GranoScan: an AI-powered mobile app for in-field identification of biotic threats of wheat.

Capitalizing on the widespread adoption of smartphones among farmers and the application of artificial intelligence in computer vision, a variety of mobile applications have recently emerged in the agricultural domain. This paper introduces GranoScan, a freely available mobile app accessible on major online platforms, specifically designed for the real-time detection and identification of over 80 threats affecting wheat in the Mediterranean region. Developed through a co-design methodology involving direct collaboration with Italian farmers, this participatory approach resulted in an app featuring: (i) a graphical interface optimized for diverse in-field lighting conditions, (ii) a user-friendly interface allowing swift selection from a predefined menu, (iii) operability even in low or no connectivity, (iv) a straightforward operational guide, and (v) the ability to specify an area of interest in the photo for targeted threat identification. Underpinning GranoScan is a deep learning architecture named efficient minimal adaptive ensembling that was used to obtain accurate and robust artificial intelligence models. The method is based on an ensembling strategy that uses as core models two instances of the EfficientNet-b0 architecture, selected through the weighted F1-score. In this phase a very good precision is reached with peaks of 100% for pests, as well as in leaf damage and root disease tasks, and in some classes of spike and stem disease tasks. For weeds in the post-germination phase, the precision values range between 80% and 100%, while 100% is reached in all the classes for pre-flowering weeds, except one. Regarding recognition accuracy towards end-users in-field photos, GranoScan achieved good performances, with a mean accuracy of 77% and 95% for leaf diseases and for spike, stem and root diseases, respectively. Pests gained an accuracy of up to 94%, while for weeds the app shows a great ability (100% accuracy) in recognizing whether the target weed is a dicot or monocot and 60% accuracy for distinguishing species in both the post-germination and pre-flowering stage. Our precision and accuracy results conform to or outperform those of other studies deploying artificial intelligence models on mobile devices, confirming that GranoScan is a valuable tool also in challenging outdoor conditions.

Efficient Lung Ultrasound Classification.

A machine learning method for classifying lung ultrasound is proposed here to provide a point of care tool for supporting a safe, fast, and accurate diagnosis that can also be useful during a pandemic such as SARS-CoV-2. Given the advantages (e.g., safety, speed, portability, cost-effectiveness) provided by the ultrasound technology over other examinations (e.g., X-ray, computer tomography, magnetic resonance imaging), our method was validated on the largest public lung ultrasound dataset. Focusing on both accuracy and efficiency, our solution is based on an efficient adaptive ensembling of two EfficientNet-b0 models reaching 100% of accuracy, which, to our knowledge, outperforms the previous state-of-the-art models by at least 5%. The complexity is restrained by adopting specific design choices: ensembling with an adaptive combination layer, ensembling performed on the deep features, and minimal ensemble using two weak models only. In this way, the number of parameters has the same order of magnitude of a single EfficientNet-b0 and the computational cost (FLOPs) is reduced at least by 20%, doubled by parallelization. Moreover, a visual analysis of the saliency maps on sample images of all the classes of the dataset reveals where an inaccurate weak model focuses its attention versus an accurate one.

Raman spectroscopy and topological machine learning for cancer grading.

In the last decade, Raman Spectroscopy is establishing itself as a highly promising technique for the classification of tumour tissues as it allows to obtain the biochemical maps of the tissues under investigation, making it possible to observe changes among different tissues in terms of biochemical constituents (proteins, lipid structures, DNA, vitamins, and so on). In this paper, we aim to show that techniques emerging from the cross-fertilization of persistent homology and machine learning can support the classification of Raman spectra extracted from cancerous tissues for tumour grading. In more detail, topological features of Raman spectra and machine learning classifiers are trained in combination as an automatic classification pipeline in order to select the best-performing pair. The case study is the grading of chondrosarcoma in four classes: cross and leave-one-patient-out validations have been used to assess the classification accuracy of the method. The binary classification achieves a validation accuracy of 81% and a test accuracy of 90%. Moreover, the test dataset has been collected at a different time and with different equipment. Such results are achieved by a support vector classifier trained with the Betti Curve representation of the topological features extracted from the Raman spectra, and are excellent compared with the existing literature. The added value of such results is that the model for the prediction of the chondrosarcoma grading could easily be implemented in clinical practice, possibly integrated into the acquisition system.

Improving plant disease classification by adaptive minimal ensembling.

A novel method for improving plant disease classification, a challenging and time-consuming process, is proposed. First, using as baseline EfficientNet, a recent and advanced family of architectures having an excellent accuracy/complexity trade-off, we have introduced, devised, and applied refined techniques based on transfer learning, regularization, stratification, weighted metrics, and advanced optimizers in order to achieve improved performance. Then, we go further by introducing adaptive minimal ensembling, which is a unique input to the knowledge base of the proposed solution. This represents a leap forward since it allows improving the accuracy with limited complexity using only two EfficientNet-b0 weak models, performing ensembling on feature vectors by a trainable layer instead of classic aggregation on outputs. To the best of our knowledge, such an approach to ensembling has never been used before in literature. Our method was tested on PlantVillage, a public reference dataset used for benchmarking models' performances for crop disease diagnostic, considering both its original and augmented versions. We noticeably improved the state of the art by achieving 100% accuracy in both the original and augmented datasets. Results were obtained using PyTorch to train, test, and validate the models; reproducibility is granted by providing exhaustive details, including hyperparameters used in the experimentation. A Web interface is also made publicly available to test the proposed methods.

Anorexia Nervosa, Body Image Perception and Virtual Reality Therapeutic Applications: State of the Art and Operational Proposal.

Anorexia Nervosa (AN) patients exhibit distorted body representation. The purpose of this study was to explore studies that analyze virtual reality (VR) applications, related to body image issues, to propose a new tool in this field. We conducted a systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. PubMed, EMBASE, Scopus, and Web of Science databases were explored; the review included 25 studies. Research has increased over the last five years. The selected studies, clinical observational studies (n = 16), mostly concerning patients' population with AN (n = 14) or eating disorders (EDs) diagnosis, presented multiple designs, populations involved, and procedures. Some of these studies included healthy control groups (n = 7). Studies on community sample populations were also selected if oriented toward clinical applications (n = 9). The VR technologies in the examined period (about 20 years) have evolved significantly, going from very complex and bulky systems, requiring very powerful computers, to agile systems. The advent of low-cost VR devices has given a big boost to research works. Moreover, the operational proposal that emerges from this work supports the use of biofeedback techniques aimed at evaluating the results of therapeutic interventions in the treatment of adolescent patients diagnosed with AN.

A Case Study of Upper Limb Robotic-Assisted Therapy Using the Track-Hold Device.

The Track-Hold System (THS) project, developed in a healthcare facility and therefore in a controlled and protected healthcare environment, contributes to the more general and broad context of Robotic-Assisted Therapy (RAT). RAT represents an advanced and innovative rehabilitation method, both motor and cognitive, and uses active, passive, and facilitating robotic devices. RAT devices can be equipped with sensors to detect and track voluntary and involuntary movements. They can work in synergy with multimedia protocols developed ad hoc to achieve the highest possible level of functional re-education. The THS is based on a passive robotic arm capable of recording and facilitating the movements of the upper limbs. An operational interface completes the device for its use in the clinical setting. In the form of a case study, the researchers conducted the experimentation in the former Tabarracci hospital (Viareggio, Italy). The case study develops a motor and cognitive rehabilitation protocol. The chosen subjects suffered from post-stroke outcomes affecting the right upper limb, including strength deficits, tremors, incoordination, and motor apraxia. During the first stage of the enrolment, the researchers worked with seven patients. The researchers completed the pilot with four patients because three of them got a stroke recurrence. The collaboration with four patients permitted the generation of an enlarged case report to collect preliminary data. The preliminary clinical results of the Track-Hold System Project demonstrated good compliance by patients with robotic-assisted rehabilitation; in particular, patients underwent a gradual path of functional recovery of the upper limb using the implemented interface.

High-altitude mountain telemedicine.

INTRODUCTION: An innovative teleconsultation platform has been designed, developed and validated between summer 2017 and winter 2018, in five mountain huts and in three remote outpatient clinical centres of the Italian region Valle d'Aosta of the Mont Blanc massif area. METHODS: An ad-hoc videoconference system was developed within the framework of the e-Rés@MONT (Interreg ALCOTRA) European project, to tackle general health problems and high-altitude diseases (such as acute mountain sickness, high-altitude pulmonary and cerebral oedema). The system allows for contacting physicians at the main hospital in Aosta to perform a specific diagnosis and to give specific advice and therapy to the patients in an extreme environment out-hospital setting. At an altitude between 1500-3500 m, five trained nurses performed clinical evaluations (anamnesis, blood pressure, heart rate, oxygen saturation), electrocardiographic and echography monitoring on both tourists and residents as necessary; all of the collected data were sent to the physicians in Aosta. RESULTS: A total of 702 teleconsultation cases were performed: 333 dismissed (47%), 356 observed (51%) and 13 immediate interventions (2%). In 30 cases the physicians decided there was no need for helicopter and ambulance rescue intervention and hospital admissions. The main physiological measures, the classified pathologies, the severe cases and the cost savings are described in this article. DISCUSSION: The e-Rés@MONT teleconsultation platform has been discussed in terms of treated cases, feasibility, proactivity in reducing complexities, direct and indirect advantages, and diagnostics help; moreover, general and specific pros and cons have been debated, and future steps have been exposed.

Privacy by design in systems for assisted living, personalised care, and wellbeing: A stakeholder analysis.

Privacy by design within a system for assisted living, personalised care, and wellbeing is crucial to protect users from misuse of the data collected about their health. Especially if the information is collected through audio-video devices, the question is even more delicate due to the nature of these data. In addition to guaranteeing a high level of privacy, it is necessary to reassure end users about the correct use of these streams. The evolution of data analysis techniques began to take on an important role and increasingly defined characteristics in recent years. The purpose of this paper is twofold: on the one hand, it presents a state of the art about privacy in European Active Healthy Ageing/Active Healthy Ageing projects, with a focus on those related to audio and video processing. On the other hand, it proposes a methodology, developed in the context of the European project PlatfromUptake.eu, to identify clusters of stakeholders and application dimensions (technical, contextual, and business), define their characteristics, and show how privacy constraints affect them. From this study, we then generated a Strengths, Weaknesses, Opportunities, and Threats analysis in which we aim to identify the critical features connected to the selection and involvement of relevant stakeholders for the success of a project. Applying this type of methodology to the initial stages of a project allows understanding of which privacy issues could be related to the various stakeholder groups and which problems can then affect the correct development of the project. The idea is, therefore, to suggest a privacy-by-design approach according to the categories of stakeholders and project dimensions. The analysis will cover technical aspects, legislative and policies-related aspects also regarding the point of view of the municipalities, and aspects related to the acceptance and, therefore, to the perception of the safety of these technologies by the final end users.

A System for Neuromotor Based Rehabilitation on a Passive Robotic Aid.

In the aging world population, the occurrence of neuromotor deficits arising from stroke and other medical conditions is expected to grow, demanding the design of new and more effective approaches to rehabilitation. In this paper, we show how the combination of robotic technologies with progress in exergaming methodologies may lead to the creation of new rehabilitation protocols favoring motor re-learning. To this end, we introduce the Track-Hold system for neuromotor rehabilitation based on a passive robotic arm and integrated software. A special configuration of weights on the robotic arm fully balances the weight of the patients' arm, allowing them to perform a purely neurological task, overcoming the muscular effort of similar free-hand exercises. A set of adaptive and configurable exercises are proposed to patients through a large display and a graphical user interface. Common everyday tasks are also proposed for patients to learn again the associated actions in a persistent way, thus improving life independence. A data analysis module was also designed to monitor progress and compute indices of post-stroke neurological damage and Parkinsonian-type disorders. The system was tested in the lab and in a pilot project involving five patients in the post-stroke chronic stage with partial paralysis of the right upper limb, showing encouraging preliminary results.

Thermal vulnerability detection in integrated electronic and photonic circuits using infrared thermography.

Failure prediction of any electrical/optical component is crucial for estimating its operating life. Using high temperature operating life (HTOL) tests, it is possible to model the failure mechanisms for integrated circuits. Conventional HTOL standards are not suitable for operating life prediction of photonic components owing to their functional dependence on the thermo-optic effect. This work presents an infrared (IR)-assisted thermal vulnerability detection technique suitable for photonic as well as electronic components. By accurately mapping the thermal profile of an integrated circuit under a stress condition, it is possible to precisely locate the heat center for predicting the long-term operational failures within the device under test. For the first time, the reliability testing is extended to a fully functional microwave photonic system using conventional IR thermography. By applying image fusion using affine transformation on multimodal acquisition, it was demonstrated that by comparing the IR profile and GDSII layout, it is possible to accurately locate the heat centers along with spatial information on the type of component. Multiple IR profiles of optical as well as electrical components/circuits were acquired and mapped onto the layout files. In order to ascertain the degree of effectiveness of the proposed technique, IR profiles of complementary metal-oxide semiconductor RF and digital circuits were also analyzed. The presented technique offers a reliable automated identification of heat spots within a circuit/system.

Fault Detection in Power Equipment via an Unmanned Aerial System Using Multi Modal Data.

The power transmission lines are the link between power plants and the points of consumption, through substations. Most importantly, the assessment of damaged aerial power lines and rusted conductors is of extreme importance for public safety; hence, power lines and associated components must be periodically inspected to ensure a continuous supply and to identify any fault and defect. To achieve these objectives, recently, Unmanned Aerial Vehicles (UAVs) have been widely used; in fact, they provide a safe way to bring sensors close to the power transmission lines and their associated components without halting the equipment during the inspection, and reducing operational cost and risk. In this work, a drone, equipped with multi-modal sensors, captures images in the visible and infrared domain and transmits them to the ground station. We used state-of-the-art computer vision methods to highlight expected faults (i.e., hot spots) or damaged components of the electrical infrastructure (i.e., damaged insulators). Infrared imaging, which is invariant to large scale and illumination changes in the real operating environment, supported the identification of faults in power transmission lines; while a neural network is adapted and trained to detect and classify insulators from an optical video stream. We demonstrate our approach on data captured by a drone in Parma, Italy.

A Pilot Study of Infrared Thermography Based Assessment of Local Skin Temperature Response in Overweight and Lean Women during Oral Glucose Tolerance Test.

Obesity is recognized as a major public health issue, as it is linked to the increased risk of severe pathological conditions. The aim of this pilot study is to evaluate the relations between adiposity (and biophysical characteristics) and temperature profiles under thermoneutral conditions in normal and overweight females, investigating the potential role of heat production/dissipation alteration in obesity. We used Infrared Thermography (IRT) to evaluate the thermogenic response to a metabolic stimulus performed with an oral glucose tolerance test (OGTT). Thermographic images of the right hand and of the central abdomen (regions of interests) were obtained basally and during the oral glucose tolerance test (3 h OGTT with the ingestion of 75 g of oral glucose) in normal and overweight females. Regional temperature vs BMI, % of body fat and abdominal skinfold were statistically compared between two groups. The study showed that mean abdominal temperature was significantly greater in lean than overweight participants (34.11 ± 0.70 °C compared with 32.92 ± 1.24 °C, p < 0.05). Mean hand temperature was significantly greater in overweight than lean subjects (31.87 ± 3.06 °C compared with 28.22 ± 3.11 °C, p < 0.05). We observed differences in temperature profiles during OGTT between lean and overweight subjects: The overweight individuals depict a flat response as compared to the physiological rise observed in lean individuals. This observed difference in thermal pattern suggests an energy rate imbalance towards nutrients storage of the overweight subjects.

An Intelligent Cooperative Visual Sensor Network for Urban Mobility.

Smart cities are demanding solutions for improved traffic efficiency, in order to guarantee optimal access to mobility resources available in urban areas. Intelligent video analytics deployed directly on board embedded sensors offers great opportunities to gather highly informative data about traffic and transport, allowing reconstruction of a real-time neat picture of urban mobility patterns. In this paper, we present a visual sensor network in which each node embeds computer vision logics for analyzing in real time urban traffic. The nodes in the network share their perceptions and build a global and comprehensive interpretation of the analyzed scenes in a cooperative and adaptive fashion. This is possible thanks to an especially designed Internet of Things (IoT) compliant middleware which encompasses in-network event composition as well as full support of Machine-2-Machine (M2M) communication mechanism. The potential of the proposed cooperative visual sensor network is shown with two sample applications in urban mobility connected to the estimation of vehicular flows and parking management. Besides providing detailed results of each key component of the proposed solution, the validity of the approach is demonstrated by extensive field tests that proved the suitability of the system in providing a scalable, adaptable and extensible data collection layer for managing and understanding mobility in smart cities.

Detection and localization of gold nanoshells inside cells: near-field approximation.

The optical properties of metal nanoparticles play a fundamental role for their use in a wide range of applications. In hyperthermia treatment, for example, gold nanoshells (NSs, dielectric core+gold shell) pre-embedded in a cancer cell absorb energy when exposed to appropriate wavelengths of a laser beam and heat up, thereby destroying the cancer cell. In this process, nevertheless, healthy tissues (not targeted by the NSs) along the laser path are not affected; this is because most biological soft tissues have a relatively low light absorption coefficient in the near-infrared (NIR) regions-a characteristic known as the tissue optical window. Over such a window, NIR light transmits through the tissues with scattering-limited attenuation and minimal heating, thereby avoiding damage to healthy tissues. As a consequence, the identification of NSs assumed a fundamental role for the further development of such cancer treatment. Recently, we have demonstrated the possibility to identify 100-150 nm diameter gold NSs inside mouse cells using a scanning near-optical microscope (SNOM). In this paper, we provide a numerical demonstration that the SNOM is able to locate NSs inside the cell with a particle-aperture distance of about 100 nm. This result was obtained by developing an analytical approach based on the calculation of the dyadic Green function in the near-field approximation. The implications of our findings will remarkably affect further investigations on the interaction between NSs and biological systems.

A proactive system for maritime environment monitoring.

The ability to remotely detect and monitor oil spills is becoming increasingly important due to the high demand of oil-based products. Indeed, shipping routes are becoming very crowded and the likelihood of oil slick occurrence is increasing. In this frame, a fully integrated remote sensing system can be a valuable monitoring tool. We propose an integrated and interoperable system able to monitor ship traffic and marine operators, using sensing capabilities from a variety of electronic sensors, along with geo-positioning tools, and through a communication infrastructure. Our system is capable of transferring heterogeneous data, freely and seamlessly, between different elements of the information system (and their users) in a consistent and usable form. The system also integrates a collection of decision support services providing proactive functionalities. Such services demonstrate the potentiality of the system in facilitating dynamic links among different data, models and actors, as indicated by the performed field tests.

Decision support in heart failure through processing of electro- and echocardiograms.

OBJECTIVE: Signal and imaging investigations are currently key components in the diagnosis, prognosis and follow up of heart diseases. Nowadays, the need for more efficient, cost-effective and personalised care has led to a renaissance of clinical decision support systems (CDSSs). The purpose of this paper is to present an effective way of achieving a high-level integration of signal and image processing methods in the general process of care, by means of a clinical decision support system, and to discuss the advantages of such an approach. From the wide range of heart diseases, heart failure, whose complexity best highlights the benefits of this integration, has been selected. METHODS: After an analysis of users' needs and expectations, significant and suitably designed image and signal processing algorithms are introduced to objectively and reliably evaluate important features involved in decisional problems in the heart failure domain. Then, a CDSS is conceived so as to combine the domain knowledge with advanced analytical tools for data processing. In particular, the relevant and significant medical knowledge and experts' knowhow are formalised according to an ontological formalism, suitably augmented with a base of rules for inferential reasoning. RESULTS: The proposed methods were tested and evaluated in the daily practice of the physicians operating at the Department of Cardiology, University Magna Graecia, Catanzaro, Italy, on a population of 79 patients. Different scenarios, involving decisional problems based on the analysis of biomedical signals and images, were considered. In these scenarios, after some training and 3 months of use, the CDSS was able to provide important and useful suggestions in routine workflows, by integrating the clinical parameters computed through the developed methods for echocardiographic image segmentation and the algorithms for electrocardiography processing. CONCLUSIONS: The CDSS allows the integration of signal and image processing algorithms into the general process of care. Feedback from end-users has been positive.

Quantification of Epicardial Fat by Cardiac CT Imaging.

The aim of this work is to introduce and design image processing methods for the quantitative analysis of epicardial fat by using cardiac CT imaging.Indeed, epicardial fat has recently been shown to correlate with cardiovascular disease, cardiovascular risk factors and metabolic syndrome. However, many concerns still remain about the methods for measuring epicardial fat, its regional distribution on the myocardium and the accuracy and reproducibility of the measurements.In this paper, a method is proposed for the analysis of single-frame 3D images obtained by the standard acquisition protocol used for coronary calcium scoring. In the design of the method, much attention has been payed to the minimization of user intervention and to reproducibility issues.In particular, the proposed method features a two step segmentation algorithm suitable for the analysis of epicardial fat. In the first step of the algorithm, an analysis of epicardial fat intensity distribution is carried out in order to define suitable thresholds for a first rough segmentation. In the second step, a variational formulation of level set methods - including a specially-designed region homogeneity energy based on Gaussian mixture models- is used to recover spatial coherence and smoothness of fat depots.Experimental results show that the introduced method may be efficiently used for the quantification of epicardial fat.