The early detection of handguns and knives from surveillance videos is crucial to enhance people’s safety. Despite the increasing development of Deep Learning (DL) methods for general object detection, weapon detection from surveillance videos still presents open challenges. Among these, the most significant are: (i) the very small size of the weapons with respect to the camera field of view and (ii) the need of a real-time feedback, even when using low-cost edge devices for computation. Complex and recently-developed DL architectures could mitigate the former challenge but do not satisfy the latter one. To tackle such limitation, the proposed work addresses the weapon-detection task from an edge perspective. A double-step DL approach was developed and evaluated against other state-of-the-art methods on a custom indoor surveillance dataset. The approach is based on a first Convolutional Neural Network (CNN) for people detection which guides a second CNN to identify handguns and knives. To evaluate the performance in a real-world indoor environment, the approach was deployed on a NVIDIA Jetson Nano edge device which was connected to an IP camera. The system achieved near real-time performance without relying on expensive hardware. The results in terms of both COCO Average Precision (AP = 79.30) and Frames per Second (FPS = 5.10) on the low-power NVIDIA Jetson Nano pointed out the goodness of the proposed approach compared with the others, encouraging the spread of automated video surveillance systems affordable to everyone.

Background: Heartbeat detection is a crucial step in several clinical fields. Laser Doppler Vibrometer (LDV) is a promising non-contact measurement for heartbeat detection. The aim of this work is to assess whether machine learning can be used for detecting heartbeat from the carotid LDV signal. Methods: The performances of Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF) and K-Nearest Neighbor (KNN) were compared using the leave-one-subject-out cross-validation as the testing protocol in an LDV dataset collected from 28 subjects. The classification was conducted on LDV signal windows, which were labeled as beat, if containing a beat, or no-beat, otherwise. The labeling procedure was performed using electrocardiography as the gold standard. Results: For the beat class, the f1-score ($f1$) values were 0.93, 0.93, 0.95, 0.96 for RF, DT, KNN and SVM, respectively. No statistical differences were found between the classifiers. When testing the SVM on the full-length (10 min long) LDV signals, to simulate a real-world application, we achieved a median macro-$f1$ of 0.76. Conclusions: Using machine learning for heartbeat detection from carotid LDV signals showed encouraging results, representing a promising step in the field of contactless cardiovascular signal analysis.

Laryngeal Cancer (LC) constitutes approximately one third of head and neck cancers. Detecting early-stage lesions in this anatomical region is crucial for achieving a high survival rate. However, it poses significant diagnostic challenges owing to the varied appearance of lesions and the need for precise characterization for appropriate clinical management. Conventional diagnostic approaches rely heavily on endoscopic examination, which often requires expert interpretation and may be limited by subjective assessment. Deep learning (DL) approaches offer promising opportunities for automating lesion detection, but their efficacy in handling multi-modal imaging data and accurately localizing small lesions remains a subject of investigation. Furthermore, the clinical domain may largely benefit from the deployment of efficient DL methods that can ensure equitable access to advanced technologies, regardless of the availability of resources that can often be limited. In this study, a DL-based approach, named SRE-YOLO, was introduced to provide real-time assistance to less-experienced personnel during laryngeal assessment, by automatically detecting lesions at different scales from endoscopic White Light (WL) and Narrow-Band Imaging (NBI) images. During the training, the SRE-YOLO integrates a YOLOv8 nano (YOLOv8n) baseline with a Super-Resolution (SR) branch to enhance lesion detection. This last component is decoupled during inference to preserve the low computational demand of the YOLOv8n baseline. The evaluation was conducted on a multi-center dataset, encompassing diverse laryngeal pathologies and acquisition modalities. The SRE-YOLO method improved the Average Precision (AP_@IoU=0.5) in lesion detection by 5% with respect to the YOLOv8n baseline, while maintaining the inference speed of 58.8 Frames Per Second (FPS). Comparative analyses against state-of-the-art DL methods highlighted the efficacy of the SRE-YOLO approach in balancing detection accuracy, computational efficiency, and real-time applicability. This research underscores the potential of SRE-YOLO in developing efficient DL-driven decision support systems for real-time detection of laryngeal lesions at different scales from both WL and NBI endoscopic data.

The generation of Synthetic Computed Tomography (sCT) images has become a pivotal methodology in modern clinical practice, particularly in the context of Radiotherapy (RT) treatment planning. The use of sCT enables the calculation of doses, pushing towards Magnetic Resonance Imaging (MRI) guided radiotherapy treatments. Moreover, with the introduction of MRI-Positron Emission Tomography (PET) hybrid scanners, the derivation of sCT from MRI can improve the attenuation correction of PET images. Deep learning methods for MRI-to-sCT have shown promising results, but their reliance on single-centre training dataset limits generalisation capabilities to diverse clinical settings. Moreover, creating centralised multi-centre datasets may pose privacy concerns. To address the aforementioned issues, we introduced FedSynthCT-Brain, an approach based on the Federated Learning (FL) paradigm for MRI-to-sCT in brain imaging. This is among the first applications of FL for MRI-to-sCT, employing a cross-silo horizontal FL approach that allows multiple centres to collaboratively train a U-Net-based deep learning model. We validated our method using real multicentre data from four European and American centres, simulating heterogeneous scanner types and acquisition modalities, and tested its performance on an independent dataset from a centre outside the federation. In the case of the unseen centre, the federated model achieved a median Mean Absolute Error (MAE) of 102.0 HU across 23 patients, with an interquartile range of 96.7-110.5 HU. The median (interquartile range) for the Structural Similarity Index (SSIM) and the Peak Signal to Noise Ratio (PNSR) were 0.89 (0.86-0.89) and 26.58 (25.52-27.42), respectively. The analysis of the results showed acceptable performances of the federated approach, thus highlighting the potential of FL to enhance MRI-to-sCT to improve generalisability and advancing safe and equitable clinical applications while fostering collaboration and preserving data privacy.

The database here described contains data relevant to preterm infants' movement acquired in neonatal intensive care units (NICUs). The data consists of 16 depth videos recorded during the actual clinical practice. Each video consists of 1000 frames (i.e., 100s). The dataset was acquired at the NICU of the Salesi Hospital, Ancona (Italy). Each frame was annotated with the limb-joint location. Twelve joints were annotated, i.e., left and right shoul- der, elbow, wrist, hip, knee and ankle. The database is freely accessible at http://doi.org/10.5281/zenodo.3891404. This dataset represents a unique resource for artificial intelligence researchers that want to develop algorithms to provide healthcare professionals working in NICUs with decision support. Hence, the babyPose dataset is the first annotated dataset of depth images relevant to preterm infants' movement analysis.

Patients suffering from neurological diseases may develop dysarthria, a motor speech disorder affecting the execution of speech. Close and quantitative monitoring of dysarthria evolution is crucial for enabling clinicians to promptly implement patients' management strategies and maximizing effectiveness and efficiency of communication functions in term of restoring, compensating or adjusting. In the clinical assessment of orofacial structures and functions, at rest condition or during speech and non-speech movements, a qualitative evaluation is usually performed, throughout visual observation. To overcome limitations posed by qualitative assessments, this work presents a store-and-forward self-service telemonitoring system that integrates, within its cloud architecture, a convolutional neural network (CNN) for analyzing video recordings acquired by individuals with dysarthria. This architecture - called facial landmark Mask RCNN - aims at locating facial landmarks as a prior for assessing the orofacial functions related to speech and examining dysarthria evolution in neurological diseases. When tested on the Toronto NeuroFace dataset, a publicly available annotated dataset of video recordings from patients with amyotrophic lateral sclerosis (ALS) and stroke, the proposed CNN achieved a normalized mean error equal to 1.79 on localizing the facial landmarks. We also tested our system in a real-life scenario on 11 bulbar-onset ALS subjects, obtaining promising outcomes in terms of facial landmark position estimation. This preliminary study represents a relevant step towards the use of remote tools to support clinicians in monitoring the evolution of dysarthria.