A5052086841- Radiomics and Machine Learning in Medical Imaging
- Video Analysis and Summarization
- 3D Surveying and Cultural Heritage
- COVID-19 diagnosis using AI
- Advanced Vision and Imaging
- Cinema and Media Studies
- Advanced Image Processing Techniques
- Image Processing Techniques and Applications
- Artificial Intelligence in Healthcare and Education
- Cardiac Imaging and Diagnostics
- Image Processing and 3D Reconstruction
- Phonocardiography and Auscultation Techniques
- Generative Adversarial Networks and Image Synthesis
- 3D Shape Modeling and Analysis
- Robotics and Sensor-Based Localization
- Cell Image Analysis Techniques
- Radiology practices and education
- Machine Learning in Healthcare
- Media Influence and Health
- Cardiac Arrest and Resuscitation
- Advanced Image and Video Retrieval Techniques
- Medical Image Segmentation Techniques
- Hair Growth and Disorders
- Functional Brain Connectivity Studies
- Neural dynamics and brain function
University of Brescia
2017-2025
Brescia University
2019-2024
Faculty of Media
2020
Vrije Universiteit Amsterdam
2020
Leipzig University
2020
Full Laboratory Automation is revolutionizing work habits in an increasing number of clinical microbiology facilities worldwide, generating huge streams digital images for interpretation. Contextually, deep learning architectures are leading to paradigm shifts the way computers can assist with difficult visual interpretation tasks several domains. At crossroads these epochal trends, we present a system able tackle core task microbiology, namely global diagnostic bacterial culture plates,...
This article's main contributions are twofold: 1) to demonstrate how apply the general European Union's High-Level Expert Group's (EU HLEG) guidelines for trustworthy AI in practice domain of healthcare and 2) investigate research question what does "trustworthy AI" mean at time COVID-19 pandemic. To this end, we present results a post-hoc self-assessment evaluate trustworthiness an system predicting multiregional score conveying degree lung compromise patients, developed verified by...
Many clinical and research studies of the human brain require accurate structural MRI segmentation. While traditional atlas-based methods can be applied to volumes from any acquisition site, recent deep learning algorithms ensure high accuracy only when tested on data same sites exploited in training (i.e., internal data). Performance degradation experienced external unseen sites) is due inter-site variability intensity distributions, unique artefacts caused by different MR scanner models...
IntroductionThe healthcare sector invests significantly in communication skills training, but not always with satisfactory results. Recently, generative Large Language Models, have shown promising results medical education. This study aims to use ChatGPT simulate radiographer-patient conversations about the critical moment of claustrophobia management during MRI, exploring how Artificial Intelligence can improve radiographers' skills.MethodsThis exploits specifically designed prompts on...
The apparent distance of the camera from subject a filmed scene, namely shot scale, is one prominent formal features any filmic product, endowed with both stylistic and narrative functions. In this work we propose to use Convolutional Neural Networks for automatic classification scale into Close-, Medium-, or Long-shots. development such tool allows investigating relationship between computed large movie corpora viewers' emotional involvement, purposes as recommendation, analysis, film...
We provide a database containing shot scale annotations (i.e., the apparent distance of camera from subject filmed scene) for more than 792,000 image frames. Frames belong to 124 full movies entire filmographies by 6 important directors: Martin Scorsese, Jean-Luc Godard, Béla Tarr, Federico Fellini, Michelangelo Antonioni, and Ingmar Bergman. Each frame, extracted videos at 1 frame per second, is annotated on following categories: Extreme Close Up (ECU), (CU), Medium (MCU), Shot (MS), Long...
Deep neural networks have revolutionised machine learning, with unparalleled performance in object classification. However, brain imaging (e.g., fMRI), the direct application of Convolutional Neural Networks (CNN) to decoding subject states or perception from data seems impractical given scarcity available data. In this work we propose a robust method transfer information deep learning (DL) features fMRI goal decoding. By adopting Reduced Rank Regression Ridge Regularisation establish...
The perceived distance of the camera from subject a filmed scene, namely <i>shot scale</i>, is prominent formal feature any filmic product, endowed with both stylistic and narrative functions. To measure how shot scale affects lower higher complexity responses in viewers, we first investigate distribution rotation Close, Medium, Long Shots relate to viewers’ rating on <i>film mood</i>, assessed terms hedonic tone, energetic arousal, tense arousal an extensive set 50 film clips. Then...
Huge streams of diagnostic images are expected to be produced daily in the emerging field digital microbiology imaging because ongoing worldwide spread Full Laboratory Automation systems. This is redefining way microbiologists execute tasks. In this context, authors want assess suitability and effectiveness a deep learning approach solve diagnostically relevant but visually challenging task directly identifying pathogens on bacterial growing plates. particular, starting from hyperspectral...
Previous research suggests that particular formal features of film, such as the use close-ups, can affect levels empathy experienced by viewers.Because is a key aspect audience's filmic experience, creative decisions in editing and cinematography may be motivated filmmaker's intention eliciting empathy.The goal this study was to investigate what film scenes intended elicit look like terms those visual theoretically or empirically linked viewer whether these converge on something might dubbed...
The first objective of this study was to apply computer vision and machine learning techniques quantify the effects haircare treatments on hair assembly identify correctly whether unknown tresses were treated or not. second explore compare performance human assessment with that obtained from artificial intelligence (AI) algorithms.Machine applied a data set tress images (virgin bleached), both untreated shampoo conditioner set, aimed at increasing volume whilst improving alignment reducing...
The position and orientation of the camera in relation to subject(s) a movie scene, namely
We provide a database aimed at real-time quantitative analysis of 3D reconstruction and alignment methods, containing 3140 point clouds from 10 subjects/objects. These scenes are acquired with high-resolution scanner. It contains depth maps that produce more than 500k points on average. This dataset is useful to develop new models strategies automatically reconstruct data optical scanners or benchmarking purposes.
The provided database of 260 ECG signals was collected from patients with out-of-hospital cardiac arrest while treated by the emergency medical services. Each signal contains a 9 second waveform showing ventricular fibrillation, followed 1 min post-shock waveform. Patients' ECGs are made available in multiple formats. All recorded during prehospital treatment PFD files, after being anonymized, printed paper, and scanned. For each ECG, dataset also includes whole digitized (9 s pre- each)...
The incorporation of LiDAR technology into some high-end smartphones has unlocked numerous possibilities across various applications, including photography, image restoration, augmented reality, and more. In this paper, we introduce a novel direction that harnesses depth maps to enhance the compression corresponding RGB camera images. To best our knowledge, represents initial exploration in particular research direction. Specifically, propose Transformer-based learned system capable...
The hypothesis that a deep learning (DL) model can produce long-term prognostic information from chest X-ray (CXR) has already been confirmed within cancer screening programs. We summarize our experience with DL prediction of mortality, plain CXR, in patients referred for angina and coronary angiography.
Deep learning models have shown remarkable performance in electrocardiogram (ECG) analysis, but their success has been constrained by the limited availability and size of ECG datasets, resulting systems that are more task specialists than versatile generalists. In this work, we introduce HuBERT-ECG, a foundation model pre-trained self-supervised manner on large diverse dataset 9.1 million 12-lead ECGs encompassing 164 cardiovascular conditions. By simply adding an output layer, HuBERT-ECG...