Embedded artificial intelligence (EAI)-based prognostics
The goal of this thesis is to reduce the clinical risk in surgical procedures by implementing an embedded artificial intelligence (EAI) technique for helping the surgeon in prognostics by assessing the quality of the outcome of the surgical procedure.
As a case study, in this work, we considered the AI-based automatic evaluation of the quality of the suture of the gastrointestinal tract, based on the vascularization analysis applied to abdominal laparoscopic surgery. The system has been already tested during a surgical procedure: results showed that the developed algorithm successfully identifies, in real time, well- and low-vascularized tracts.
Further research is dedicated to implement different thresholds levels to quantify the success of the procedure.
Augmented Reality-based telementoring
The goal of this research activity is to design and develop a complete set of functional and efficient software and hardware system for telementoring, to distribute to every small centre without much experience in correcting complex pathologies in pediatric surgery, with a focus on minimally invasive surgery in urology and oncology.
The objective is to include advanced interaction techniques and tools to remotely guide the surgeon, not only ask for a second opinion: the remote surgeon could use his laparoscopic tools (tracked) in a simulator, or an ad hoc devised haptic tool, to show on the mentee display exactly how to intervene and with which movements.
Augmented Reality for anesthetist in surgery
In this research work, an augmented-reality (AR) system for monitoring patient’s vitals in real time during surgical procedures is proposed and metrologically characterized in terms of transmission error rates and latency. These specifications, in fact, are crucial to ensure real-time response which is a critical requirement for Health applications. The proposed system automatically acquires the vitals from the operating room instrumentation, and shows them in real time directly on a set of wearable AR glasses. In this way, the surgical team has a real-time visualization of a comprehensive set of patient’s information, without constantly looking at the instrumentation. The goal of this work is to design (taking into account the metrologocial characteristics) and implement a system was designed to ensure modularity, flexibility, ease of use and, most importantly, a reliable communication.
Metabolic simulator based on neural networks and nutritional factors for artificial pancreas
When a person is affected by type 1 diabetes mellitus, the lack of insulin requires to introduce in the body the correct amount of insulin particularly after meals, when blood glucose typically increases. Many algorithms in the literature have been used to predict post-prandial insulin boluses, but they only consider meal carbohydrates (which are the main nutritional factor influencing postprandial glucose response). The aim of this work is to improve the accuracy of the prediction also considering the contribution of other nutritional factors such as, for example, sugars, proteins, lipids and fats. In fact, all these factors are expected to affect the glycemic response in different ways. Based on these considerations, the goal of this study is to develop an artificial intelligence algorithm to predict post-prandial glycemia levels taking into account not only nutritional factors but also glycemia before the meal and infused insulin.
Impedance Spectroscopy for drug absorption measurement
The research activity aims to develop innovative methodologies for assessing the efficacy of topical and transdermal administration of drugs. At present, shared and recognized in vivo assessment methods are missing for topical and transdermal drugs. The impedance spectroscopy applied to biological tissue is proposed as quantitative method for assessing the dosage. The research will also aim at developing a method for assessing the rate of drug bioavailability to provide immediate evidence on the rapid systemic absorption of a drug (e.g., diabetic patient receiving ultra-fast insulin). In this research, biological tissue models will be realized (e.g. by finite element simulations) and validated through experimental campaign. Different kinds of drugs will be classified according their electrical behavior. Algorithms will be created for assessing concentration and localization of a drug by elaborating the variation of the impedance spectrum in a biological tissue before and after administration. Portable and wearable devices able to quantify and localize a drug administered topically or transdermally delivered will be prototyped.
EEG-based emotion, attention and engagement measurement
Carrying out rehabilitation exercises, while maintaining the attention focus in a sustained and selective way, promotes neuronal plasticity and motor learning. The attention to the motor task has an enhanced effect on rehabilitation performance output. A distraction detection system allows the therapist or an automated system to know when to stimulate the patient’s attention for enhancing the therapy effectiveness.
Discrimination of emotional valence is a broad issue widely addressed in recent decades, affecting the most varied sectors and finding application in multiple domains. Some application fields are for example, car driving, working, medicine, and entertainment.
Engagement assessment is fundamental in clinical practice to personalize a treatment and improve its effectiveness. Int he last years, scientific literature addressed particular attention to engagement.
Several bio-signals have been studied over the years for emotions, attention and engagement recognition. In recent years, several studies has focused on the brain signal, in particular on electroencephalographic (EEG) signals.
Based on these considerations, the goal of this research is prototyping and/or experimental validating EEG-based system (hardware and software) for measuring emotion, attention and engagement.
Artifacts detection and removal in wearable BCI
Brain-computer interfaces are recently attracting more and more investments from the technological and scientific community. However, the path toward a daily-life system has still several obstacles. Many effort have been made to properly decode neural activity and understand users intentions by mainly focusing on processing algorithms. Nonetheless, it was rightfully assumed that brain activity could be properly recorded. Although this is true for clinical setups, in which wet electrodes (with conductive gels) are employed, data quality is a major issue in wearable brain-computer interfaces targeting daily-life applications. In such a case, dry electrodes are typically considered for user-friendliness, and this poses some issues in terms of setup stability. To overcome this limits, a proper setup must be adopted. Electroencephalography is generally exploited to guarantee low cost, wearability, and portability of the setup. Nevertheless, setup stability must be guaranteed as well as robust processing algorithms must be developed in order to detect and remove unavoidable artifacts. This thesis project thus aims to overcome these metrological issues in order to reach high data quality for further analyses.
Neurofeedback in motor imagery-based brain-computer interfaces
A brain-computer interface offers a mean for communication and control that is alternative to normal brain pathways. Among the several paradigms under study, motor imagery-based BCIs (MI-BCI) rely on spontaneous sensorimotor rhythms and do not need for external stimuli in eliciting the brain activity to be measured. In this context, neurofeedback is essential in helping the user to better focus the mental task, and hence guarantee proper functionality of the system. Many BCI technologies rely on vision for their functionality. The need to investigate further feedback paradigms arises in aiming to create a more immersive experience, which would lead to a stronger engagement of the user. In particular, haptic feedback is of great interest because it is claimed that it could naturally close the sensorimotor loop.
Therefore, in this project the brain activity will be measured by means of electroencephalography (EEG), with few dry electrodes placed around the sensorimotor area and neurofeedback will be provided in real-time in accordance with the measured biosignals. Attention will be given to the algorithm for online classification of EEG signals, in which the information about sensorimotor rhythms is encoded. The proposed technology could find interesting applications in rehabilitation, gaming, and even in robotics.