neural network based computational workflow for inline myocardial perfusion analysis. Adenosine stress and rest perfusion scans were acquired from three hospitals. Training set included N=1,825 perfusion series from 1,034 patients. Independent test set included 200 scans from 105 patients. Data were consecutively acquired at each site. A convolution neural net (CNN) model was trained to provide segmentation for LV cavity, myocardium and right ventricular by processing incoming 2D+T perfusion Gd series. Model outputs were compared to manual ground-truth for accuracy of segmentation and flow measures derived on global and per-sector basis. The trained models were integrated onto MR scanners for effective inference. Segmentation accuracy and myocardial flow measures were compared between CNN models and manual ground-truth. The mean Dice ratio of CNN derived myocardium was 0.93 +/- 0.04. Both global flow and per-sector values showed no significant difference, compared to manual results. The AHA 16 segment model was automatically generated and reported on the MR scanner. As a result, the fully automated analysis of perfusion flow mapping was achieved. This solution was integrated on the MR scanner, enabling 'one-click' analysis and reporting of myocardial blood flow.

q-bio.QM↗

arXiv2019

Automatic In-line Quantitative Myocardial Perfusion Mapping: processing algorithm and implementation

Quantitative myocardial perfusion mapping has advantages over qualitative assessment, including the ability to detect global flow reduction. However, it is not clinically available and remains as a research tool. Building upon the previously described imaging sequence, this paper presents algorithm and implementation of an automated solution for inline perfusion flow mapping with step by step performance characterization. An inline perfusion flow mapping workflow is proposed and demonstrated on normal volunteers. Initial evaluation demonstrates the fully automated proposed solution for the respiratory motion correction, AIF LV mask detection and pixel-wise mapping, from free-breathing myocardial perfusion imaging.

eess.IV↗

arXiv2018

Selection and processing of calibration samples to measure the particle identification performance of the LHCb experiment in Run 2

Since 2015, with the restart of the LHC for its second run of data taking, the LHCb experiment has been empowered with a dedicated computing model to select and analyse calibration samples to measure the performance of the particle identification (PID) detectors and algorithms. The novel technique was developed within the framework of the innovative trigger model of the LHCb experiment, which relies on online event reconstruction for most of the datasets, reserving offline reconstruction to special physics cases. The strategy to select and process the calibration samples, which includes a dedicated data-processing scheme combining online and offline reconstruction, is discussed. The use of the calibration samples to measure the detector PID performance, and the efficiency of PID requirements across a large range of decay channels, is described. Applications of the calibration samples in data-quality monitoring and validation procedures are also detailed.

hep-ex↗