Safety and efficiency of COVID-19 vaccine in North Africa.

Background In the absence of a specific treatment for COVID-19, preventive measures have been implemented to control this pandemic and vaccination is one of them. However, it is crucial to verify the safety and efficiency of every vaccine. The aim was to determinate the predictive factors of side effects and reinfection after COVID-19 vaccine. Methods A cross-sectional study was conducted in February 2022 among Tunisians infected with COVID-19 between March 2020 and February 2022, using an online self-administered questionnaire. We conducted univariate and multivariate analyses using binary stepwise logistic regression. Results A total of 1541 was selected from 1911 individuals. Comorbidities affected a quarter of the population (22.3%). Before the initial infection, 39.3% had received full vaccination, and 8.7% had received partial vaccination. By February 2022, the majority (82.9%) had received at least two vaccine doses. The reinfection rate was 30.6%. All vaccines prior to the first infection was identified as a protective factor against reinfection. Inactivated virus vaccinations were less likely to induce adverse effects. Conclusion ach vaccine has its own set of advantages and disadvantages: mRNA-based vaccines had a higher incidence of side effects but all vaccines provided better protection against reinfection.

Complementary and alternative medicine in oncology: knowledge, attitude and practice among Tunisian healthcare workers.

PURPOSE: Over the past years, there has been a considerable increase in complementary and alternative medicine (CAM) use among cancer patients. However, guidance from health care workers (HCWs) is not always provided. We aimed to determine the knowledge, attitude and practice of Tunisian HCWs regarding the use of CAM in cancer patients. METHODS: We conducted a multicenter cross-sectional study over 5 months from February to June 2022 among HCWs caring for cancer patients in the Tunisian center region. Data were collected using a self-administered questionnaire developed by our investigators. RESULTS: The level of knowledge about CAM was declared limited by 78.4% of our population. The best known CAM therapies were herbal medicine and homeopathy while chiropractic and hypnosis where the least. HCWs who had sought information on CAM represented 54.3% of our sample and the main source of information was the Internet (37.1%). A positive attitude towards the use of CAM was found in 56% of HCWs. The integration of CAM into supportive care in oncology was approved by 78% of HCWs. Concerning training on CAM, 78% declared its necessity for HCWs and 73.3% expressed a desire to have it. A personal use of CAM was found in 53% of HCWs while 38.8% had previously used CAM in the treatment of their cancer patients. CONCLUSION: The majority of HCWs had a positive attitude towards the use of CAM in oncology despite their poor knowledge about it. Our study emphasizes the need to train HCWs dealing with cancer patients on CAM.

Symptoms and risk factors for long COVID in Tunisian population.

BACKGROUND: The COVID-19 pandemic has presented various challenges, one of which is the discovery that after the acute episode, around 30% of patients experience persistent symptoms or develop new ones, now known as long COVID. This new disease has significant social and financial impacts. The objective is to determine the prevalence of long COVID in the Tunisian population and identify its predictive factors. METHODS: This was a cross-sectional study conducted among Tunisians who were infected with COVID-19 between March 2020 and February 2022. An online self-administered questionnaire was distributed through social media, radio, and television channels over the course of one month (February 2022). Long COVID was defined as the persistence of existing symptoms or the development of new symptoms within three months after onset, lasting for at least two months, and with no differential diagnosis. We performed univariate and multivariate analyses using binary stepwise logistic regression with a significance level set at 5%. RESULTS: A total of 1911 patients participated in our study, and the prevalence of long COVID was 46.5%. The two most frequent categories were general and neurological post-COVID syndrome, with a prevalence of 36.7% each. The most commonly observed symptoms were fatigue (63.7%) and memory problems (49.1%). In the multivariate analysis, the predictive factors for long COVID were female gender and age of 60 years or older, while complete anti-COVID vaccination was found to be a protective factor. CONCLUSIONS: Our study found that complete vaccination was a protective factor against long COVID, while female gender and age of 60 years or older were identified as the main risk factors. These findings are consistent with studies conducted on other ethnic groups. However, many aspects of long COVID remain unclear, including its underlying mechanisms, the identification of which could guide the development of potential effective treatments.

Graph Neural Networks in Network Neuroscience.

Noninvasive medical neuroimaging has yielded many discoveries about the brain connectivity. Several substantial techniques mapping morphological, structural and functional brain connectivities were developed to create a comprehensive road map of neuronal activities in the human brain -namely brain graph. Relying on its non-euclidean data type, graph neural network (GNN) provides a clever way of learning the deep graph structure and it is rapidly becoming the state-of-the-art leading to enhanced performance in various network neuroscience tasks. Here we review current GNN-based methods, highlighting the ways that they have been used in several applications related to brain graphs such as missing brain graph synthesis and disease classification. We conclude by charting a path toward a better application of GNN models in network neuroscience field for neurological disorder diagnosis and population graph integration. The list of papers cited in our work is available at https://github.com/basiralab/GNNs-in-Network-Neuroscience.

Fatal falls from great height in Sousse (Tunisia): Study of 141 medicolegal autopsy cases.

INTRODUCTION: Falls from great heights constitute a violent trauma that can lead to death. This represents a suspicious death, leading to initiate legal proceedings with in particular the practice of an autopsy. AIMS: to determine the features of victims of falls from height and relations between medico-legal form of the death, the height of the fall and the nature of traumatic injuries Methods: A retrospective study about 141 cases of death after fall from great height. Data were collected at the Legal Medicine Department of the Farhat Hached University Hospital in Sousse (Tunisia) over a period of 14 years from 2007 to 2020. RESULTS: The average age of the victims was 37±12.8 years with a sex ratio of 6.05. Half of them were day laborers. The majority had no history of psychiatric illness (91.5%). The majority of victims (41.8%) fell from a height of 3 to 6 meters. Regarding injuries, rib fractures (52.4) were predominant, especially on the right side followed by skull fractures (31.2%). A significant difference in the prevalence of rib cage lesions in the groups over 9 meters in height (p<0.05) was found. The lesions of the lower limbs were proportional to the increase in the height of the fall. Deaths were accidental in 80.8% and suicides in 13.5%. CONCLUSION: In cases of high falls, a forensic autopsy is essential to make a complete evaluation of the injuries, to search a correlation between severity of injuries and height of the fall and finally to orientate towards the medico legal form of the fall.

Patient safety culture in Tunisian maternity care: Analytic study.

INTRODUCTION: Given the potential risks involved in childbirth, patient safety is of utmost importance in maternity care. AIM: To compare the level of patient safety culture between physicians and paramedics in public maternity care structures in Sousse, Tunisia. METHODS: An observational descriptive and cross-sectional study was conducted among health professionals working in all public health maternities of Sousse, Tunisia. A valid Hospital Survey On Patient Safety Culture (HSOPSC) questionnaire was used to gather data, and a score was calculated for each dimension by taking the average of the positive response proportions per item. RESULTS: The global response rate was 86.4%. Paramedics had a higher response rate compared to physicians (90.6% versus 62.1%). The overall scores for the ten dimensions of patient safety culture showed significantly higher scores for physicians compared to paramedics for the dimensions of "Expectations and actions of superiors regarding care safety" and "healthcare professional-patient relationship and safety culture" (88.43% versus 63.73%; p=0.027 and 75.38% versus 65.73%; p=0.041 respectively). Conversely, a significant difference was found in favor of paramedics compared to physicians regarding the dimension of "Management support for care safety" (37.3% versus 13%; p=0.019). CONCLUSION: Significant differences in patient safety culture scores among healthcare professionals. It suggest that efforts should be made to improve management support for care safety for physicians, while paramedics could benefit from increased attention to expectations and actions of superiors regarding care safety and healthcare professional-patient relationship.

Medical students' resilience level and its associated factors: A Tunisian study.

INTRODUCTION: Resilience is one's ability to adapt to internal and external stressors and cope with challenges encountered throughout life. AIM: Our work aimed to determine resilience levels at the Medical University of Ibn El Jazzar-Sousse (Faculty of Medicine of Sousse) Tunisia and to identify the key factors influencing resilience in order to help students improve their college experience, as well as their future career and eventually their quality of life. METHODS: It's a cross-sectional study conducted during October and November 2021 at the FMS including all undergraduate medical students using a questionnaire elaborated in French language and composed of 02 major parts Socio-demographic and general health data and The Connor-Davidson Resilience Scale (CD-RISC). Data were collected using Google Forms platform via social networks (Facebook students' groups). RESULTS: A total of 225 participants filled the questionnaire; the mean age was 21±5 years. Among them 75.1% were females (sex ratio=0.33). The total resilience mean score was 56.36±12.43. Comparison of resilience scores according to different covariates showed that resilience was positively associated with male gender, extracurricular activities, relationships with both colleagues and teachers and physical exercise, but negatively associated with imposed course of study and perception of both study difficulties and personal academic results. No correlation was found between resilience score and age. CONCLUSIONS: This work encourages university administrators to devote more resources to promote resilience, and it emphasizes the importance of implementing new educational and entertaining interventions to improve students' ability to deal with academic challenges.

Risk factors of COVID-19 severity among Tunisian patients in the Emergency Department of Sousse, Tunisia.

INTRODUCTION: Despite the spread of COVID-19 in Tunisia and its impact on people, health and economy, few studies have investigated the profile of COVID-19 Tunisian patients. AIM: Determine the epidemiological, clinical, para-clinical and therapeutic characteristics patients and identify the associated factors of severity. METHODS: This is a retrospective study, conducted among confirmed COVID-19 patients consulting the hospital emergency department. We collected Data using from the patients' computerized files. We performed Data entry and analysis using SPSS 22. RESULTS: We included 375 patients. The average age was 66.7±11.43 years with a sex ratio of 1.6. The most frequent comorbidities were diabetes (100%), hypertension (64.5%), and chronic heart disease (25.9%). The most frequent clinical signs were dyspnea (75.2%), asthenia (66.9%), cough (66.7%) and fever (60.3%). The most frequent biological abnormalities were biological inflammatory syndrome (96%) and elevation of troponin (69.3%). CT scans revealed lung damage in 34.1% of patients. As for treatments, 91.7% received antibiotics, 89% received corticosteroids, 89.3% received anticoagulants, and 85.1% received ventilation (42.6% non-invasive ventilation and 1.9% were intubated). Risk factors of severity were age, chronic heart disease and hypertension. CONCLUSION: Knowing the particularities of Tunisian patients will help to install recommendations to improve the process of care and prevention.

Compassion fatigue among frontline healthcare workers during the covid-19 pandemic in Tunisia.

BACKGROUND: Healthcare workers (HCWs) are highly vulnerable to compassion fatigue (CF), which not only leads to decreased mental and physical health, but also to deterioration in the safety of care delivered. Our study aims to measure compassion satisfaction (CS), CF levels and their predictors among Tunisian HCWs. METHODS: We conducted a cross-sectional study among HCWs caring for confirmed and suspected Covid-19 patients, staff at two university hospitals in Sousse, Tunisia during the 4thwave of coronavirus through a self-administrated Questionnaire, using the French version of the Professional Quality of Life scale ProQol, version 5. RESULTS: A total of 274 professionals were recruited with a mean age of 32.87±8.35 years. HCWs tend to have an overall moderate levels of compassion satisfaction, secondary traumatic stress and burnout with mean scores 35.09±7.08, 29.72±7.62, 28.54±5.44 respectively. Self-reported resilience (ß = 0.14, p = 10-3), work engagement (ß = 0.39, p = 10-3) and burnout (ß = -0.32, p = 10-3) were the predictors of compassion satisfaction in the linear regression analysis (adjusted r2 = 0.45). Similarly, limited work experience, compassion satisfaction and secondary traumatic sub-scores were the determinants of burnout (ß = -0.1, p = 0.04; ß = -0.54, p = 10-3; ß = 0.35, p = 10-3 respectively); (adjusted r2 = 0.48). Regarding STS, female professionals (ß = 0.20, p = 10-3), being married (ß = 0.19, p = 10-3), the fear of transmitting the infection (ß = 0.11, p = 0.03) and burnout (ß = 0.39, p = 10-3) were the predictors for the occurrence of secondary traumatic stress (adjusted r2 = 0.48). CONCLUSION: More resilience promoting interventions and more coping skills programs must be implemented to fulfill HCWs' psychological well-being needs.

Patient safety in medical education: Tunisian students' attitudes.

Health care delivery continues to be unsafe despite major patient safety (PS) improvement efforts over the past decade. Medical school education plays an important role in promoting this culture during initial training. To determine undergraduate medical students' attitudes toward PS at a Tunisian medical school. We carried out a cross-sectional study among undergraduate medical students at Ibn Al Jazzar Medical School in Sousse, Tunisia, using a self-administered questionnaire inspired from the valid tool: Attitudes to Patient Safety Questionnaire (APSQ III). A total of 178 medical students responded to the questionnaire. Medical students tend to have an overall positive perceptions of PS culture with a global mean score 5.33 ± 0.5. Among the individual domains 'Working hours as a cause of error' earned the highest score (6.38 ± 1.0) followed in order by 'Team functioning' (6.24 ± 0.8), 'Error inevitability' (5.91 ± 1.0) and 'Patient involvement in reducing error' (5.50 ± 1.0). The lowest score was for 'Professional incompetence as a cause of error' (4.01 ± 1.0). A PS domain's mean scores comparison based on socio-demographic variables: gender, age, academic year and on PS training revealed a statistically significant difference (p < 0.05) for five PS key dimensions: ' Error reporting confidence ', ' Working hours as a cause of error ', ' Professional incompetence as a cause of error ', ' Team functioning ' and 'PS training received'. Tunisian medical students showed positive attitude towards PS. Nevermore, intensive in terms of frequency and duration sessions, based on various teaching methods may be needed to fulfill students' educational needs.

Hesitation of the Tunisian population to COVID-19 vaccination, and its associated factors.

BACKGROUND: Many people are reluctant to be vaccinated against COVID-19. AIM: To determine the intention to accept COVID19 vaccine and its associated factors among Tunisians. METHODS: We conducted a cross-sectional study among Tunisians from December 2020 to January 2021 using an online questionnaire. Factors associated with intention to accept coronavirus vaccine were analysed using multinomial logistic regression. RESULTS: In total, 169 Tunisians participated in our study. The majority were female (85.2%). The mean age was 48.3 ± 11.8 years. Only 33.1% intended to accept to be vaccinated when COVID-19 vaccine will be available in Tunisia and 22.5% were still hesitant. In multinomial logistic regression, participants having high or very high perceived personal risk of COVID-19 infection (aOR: 3.257, 95% CI : 1.204 - 8.815) were more prone to hesitate to accept COVID-19 vaccine rather than those being willing to accept it. Respondents undergoing seasonal influenza vaccination (aOR: 0.091, 95% CI : 0.019 - 0.433) were less prone to refuse COVID-19 vaccine rather than those being willing to accept it. Young ones aged less than 40 years (aOR: 4.324, 95% CI: 1.180 - 15.843) were more prone to refuse COVID-19 vaccine rather than those being willing to accept it. CONCLUSION: The acceptance rate of coronavirus vaccination was moderate. Therefore, a good communication and health education at a community level are needed.

A New Collaborative Classification Process for Microcalcification Detection Based on Graphs and Knowledge Propagation.

In this paper, we propose a new collaborative process that aims to detect macrocalcifications from mammographic images while minimizing false negative detections. This process is made up of three main phases: suspicious area detection, candidate object identification, and collaborative classification. The main concept is to operate on the entire image divided into homogenous regions called superpixels which are used to identify both suspicious areas and candidate objects. The collaborative classification phase consists in making the initial results of different microcalcification detectors collaborate in order to produce a new common decision and reduce their initial disagreements. The detectors share the information about their detected objects and associated labels in order to refine their initial decisions based on those of the other collaborators. This refinement consists of iteratively updating the candidate object labels of each detector following local and contextual analyses based on prior knowledge about the links between super pixels and macrocalcifications. This process iteratively reduces the disagreement between different detectors and estimates local reliability terms for each super pixel. The final result is obtained by a conjunctive combination of the new detector decisions reached by the collaborative process. The proposed approach is evaluated on the publicly available INBreast dataset. Experimental results show the benefits gained in terms of improving microcalcification detection performances compared to existing detectors as well as ordinary fusion operators.

[Strength of academic motivation among medical students in the Faculty of Medicine of Sousse (Tunisia)].

INTRODUCTION: Motivation is an important component of learning. It remains a complex phenomenon to explore, largely influenced by multiple external and internal factors. It is important to measure the strength of student motivation in a long training course such as medical studies and its influencing factors. AIM: to measure strength of motivation among medical students the Faculty of Medicine of Sousse (FMS). METHODS: It was a cross-sectional study conducted among medical students enrolled at the FMS during the 2021/2022 academic year for 3 months using a questionnaire based on a validated scale: Strength of Motivation for Medical School-Revised (SMMS-R). RESULTS: A total of 185 students participated in the study. The mean age was 20.97 ± 1.8 years. The sex ratio was 0.34. The SMMS-R score was 55[47-63]. This score was higher international students (p=0.029), students who chose medical studies before passing the baccalaureate (p<10-3) and students satisfied with their choice of medical studies (p<10-3). CONCLUSION: Our results revealed a strong association between students' satisfaction and motivation. Thus, the learning environment, governed mainly by institutional rules, educational activities and evaluative practices, greatly influences satisfaction and therefore motivation of medical students.

Toward new multi-wavelets: associated filters and algorithms. Part I: theoretical framework and investigation of biomedical signals, ECG, and coronavirus cases.

Biosignals are nowadays important subjects for scientific researches from both theory, and applications, especially, with the appearance of new pandemics threatening the humanity such as the new coronavirus. One aim in the present work is to prove that wavelets may be a successful machinery to understand such phenomena by applying a step forward extension of wavelets to multi-wavelets. We proposed in a first step to improve multi-wavelet notion by constructing more general families using independent components for multi-scaling and multi-wavelet mother functions. A special multi-wavelet is then introduced, continuous, and discrete multi-wavelet transforms are associated, as well as new filters, and algorithms of decomposition, and reconstruction. Applied breakthroughs of the paper may be summarized in three aims. In a first direction, an approximation (reconstruction) of a classical (stationary, periodic) example dealing with Fourier modes has been conducted in order to confirm the efficiency of the HSch multi-wavelets in approximating such signals and in providing fast algorithms. The second experimentation is concerned with the decomposition and reconstruction application of the HSch multi-wavelet on an ECG signal. The last experimentation is concerned with a de-noising application on a strain of coronavirus signal permitting to localize approximately the transmembrane segments of such a series as neighborhoods of the local maxima of an numerized version of the strain. Accuracy of the method has been evaluated by means of error estimates and statistical tests.

Good hygiene practice application in the private sector, Tunisia.

BACKGROUND: Healthcare-associated infections (HCAIs) occurring outside of health facilities are underestimated because there are a lack of structured preventive organization and absence of epidemiological surveillance. HCAI prevalence is likely to grow with the increase in patient care outside of health institutions. AIMS: To set up a situational analysis of good hygiene practices among private general practitioners (GPs) to better organize HCAI prevention in this sector. METHODS: A descriptive cross-sectional study was conducted between November 2017 and March 2018, using a self-administered questionnaire among all GPs in Sousse City, Tunisia. RESULTS: Participation rate was 93.1%. There was a predominance of male GPs (63%), with a sex ratio of 1.7:1. Up-to-date vaccination status was reported by 82 (75.9%) of GPs. Fifty-six (51.3%) GPs used hydroalcoholic solutions, 13 (12.1%) adopted autoclaving, and 106 (98.1%) wore gloves during invasive care. Blood exposure accidents (BEAs) were reported by 38 (35.2%; declared in 26.3% of cases) and were more prevalent in the group aged > 50 years who used significantly more reusable equipment. BEAs were primarily due to needle-stick injuries (86.8%). CONCLUSION: We identified the priority axes to be considered in organizing HCAI prevention in the private sector, which allows guidance of GPs, avoiding their isolation and compensating for their lack of training and information. This requires willingness and a culture of improving the quality and safety of care in this sector. Committed involvement of several stakeholders at different levels of decision-making in health care is needed.

A new conditional region growing approach for microcalcification delineation in mammograms.

Microcalcifications (MCs) are considered as the first indicator of breast cancer development. Their morphology, in terms of shape and size, is considered as the most important criterion that determines their malignity degrees. Therefore, the accurate delineation of MC is a cornerstone step in their automatic diagnosis process. In this paper, we propose a new conditional region growing (CRG) approach with the ability of finding the accurate MC boundaries starting from selected seed points. The starting seed points are determined based on regional maxima detection and superpixel analysis. The region growing step is controlled by a set of criteria that are adapted to MC detection in terms of contrast and shape variation. These criteria are derived from prior knowledge to characterize MCs and can be divided into two categories. The first one concerns the neighbourhood searching size. The second one deals with the analysis of gradient information and shape evolution within the growing process. In order to prove the effectiveness and the reliability in terms of MC detection and delineation, several experiments have been carried out on MCs of various types, with both qualitative and quantitative analysis. The comparison of the proposed approach with state-of-the art proves the importance of the used criteria in the context of MC delineation, towards a better management of breast cancer. Graphical Abstract Flowchart of the proposed approach.

Brain multigraph prediction using topology-aware adversarial graph neural network.

Brain graphs (i.e, connectomes) constructed from medical scans such as magnetic resonance imaging (MRI) have become increasingly important tools to characterize the abnormal changes in the human brain. Due to the high acquisition cost and processing time of multimodal MRI, existing deep learning frameworks based on Generative Adversarial Network (GAN) focused on predicting the missing multimodal medical images from a few existing modalities. While brain graphs help better understand how a particular disorder can change the connectional facets of the brain, synthesizing a target brain multigraph (i.e, multiple brain graphs) from a single source brain graph is strikingly lacking. Additionally, existing graph generation works mainly learn one model for each target domain which limits their scalability in jointly predicting multiple target domains. Besides, while they consider the global topological scale of a graph (i.e., graph connectivity structure), they overlook the local topology at the node scale (e.g., how central a node is in the graph). To address these limitations, we introduce topology-aware graph GAN architecture (topoGAN), which jointly predicts multiple brain graphs from a single brain graph while preserving the topological structure of each target graph. Its three key innovations are: (i) designing a novel graph adversarial auto-encoder for predicting multiple brain graphs from a single one, (ii) clustering the encoded source graphs in order to handle the mode collapse issue of GAN and proposing a cluster-specific decoder, (iii) introducing a topological loss to force the prediction of topologically sound target brain graphs. The experimental results using five target domains demonstrated the outperformance of our method in brain multigraph prediction from a single graph in comparison with baseline approaches.

Severity factors in Algerian patients hospitalized for COVID-19.

INTRODUCTION: Determining the profile of COVID-19 patients with low pulsed hemoglobin saturation in oxygen (SpO2) could help clinicians identify those with a poor prognosis. AIM: To identify and to compare the clinical, biological and radiological data of Algerian patients hospitalized for COVID-19 and divided according to the SpO2 measured at admission, at rest, and in ambient air. METHODS: A prospective study was carried out on Algerian patients hospitalized for COVID-19 during the period from March 9 to April 30, 2020. The general characteristics of the patients and the clinical, biological and radiological data were determined. RESULTS: 86 patients were included in the study [G1: SpO2 >95% (n=51) and G2: SpO2 ≤95% (n=35)]. Compared to G1, G2 was older (48±14 vs. 61±12 years, p=0.0001), included more patients aged ≥ 50 years (37.2 vs. 80.0%, p=0.0001), having an arterial-hypertension (21.6 vs. 45.7%, p=0.0180), a cancer (0.0 vs. 14.3%, p=0.0054), an anemia (25.6 vs. 56.3%, p=0.0069), a leukocytosis (4.7 vs. 21.9%, p=0.0236), a biological inflammatory syndrome (82.5 vs. 100%, p=0.0142), a hyper-uremia (7.0 vs. 37.5%, p=0.0185), a hyper-creatininaemia (4.7 vs. 18.8%, p=0.0356), a tissue damage (41.0 vs. 66.7%, p=0.0341), a diffuse ground-glass opacity (52.0 vs. 71.4%, p=0.0397), band condensations (30.0 vs. 54.3%, p=0.0244), a severe extension (2.0 vs. 25.7%, p=0.0008), and included fewer patients who complained from diarrhea (49.0 vs. 22.9%, p=0.0145), having a nodular ground-glass (66.0 vs. 40.0%, p=0.0177) and a slight extension (78.0 vs. 40.0%, p=0.0004). CONCLUSION: Criteria associated with low SpO2 in hospitalized COVID-19 patients were advanced age, a history of arterial-hypertension and cancer, high frequencies of certain biological abnormalities or radiological signs. The diarrhea symptom, the radiological appearance of nodular ground glass, and a slight extension of the radiological lesions appear as protective elements.

Brain graph synthesis by dual adversarial domain alignment and target graph prediction from a source graph.

Developing predictive intelligence in neuroscience for learning how to generate multimodal medical data from a single modality can improve neurological disorder diagnosis with minimal data acquisition resources. Existing deep learning frameworks are mainly tailored for images, which might fail in handling geometric data (e.g., brain graphs). Specifically, predicting a target brain graph from a single source brain graph remains largely unexplored. Solving such problem is generally challenged with domain fracturecaused by the difference in distribution between source and target domains. Besides, solving the prediction and domain fracture independently might not be optimal for both tasks. To address these challenges, we unprecedentedly propose a Learning-guided Graph Dual Adversarial Domain Alignment (LG-DADA) framework for predicting a target brain graph from a source brain graph. The proposed LG-DADA is grounded in three fundamental contributions: (1) a source data pre-clustering step using manifold learning to firstly handle source data heterogeneity and secondly circumvent mode collapse in generative adversarial learning, (2) a domain alignment of source domain to the target domain by adversarially learning their latent representations, and (3) a dual adversarial regularization that jointly learns a source embedding of training and testing brain graphs using two discriminators and predict the training target graphs. Results on morphological brain graphs synthesis showed that our method produces better prediction accuracy and visual quality as compared to other graph synthesis methods.

Brain graph super-resolution for boosting neurological disorder diagnosis using unsupervised multi-topology connectional brain template learning.

Existing graph analysis techniques generally focus on decreasing the dimensionality of graph data (i.e., removing nodes, edges, or both) in diverse predictive learning tasks in pattern recognition, computer vision, and medical data analysis such as dimensionality reduction, filtering and embedding techniques. However, graph super-resolution is strikingly lacking, i.e., the concept of super-resolving low-resolution (LR) graphs with nr nodes into high-resolution graphs (HR) with [Formula: see text] nodes. Particularly, learning how to automatically generate HR brain connectomes, without resorting to the computationally expensive MRI processing steps such as image registration and parcellation, remains unexplored. To fill this gap, we propose the first technique to super-resolve undirected fully connected graphs with application to brain connectomes. First, we root our brain graph super-resolution (BGSR) framework in learning how to estimate a centered LR population-based brain graph representation, coined as connectional brain template (CBT), acting as a proxy in the target BGSR task. Specifically, we hypothesize that the estimation of a well-representative and centered CBT would help better capture the individuality of each LR brain graph via its residual distance from the population-based CBT. This will eventually allow an accurate identification of the most similar individual graphs to a new testing graph in the LR domain for the target prediction task. Second, we leverage the estimated LR CBT (i.e., population mean) to derive residual LR brain graphs, capturing the deviation of all subjects from the estimated CBT. Third, we learn multi-topology LR graph manifolds using different graph topological measurements (e.g., degree, closeness, betweenness) by estimating residual LR similarity matrices modeling the relationship between pairs of residual LR graphs. These are then fused so we can effectively identify for each testing LR subject its most K similar training LR graphs. Last, the missing testing HR graph is predicted by averaging the HR graphs of the K selected training subjects. Predicted HR from LR functional brain graphs boosted classification results for autistic subjects by 16.48% compared with LR functional graphs.