Graph-based analysis of brain resting-state fMRI data in nocturnal frontal lobe epileptic patients

Il lavoro che ho sviluppato presso l'unità di RM funzionale del Policlinico S.Orsola-Malpighi, DIBINEM, è incentrato sull'analisi dati di resting state - functional Magnetic Resonance Imaging (rs-fMRI) mediante l'utilizzo della graph theory, con lo scopo di valutare eventuali differenze in termini di connettività cerebrale funzionale tra un campione di pazienti affetti da Nocturnal Frontal Lobe Epilepsy (NFLE) ed uno di controlli sani. L'epilessia frontale notturna è una peculiare forma di epilessia caratterizzata da crisi che si verificano quasi esclusivamente durante il sonno notturno. Queste sono contraddistinte da comportamenti motori, prevalentemente distonici, spesso complessi, e talora a semiologia bizzarra. L'fMRI è una metodica di neuroimaging avanzata che permette di misurare indirettamente l'attività neuronale. Tutti i soggetti sono stati studiati in condizioni di resting-state, ossia di veglia rilassata. In particolare mi sono occupato di analizzare i dati fMRI con un approccio innovativo in campo clinico-neurologico, rappresentato dalla graph theory. I grafi sono definiti come strutture matematiche costituite da nodi e links, che trovano applicazione in molti campi di studio per la modellizzazione di strutture di diverso tipo. La costruzione di un grafo cerebrale per ogni partecipante allo studio ha rappresentato la parte centrale di questo lavoro. L'obiettivo è stato quello di definire le connessioni funzionali tra le diverse aree del cervello mediante l'utilizzo di un network. Il processo di modellizzazione ha permesso di valutare i grafi neurali mediante il calcolo di parametri topologici che ne caratterizzano struttura ed organizzazione. Le misure calcolate in questa analisi preliminare non hanno evidenziato differenze nelle proprietà globali tra i grafi dei pazienti e quelli dei controlli. Alterazioni locali sono state invece riscontrate nei pazienti, rispetto ai controlli, in aree della sostanza grigia profonda, del sistema limbico e delle regioni frontali, le quali rientrano tra quelle ipotizzate essere coinvolte nella fisiopatologia di questa peculiare forma di epilessia.

Graph-based Vision-Language Model for Multi-Modal Retrieval of Fashion Products

Artificial Intelligence is reshaping the field of fashion industry in different ways. E-commerce retailers exploit their data through AI to enhance their search engines, make outfit suggestions and forecast the success of a specific fashion product. However, it is a challenging endeavour as the data they possess is huge, complex and multi-modal. The most common way to search for fashion products online is by matching keywords with phrases in the product's description which are often cluttered, inadequate and differ across collections and sellers. A customer may also browse an online store's taxonomy, although this is time-consuming and doesn't guarantee relevant items. With the advent of Deep Learning architectures, particularly Vision-Language models, ad-hoc solutions have been proposed to model both the product image and description to solve this problems. However, the suggested solutions do not exploit effectively the semantic or syntactic information of these modalities, and the unique qualities and relations of clothing items. In this work of thesis, a novel approach is proposed to address this issues, which aims to model and process images and text descriptions as graphs in order to exploit the relations inside and between each modality and employs specific techniques to extract syntactic and semantic information. The results obtained show promising performances on different tasks when compared to the present state-of-the-art deep learning architectures.

Graph-based User Scheduling for MIMO LEO-based Satellite Communication Systems

The study of the user scheduling problem in a Low Earth Orbit (LEO) Multi-User MIMO system is the objective of this thesis. With the application of cutting-edge digital beamforming algorithms, a LEO satellite with an antenna array and a large number of antenna elements can provide service to many user terminals (UTs) in full frequency reuse (FFR) schemes. Since the number of UTs on-ground are many more than the transmit antennas on the satellite, user scheduling is necessary. Scheduling can be accomplished by grouping users into different clusters: users within the same cluster are multiplexed and served together via Space Division Multiple Access (SDMA), i.e., digital beamforming or Multi-User MIMO techniques; the different clusters of users are then served on different time slots via Time Division Multiple Access (TDMA). The design of an optimal user grouping strategy is known to be an NP-complete problem which can be solved only through exhaustive search. In this thesis, we provide a graph-based user scheduling and feed space beamforming architecture for the downlink with the aim of reducing user inter-beam interference. The main idea is based on clustering users whose pairwise great-circle distance is as large as possible. First, we create a graph where the users represent the vertices, whereas an edge in the graph between 2 users exists if their great-circle distance is above a certain threshold. In the second step, we develop a low complex greedy user clustering technique and we iteratively search for the maximum clique in the graph, i.e., the largest fully connected subgraph in the graph. Finally, by using the 3 aforementioned power normalization techniques, a Minimum Mean Square Error (MMSE) beamforming matrix is deployed on a cluster basis. The suggested scheduling system is compared with a position-based scheduler, which generates a beam lattice on the ground and randomly selects one user per beam to form a cluster.

A study of brain functional connectivity and cognitive processes in long COVID-19 patients

In this work, a prospective study conducted at the IRCCS Istituto delle Scienze Neurologiche di Bologna is presented. The aim was to investigate the brain functional connectivity of a cohort of patients (N=23) suffering from persistent olfactory dysfunction after SARS-CoV-2 infection (Post-COVID-19 syndrome), as compared to a matching group of healthy controls (N=26). In particular, starting from individual resting state functional-MRI data, different analytical approaches were adopted in order to find potential alterations in the connectivity patterns of patients’ brains. Analyses were conducted both at a whole-brain level and with a special focus on brain regions involved in the processing of olfactory stimuli (Olfactory Network). Statistical correlations between functional connectivity alterations and the results of olfactory and neuropsychological tests were investigated, to explore the associations with cognitive processes. The three approaches implemented for the analysis were the seed-based correlation analysis, the group-level Independent Component analysis and a graph-theoretical analysis of brain connectivity. Due to the relative novelty of such approaches, many implementation details and methodologies are not standardized yet and represent active research fields. Seed-based and group-ICA analyses’ results showed no statistically significant differences between groups, while relevant alterations emerged from those of the graph-based analysis. In particular, patients’ olfactory sub-graph appeared to have a less pronounced modular structure compared to the control group; locally, a hyper-connectivity of the right thalamus was observed in patients, with significant involvement of the right insula and hippocampus. Results of an exploratory correlation analysis showed a positive correlation between the graphs global modularity and the scores obtained in olfactory tests and negative correlations between the thalamus hyper-connectivity and memory tests scores.

Knowledge-based Chord Embeddings

This thesis develops AI methods as a contribution to computational musicology, an interdisciplinary field that studies music with computers. In systematic musicology a composition is defined as the combination of harmony, melody and rhythm. According to de La Borde, harmony alone "merits the name of composition". This thesis focuses on analysing the harmony from a computational perspective. We concentrate on symbolic music representation and address the problem of formally representing chord progressions in western music compositions. Informally, chords are sets of pitches played simultaneously, and chord progressions constitute the harmony of a composition. Our approach combines ML techniques with knowledge-based techniques. We design and implement the Modal Harmony ontology (MHO), using OWL. It formalises one of the most important theories in western music: the Modal Harmony Theory. We propose and experiment with different types of embedding methods to encode chords, inspired by NLP and adapted to the music domain, using both statistical (extensional) knowledge by relying on a huge dataset of chord annotations (ChoCo), intensional knowledge by relying on MHO and a combination of the two. The methods are evaluated on two musicologically relevant tasks: chord classification and music structure segmentation. The former is verified by comparing the results of the Odd One Out algorithm to the classification obtained with MHO. Good performances (accuracy: 0.86) are achieved. We feed a RNN for the latter, using our embeddings. Results show that the best performance (F1: 0.6) is achieved with embeddings that combine both approaches. Our method outpeforms the state of the art (F1 = 0.42) for symbolic music structure segmentation. It is worth noticing that embeddings based only on MHO almost equal the best performance (F1 = 0.58). We remark that those embeddings only require the ontology as an input as opposed to other approaches that rely on large datasets.

Machine Learning methods for hepatocellular malignancies segmentation and MVI prediction

The aim of this thesis project is to automatically localize HCC tumors in the human liver and subsequently predict if the tumor will undergo microvascular infiltration (MVI), the initial stage of metastasis development. The input data for the work have been partially supplied by Sant'Orsola Hospital and partially downloaded from online medical databases. Two Unet models have been implemented for the automatic segmentation of the livers and the HCC malignancies within it. The segmentation models have been evaluated with the Intersection-over-Union and the Dice Coefficient metrics. The outcomes obtained for the liver automatic segmentation are quite good (IOU = 0.82; DC = 0.35); the outcomes obtained for the tumor automatic segmentation (IOU = 0.35; DC = 0.46) are, instead, affected by some limitations: it can be state that the algorithm is almost always able to detect the location of the tumor, but it tends to underestimate its dimensions. The purpose is to achieve the CT images of the HCC tumors, necessary for features extraction. The 14 Haralick features calculated from the 3D-GLCM, the 120 Radiomic features and the patients' clinical information are collected to build a dataset of 153 features. Now, the goal is to build a model able to discriminate, based on the features given, the tumors that will undergo MVI and those that will not. This task can be seen as a classification problem: each tumor needs to be classified either as “MVI positive” or “MVI negative”. Techniques for features selection are implemented to identify the most descriptive features for the problem at hand and then, a set of classification models are trained and compared. Among all, the models with the best performances (around 80-84% ± 8-15%) result to be the XGBoost Classifier, the SDG Classifier and the Logist Regression models (without penalization and with Lasso, Ridge or Elastic Net penalization).

Deep learning based style transfer for low altitude aerial imagery

Unmanned Aerial Vehicle (UAVs) equipped with cameras have been fast deployed to a wide range of applications, such as smart cities, agriculture or search and rescue applications. Even though UAV datasets exist, the amount of open and quality UAV datasets is limited. So far, we want to overcome this lack of high quality annotation data by developing a simulation framework for a parametric generation of synthetic data. The framework accepts input via a serializable format. The input specifies which environment preset is used, the objects to be placed in the environment along with their position and orientation as well as additional information such as object color and size. The result is an environment that is able to produce UAV typical data: RGB image from the UAVs camera, altitude, roll, pitch and yawn of the UAV. Beyond the image generation process, we improve the resulting image data photorealism by using Synthetic-To-Real transfer learning methods. Transfer learning focuses on storing knowledge gained while solving one problem and applying it to a different - although related - problem. This approach has been widely researched in other affine fields and results demonstrate it to be an interesing area to investigate. Since simulated images are easy to create and synthetic-to-real translation has shown good quality results, we are able to generate pseudo-realistic images. Furthermore, object labels are inherently given, so we are capable of extending the already existing UAV datasets with realistic quality images and high resolution meta-data. During the development of this thesis we have been able to produce a result of 68.4% on UAVid. This can be considered a new state-of-art result on this dataset.

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.