Computer vision in aseptic pharmaceutical manufacturing: a deep learning-based approach to vial inspection and pack formation during a lyophilizer loading cycle

Vision systems are powerful tools playing an increasingly important role in modern industry, to detect errors and maintain product standards. With the enlarged availability of affordable industrial cameras, computer vision algorithms have been increasingly applied in industrial manufacturing processes monitoring. Until a few years ago, industrial computer vision applications relied only on ad-hoc algorithms designed for the specific object and acquisition setup being monitored, with a strong focus on co-designing the acquisition and processing pipeline. Deep learning has overcome these limits providing greater flexibility and faster re-configuration. In this work, the process to be inspected consists in vials’ pack formation entering a freeze-dryer, which is a common scenario in pharmaceutical active ingredient packaging lines. To ensure that the machine produces proper packs, a vision system is installed at the entrance of the freeze-dryer to detect eventual anomalies with execution times compatible with the production specifications. Other constraints come from sterility and safety standards required in pharmaceutical manufacturing. This work presents an overview about the production line, with particular focus on the vision system designed, and about all trials conducted to obtain the final performance. Transfer learning, alleviating the requirement for a large number of training data, combined with data augmentation methods, consisting in the generation of synthetic images, were used to effectively increase the performances while reducing the cost of data acquisition and annotation. The proposed vision algorithm is composed by two main subtasks, designed respectively to vials counting and discrepancy detection. The first one was trained on more than 23k vials (about 300 images) and tested on 5k more (about 75 images), whereas 60 training images and 52 testing images were used for the second one.

Analisi di sostenibilità di tecnologie CCS applicate a settori hard-to-abate

In questo lavoro di tesi viene presentata un’analisi di sostenibilità economica, ambientale e sociale di un sistema di cattura e purificazione della CO2 post-combustione. Al fine di ottenere i dati utili al calcolo di indici significativi di sostenibilità, si è scelto di modellare, attraverso il software Aspen Plus®, un processo di assorbimento chimico con ammine, tecnologia matura e spesso impiegata a questo scopo per settori industriali di taglie superiori. Al sistema di assorbimento viene poi fatta seguire un’unità di compressione e purificazione, finalizzata al raggiungimento di specifiche condizioni della corrente in uscita, legate alla successiva tipologia di trasporto considerata: gas pressurizzato via pipeline o gas liquefatto via nave. L’analisi economica si è basata sul calcolo di costi operativi, costi variabili, costo della CO2 evitata e consumo specifico di energia (SPECCA). La simulazione ha infatti consentito di ottenere un dimensionamento di massima delle apparecchiature e il consumo di utenze, valori necessari alla stima degli indici. Gli indici economici calcolati sono poi stati utilizzati per comprendere quale parte del processo fosse la più critica e che impatto avessero le unità di compressione e purificazione rispetto alla presenza del solo impianto di cattura. Analisi simili sono state svolte anche in termini di sostenibilità ambientale e sociale. La prima è stata svolta calcolando l’effettiva rimozione di CO2 del sistema, tenendo in considerazione le emissioni dirette e indirette connesse al funzionamento degli apparati. L’analisi sociale è stata ridotta a un’analisi qualitativa della sicurezza, attuata attraverso il calcolo dell’Indice di sicurezza intrinseca.

Motor Unit Spike Trains Recontruction from sEMG Signals for Gesture Classification on a Low-Power Processing Platform

Hand gesture recognition based on surface electromyography (sEMG) signals is a promising approach for the development of intuitive human-machine interfaces (HMIs) in domains such as robotics and prosthetics. The sEMG signal arises from the muscles' electrical activity, and can thus be used to recognize hand gestures. The decoding from sEMG signals to actual control signals is non-trivial; typically, control systems map sEMG patterns into a set of gestures using machine learning, failing to incorporate any physiological insight. This master thesis aims at developing a bio-inspired hand gesture recognition system based on neuromuscular spike extraction rather than on simple pattern recognition. The system relies on a decomposition algorithm based on independent component analysis (ICA) that decomposes the sEMG signal into its constituent motor unit spike trains, which are then forwarded to a machine learning classifier. Since ICA does not guarantee a consistent motor unit ordering across different sessions, 3 approaches are proposed: 2 ordering criteria based on firing rate and negative entropy, and a re-calibration approach that allows the decomposition model to retain information about previous sessions. Using a multilayer perceptron (MLP), the latter approach results in an accuracy up to 99.4% in a 1-subject, 1-degree of freedom scenario. Afterwards, the decomposition and classification pipeline for inference is parallelized and profiled on the PULP platform, achieving a latency < 50 ms and an energy consumption < 1 mJ. Both the classification models tested (a support vector machine and a lightweight MLP) yielded an accuracy > 92% in a 1-subject, 5-classes (4 gestures and rest) scenario. These results prove that the proposed system is suitable for real-time execution on embedded platforms and also capable of matching the accuracy of state-of-the-art approaches, while also giving some physiological insight on the neuromuscular spikes underlying the sEMG.

Analisi prestazionale della ricerca di vicini in uno spazio vettoriale con Elasticsearch su infrastruttura cloud

Negli ultimi anni, a causa degli enormi progressi dell’informatica e della sempre crescente quantità di dati generati, si è sentito sempre più il bisogno di trovare nuove tecniche, approcci e algoritmi per la ricerca dei dati. Infatti, la quantità di informazioni da memorizzare è diventata tale che ormai si sente sempre più spesso parlare di "Big Data". Questo nuovo scenario ha reso sempre più inefficaci gli approcci tradizionali alla ricerca di dati. Recentemente sono state quindi proposte nuove tecniche di ricerca, come ad esempio le ricerche Nearest Neighbor. In questo elaborato sono analizzate le prestazioni della ricerca di vicini in uno spazio vettoriale utilizzando come sistema di data storage Elasticsearch su un’infrastruttura cloud. In particolare, sono stati analizzati e messi a confronto i tempi di ricerca delle ricerche Nearest Neighbor esatte e approssimate, valutando anche la perdita di precisione nel caso di ricerche approssimate, utilizzando due diverse metriche di distanza: la similarità coseno e il prodotto scalare.

Experimental characterization of hydrogen containing flames

The current environmental and socio-economic situation promotes the development of carbon-neutral and sustainable solutions for energy supply. In this framework, the use of hydrogen has been largely indicated as a promising alternative. However, safety aspects are of concern for storage and transportation technologies. Indeed, the current know-how promotes its transportation via pipeline as compressed gas. However, the peculiar properties of hydrogen make the selection of suitable materials challenging. For these reasons, dilution with less reactive species has been considered a short and medium solution. As a way of example, methane-hydrogen mixtures are currently transported via pipelines. In this case, the hydrogen content is limited to 20% in volume, thus keeping the dependence on natural gas sources. On the contrary, hydrogen can be conveniently transported by mixing it with carbon dioxide deriving from carbon capture and storage technologies. In this sense, the interactions between hydrogen and carbon dioxide have been poorly studied. In particular, the effects of composition and operative conditions in the case of accidental release or for direct use in the energy supply chain are unknown. For these reasons, the present work was devoted to the characterization of the chemical phenomena ruling the system. To this aim, laminar flames containing hydrogen and carbon dioxide in the air were investigated experimentally and numerically. Different detailed kinetic mechanisms largely validated were considered at this stage. Significant discrepancies were observed among numerical and experimental data, especially once a fuel consisting of 40%v of hydrogen was studied. This deviation was attributed to the formation of a cellular flame increasing the overall reactivity. Hence, this observation suggests the need for combined models accounting for peculiar physical phenomena and detailed kinetic mechanisms characterizing the hydrogen-containing flames.

ICE Emissions Estimation through Machine-learning techniques to support real-world R&D experiments

The emissions estimation, both during homologation and standard driving, is one of the new challenges that automotive industries have to face. The new European and American regulation will allow a lower and lower quantity of Carbon Monoxide emission and will require that all the vehicles have to be able to monitor their own pollutants production. Since numerical models are too computationally expensive and approximated, new solutions based on Machine Learning are replacing standard techniques. In this project we considered a real V12 Internal Combustion Engine to propose a novel approach pushing Random Forests to generate meaningful prediction also in extreme cases (extrapolation, very high frequency peaks, noisy instrumentation etc.). The present work proposes also a data preprocessing pipeline for strongly unbalanced datasets and a reinterpretation of the regression problem as a classification problem in a logarithmic quantized domain. Results have been evaluated for two different models representing a pure interpolation scenario (more standard) and an extrapolation scenario, to test the out of bounds robustness of the model. The employed metrics take into account different aspects which can affect the homologation procedure, so the final analysis will focus on combining all the specific performances together to obtain the overall conclusions.

Application of active and transfer learning to wound image segmentation

Wound management is a fundamental task in standard clinical practice. Automated solutions already exist for humans, but there is a lack of applications on wound management for pets. The importance of a precise and efficient wound assessment is helpful to improve diagnosis and to increase the effectiveness of treatment plans for the chronic wounds. The goal of the research was to propose an automated pipeline capable of segmenting natural light-reflected wound images of animals. Two datasets composed by light-reflected images were used in this work: Deepskin dataset, 1564 human wound images obtained during routine dermatological exams, with 145 manual annotated images; Petwound dataset, a set of 290 wound photos of dogs and cats with 0 annotated images. Two implementations of U-Net Convolutioal Neural Network model were proposed for the automated segmentation. Active Semi-Supervised Learning techniques were applied for human-wound images to perform segmentation from 10% of annotated images. Then the same models were trained, via Transfer Learning, adopting an Active Semi- upervised Learning to unlabelled animal-wound images. The combination of the two training strategies proved their effectiveness in generating large amounts of annotated samples (94% of Deepskin, 80% of PetWound) with the minimal human intervention. The correctness of automated segmentation were evaluated by clinical experts at each round of training thus we can assert that the results obtained in this thesis stands as a reliable solution to perform a correct wound image segmentation. The use of Transfer Learning and Active Semi-Supervied Learning allows to minimize labelling effort from clinicians, even requiring no starting manual annotation at all. Moreover the performances of the model with limited number of parameters suggest the implementation of smartphone-based application to this topic, helping the future standardization of light-reflected images as acknowledge medical images.