68994 research outputs found
Sort by
SHORT-COURSE RADIOTHERAPY FOR OLDER PATIENTS WITH LOCALLY ADVANCED RECTAL CANCER AND UNFIT FOR CHEMOTHERAPY: THE SOFT STUDY
No full textBackground The treatment of locally advanced rectal cancer (LARC) in older patients is not sufficiently standardized in international guidelines. Short-course radiotherapy (SCRT) followed by delayed surgery has emerged as a promising therapeutic option for these patients, offering potential benefits in terms of reduced treatment time and toxicity, along with improved convenience. Materials and Methods From February 2019 to April 2024, a total of 141 older patients aged ≥ 70 years [median age = 79 (range, 70–91)], unfit for chemotherapy and with LARC (stage II-III) adenocarcinoma, underwent SCRT (5 daily fractions of 5 Gy each for a total dose of 25 Gy) followed by delayed surgery (no earlier than 6 weeks after the end of treatment). The study was conducted at 11 centers in Italy. Down-staging rates, relapse-free survival (RFS), overall survival (OS), cancer-specific survival (CSS), safety, and mortality were analyzed. Results Down-staging occurred in 87 cases (61.7 %). Complete radiological responses after SCRT and before surgery were reported in 10 patients (7.1 %), while partial responses were reported in 77 (54.6 %) cases. All patients underwent delayed surgery. The R0 resection rate was 93.6 %. Eight patients (5.7 %) had a pathological complete response (pCR). At a follow-up of 70.5 months, the median RFS, OS and CSS were 31.5, 40.5 and 41.5 months, respectively. Post-operative severe morbidity was 23.4 %, and mortality was 2.8 %. Conclusions Overall, SCRT offers a viable alternative to conventional long-term radiotherapy plus chemotherapy in older patients with LARC, with favorable results in terms of efficacy, limited toxicity and low mortality
Multimodal vision and Artificial Intelligence for the characterization of construction and demolition and municipal solid waste
No full textLa gestione sostenibile dei rifiuti solidi richiede sistemi di caratterizzazione automatica che integrino robustezza metrologica e scalabilità industriale. Questa tesi propone un framework basato su visione artificiale multimodale e algoritmi di intelligenza artificiale applicato a due domini: rifiuti da costruzione e demolizione (CDW) e rifiuti solidi urbani indifferenziati (MSW).
Per i CDW sono stati sviluppati modelli di riconoscimento basati su immagini RGB operanti in tempo reale su impianto pilota, affiancati da protocolli di caratterizzazione spettrale in laboratorio. Mediante termografia IR attiva è stata stimata l’emissività di materiali da costruzione (calcestruzzo, legno, mattoni, ceramica), mentre l’analisi iperspettrale ha consentito l’estrazione di firme spettrali distintive. I dati, classificati con SAM, MLP e CNN, hanno raggiunto accuratezze superiori al 90%, dimostrando la trasferibilità del modello da ambiente controllato a scenario operativo.
Per i MSW è stata integrata una stereo camera su linea di vagliatura industriale. Sono stati ottimizzati i parametri di acquisizione e addestrati modelli di segmentazione per il riconoscimento delle principali classi presenti. Parallelamente è stato sviluppato un algoritmo deterministico per la stima del volume da nuvole di punti; combinando il volume con la densità media sperimentale è stata stimata la massa. L’analisi di incertezza ha evidenziato un’incertezza <10% sul volume e circa 19% sulla massa, risultato significativo considerando l’elevata velocità del nastro e l’assenza di modifiche alle condizioni operative dell’impianto. Le performance AI risultano ~90% per CDW e ~50% per MSW, differenza attribuibile alla qualità dei dati acquisiti.
Il lavoro dimostra come l’integrazione di sensori ottici e IA possa abilitare sorting automatizzato, migliorare il recupero di materia e supportare la valutazione del rischio (es. carico d’incendio), con prospettive di trasferimento verso altri contesti industriali e sistemi smart-factory.Sustainable and safe solid waste management requires automated characterization tools combining metrological robustness with industrial scalability. This thesis proposes a framework based on multimodal computer vision and artificial intelligence applied to two complementary domains: Construction and Demolition Waste (CDW) and unsorted Municipal Solid Waste (MSW).
For CDW, AI-based recognition models using RGB images were developed and deployed in real time on a pilot plant. These were complemented by laboratory spectral characterization protocols. Active infrared thermography was used to estimate the emissivity of key construction materials (concrete, wood, bricks, ceramics), while hyperspectral imaging (400–1000 nm) enabled the extraction of distinctive spectral signatures. Data were classified using Spectral Angle Mapper (SAM), Multilayer Perceptron (MLP), and Convolutional Neural Networks (CNN), achieving accuracies above 90% and demonstrating transferability from controlled laboratory conditions to in-line industrial scenarios.
In the MSW study, a stereo camera was integrated on an industrial sorting line. Acquisition parameters were optimized according to lighting and belt speed constraints, and segmentation models were trained to recognize the most representative waste classes. A deterministic algorithm based on point cloud processing was implemented for volume estimation. By combining volume with experimentally derived average density, waste mass was calculated. Uncertainty analysis showed <10% uncertainty for volume and approximately 19% for mass. Considering the high conveyor speed and uninterrupted plant operation, these results are operationally promising.
Overall, the integration of optical sensors and AI techniques enables automated sorting, improved material recovery, and enhanced risk assessment (e.g., fire load estimation), with strong potential for scalability and transfer to other waste management and smart factory contexts
Toward Trustworthy AI in Retail: A Case Study of Vending Machines in Public Environments
No full textArtificial intelligence (AI) systems are becoming increasingly embedded in physical retail environments, enabling real-time personalisation through face and behaviour analysis. This paper critically examines the ethical implications of one such system: an intelligent vending machine (VM) infrastructure deployed in 30 real-world locations across Italy. This infrastructure uses computer vision and machine learning to adapt promotional content based on inferred age, gender and interaction patterns. While such systems are technically effective and privacy-preserving, they raise important concerns around fairness, transparency, consent, and profiling. Drawing on the EU's Ethics Guidelines for Trustworthy AI and the ALTAI framework, we evaluate the VM system against core ethical principles, particularly those of justice and explicability. Key challenges identified include demographic bias in personalisation, the absence of meaningful user awareness and implicit profiling in public spaces. Based on our findings, we propose practical enhancements to improve fairness-aware design, embedded transparency mechanisms and symbolic consent models. Our work provides a field-tested ethical assessment of AI in physical retail and offers actionable strategies for aligning real-world deployments with responsible AI standards
RISPOSTE FISIOLOGICHE E DI CRESCITA DI ALBERI DA FRUTTO A STRESS ABIOTICI
No full textGli eventi di ondate di calore e siccità sono sempre più frequenti e rappresentano una crescente minaccia per la frutticoltura perenne. Questa tesi ha valutato, in ambiente protetto su piante in vaso, gli effetti di elevate temperature della zona radicale, da sole o combinate con alte temperature dell’aria e siccità, su fisiologia, crescita e sopravvivenza del pero, nonché il potenziale della spettroscopia nel vicino infrarosso (NIR) per il rilevamento rapido e non distruttivo dello stress. In una prima prova su Conference autoradicata, 45 °C nella zona radicale hanno causato un rapido collasso dell’attività fotosintetica, danni a radici e parte aerea e morte della pianta. A 40 °C, esposizioni brevi hanno ridotto gli scambi gassosi, mentre esposizioni prolungate hanno diminuito crescita e biomassa. In un secondo esperimento, un riscaldamento acuto a 50 °C ha compromesso gravemente l’integrità radicale e la sopravvivenza di William autoradicato e del portinnesto di cotogno BA29; 40 °C prolungati hanno ridotto sviluppo radicale e accumulo di biomassa. La spettroscopia NIR ha mostrato buona accuratezza nel stimare contenuto idrico relativo e clorofilla in pero e cotogno stressati. In un terzo esperimento fattoriale, il riscaldamento radicale (40 °C) combinato con alte temperature dell’aria e siccità ha aggravato il declino fisiologico e le perdite di crescita. Questi risultati evidenziano l’importanza della temperatura del suolo nelle prestazioni complessive del portinnesto cotogno BA29 e delle piante di pero William e Conference autoradicate, evidenziando per tutti i genotipi studiati, soglie termiche critiche oltre le quali i danni diventano evidenti, fino a diventare irreversibili. Le soglie critiche di temperatura dell’apparato radicale identificate insieme al potenziale degli strumenti NIR per il rilevamento precoce dello stress, possono offrire indicazioni utili per sviluppare strategie integrate di gestione del frutteto volte ad affrontare le sfide poste dal cambiamento climatico.Heat waves and drought events are becoming increasingly frequent and pose a growing threat to perennial fruit production. This thesis evaluated, under protected conditions using potted plants, the effects of elevated root-zone temperatures, alone or combined with high air temperatures and drought, on pear physiology, growth, and survival, as well as the potential of near-infrared (NIR) spectroscopy as a rapid, non-destructive tool for stress detection. In a first trial on self-rooted ‘Conference’, a root-zone temperature of 45 °C caused a rapid collapse of photosynthetic activity, damage to roots and shoots, and plant death. At 40 °C, short-term exposure reduced gas exchange, whereas prolonged exposure decreased growth and biomass accumulation. In a second experiment, acute root-zone heating at 50 °C severely compromised root integrity and plant survival in self-rooted ‘William’ and the quince rootstock BA29; prolonged exposure to 40 °C reduced root development and biomass accumulation. NIR spectroscopy showed good accuracy in estimating relative water content and chlorophyll in stressed pear and quince plants. In a third factorial experiment, root-zone heating (40 °C) combined with high air temperatures and drought exacerbated physiological decline and growth losses of Conference pear plants. These results highlight the importance of soil temperature for the overall performance of quince rootstock BA29 and self-rooted ‘William’ and ‘Conference’, identifying critical thermal thresholds beyond which damage becomes evident and may become irreversible. The identified root-zone temperature thresholds, together with the potential of NIR tools for early stress detection, provide useful insights for developing integrated orchard management strategies to address climate change challenges
Rock strength assessment in tectonically deformed calcareous rocks integrating Equotip, ultrasound velocity, and geo-structural fracture analysis
No full textIn complex geological environments, the analysis of drill cores to determine rock strength can be challenging due to the wide variability in the degree of fracturing, leading to subjectivity in the collection of representative samples for uniaxial compressive strength testing. This study evaluates non-destructive techniques on calcareous rocks with different tectonic deformations, including Equotip hardness, ultrasound P-wave velocity, thin section analysis, and calcimetry, integrated with photogrammetric fracture analysis. The investigated carbonate rock samples are sourced from drill cores derived from the Umbria-Marche fold and thrust belt (northern Apennines, Italy), including a gently dipping limb of an anticline, a hinge zone of an anticline, and a fault zone associated with a thrust. Fracture intensity, quantified by the P21 parameter using photogrammetric techniques on pre-loading rock samples, is assessed alongside macroscopic identification of discontinuities (such as stylolites, veins, and joints) using marker colours to monitor failures during uniaxial compression testing. Empirical correlations depicted by single and multi-linear relationships indicate a strong dependence between the mechanical and physical properties of limestones. Both Equotip and P-wave velocity are influenced by fracture intensity, but P-wave velocity varies significantly with discontinuity orientation, especially at 45°–90°. To refine uniaxial compressive strength predictions and mitigate multicollinearity, statistical approaches, including linear and multilinear regression, Principal Component Analysis and Gaussian Process Regression, were tested. Findings improve the reliability of non-destructive techniques for assessing rock strength in structurally complex settings, with implications for geotechnical applications
Smart retrofit solution for the industrial sector: sustainable digitalisation on legacy machines
No full textThis work presents a systematic methodology and a modular Smart Retrofit Architecture
(SRA) to digitalise legacy machinery through interoperable hardware, software and a
communication layer that enable data acquisition, analytics, and human-centred decision
support. The approach was progressively refined through preliminary experiments,
materialised in a dedicated toolkit (Smart Retrofit Toolkit, SRT), and subsequently applied
in pilot implementations developed within the framework of the European AIDEAS project.
It was first operationalised in the AIDEAS Smart Retrofitter (AI-SR) and validated on
industrial and laboratory testbeds. At PAMA S.p.A., AI-SR was applied to large machine
tools, demonstrating non-invasive sensing, real-time monitoring, AI-based analysis, and
measurable operational and environmental improvements; scalability was verified on a
second machine through rapid model transfer and a shared ontology. At D2 Technology, AI-
SR was implemented on a testbench with a configurable database and a three-panel UI for
monitoring, historical data analysis, and scenario simulation. Finally, at IKERLAN, a Real-
Time Data Simulator digitally retrofitted a crank–slider testbench, streaming data for near-
real-time condition assessment using statistical and AI models. Overall, the methodology,
SRA, SRT and AI-SR demonstrate that smart retrofit provides an efficient and sustainable
pathway to achieving Industry 4.0 and 5.0 functionalities without full equipment
replacement
Fatigue softening mechanisms in UFG-EUROFER97 steel. A microstructural study
No full textA novel Thermo-Mechanical Treatment (TMT) developed by us leads to a microstructure with Ultra-Fine Grains (UFG) resulting in improved mechanical properties compared to those of standard EUROFER97 steel, foreseen for nuclear fusion applications. Since softening was observed after High Cycle Fatigue (HCF) tests at room temperature, this work investigated the specific microstructural mechanisms. The microstructure evolution of fatigued samples has been studied by means of Electron Back-Scattered Diffraction (EBSD), X-Ray Diffraction (XRD), Transmission Electron Microscopy (TEM) and Scanning Electron Microscopy (SEM). Softening resulted to be originated by the phenomenon of fatigue-assisted grain coarsening. HCF always leads to the increase of grain and sub-grain size together with texture change, and such modifications depend on the stress level. Under cyclic stress, boundary migration and absorption in the boundaries of dislocations produced by Frank-Read sources inside the grains lead to an excess of dislocations in the boundaries. Although the stability and mobility of High- Angle Grain Boundaries (HAGBs) and Low-Angle Grain Boundaries (LAGBs) are totally different, both the boundaries become unstable. In the final step of the process, the collapse or sliding of unstable HAGBs and LAGBs give rise to a population of grains and sub-grains of larger size. The novel TMT demonstrates improved softening behavior compared to standard EUROFER97, and the results obtained will guide future research toward further improvements
Adhesive strength of bio-inspired fibrillar arrays in the presence of contact defects
No full textThe performance of bio-inspired fibrillar adhesives can be compromised by surface roughness, manufacturing imperfections or impurities. Previous studies investigated the cases of distributed defects on the array, and defects at the level of single fibrils. However, the influence of localized, macroscopic defects remains largely unexplored. Using numerical simulations of a discrete mechanical model for a fibrillar adhesive with a thick backing layer, we investigate how the size and location of a single circular defect affect the established scaling law between the adhesion force (F) and the effective compliance of the system (β), i.e, F ∝ β−1/2. We find that edge defects are generally more detrimental than central ones, as they act as pre-cracks that amplify stress concentrations at the adhesive's edge, accelerating a crack-like failure. Consequently, the established adhesion scaling law is preserved, with the defect only reducing the effective contact area. In contrast, a central defect fundamentally alters the mechanics of detachment. By transforming the contact geometry into an annulus, it promotes more uniform load sharing across the remaining fibrils and mitigates the edge-dominated failure mechanism. This change makes the adhesive strength less sensitive to the compliance of the system, as reflected by a less negative scaling exponent. The transition between these two regimes appears to occur for defects whose boundary merges with the one of the adhesive
La transizione energetica nel prisma delle soluzioni privatistiche: il modello dell'agrivoltaico
No full textIn un contesto in cui l’Italia e l’Unione europea hanno da tempo orientato le proprie politiche nel senso della decarbonizzazione e in cui proliferano impianti di produzione di energia da fonti rinnovabili in aree agricole, il modello dell’agrivoltaico, inteso quale soluzione tecnologica innovativa capace di combinare produzione energetica e produzione agricola, appare meritevole di approfondimento per il ruolo centrale che è destinato ad assumere. Volendo sviluppare una riflessione su un tema oggetto di una normativa ancora in fieri e per certi versi disorganica e carente, si è, dunque, proceduto ad analizzarne la relativa disciplina, partendo da una ricognizione del travagliato itinerario che ha condotto all’affermazione del principio dello sviluppo sostenibile e del principio della massima diffusione delle fonti rinnovabili. Si sono poi individuate le linee fondamentali delle disposizioni relative alla tecnologica agrivoltaica, evidenziandone limiti e contraddizioni. Ci si è, infine, dedicati ad approfondire gli strumenti contrattuali astrattamente idonei a consentire l’acquisizione delle aree agricole destinate all’installazione dei predetti impianti, evidenziando gli elementi di discontinuità rispetto alla precedente esperienza maturata con riferimento agli impianti fotovoltaici con moduli a terra di cui gli impianti agrivoltaici rappresentano un’evoluzione
Dhekalos Lab to Metaverse: Exploring XR Strategies for Digital Cultural Heritage
No full textThis chapter presents the design and implementation of Meta-Dhekalos, a digital twin of the Dhekalos Laboratory at the Università Politecnica delle Marche, developed within the Spatial platform. The project aims to transpose the physical lab into the Metaverse, enabling remote access, immersive interaction, and participatory dissemination of cultural heritage research. A curated selection of six XR-based digital experiences, spanning AR, VR, MR, and AI-enhanced environments, is also presented and integrated into the new Meta-space to reflect the lab’s evolving methodological and technological approaches. Each case was chosen for its ability to represent different stages and paradigms of immersive cultural storytelling. The chapter documents the criteria used in this selection and the design strategies adopted to translate them into a shared 3D environment. Finally, it presents the preliminary results of an evaluation test conducted with educators, offering early insights into presence, usability, and educational impact across the different experiences