Episode 24: Inside the Future Brain: AI, Neurotechnology, Robotics and Beyond

00:00:10: Welcome to another episode of PhD BNM Beyond the Lab.

00:00:14: Today we'll explore the fascinating world of different disciplines which are converging toward a common goal, understanding and interacting with the brain in increasingly precise and innovative ways.

00:00:25: Let's begin!

00:00:40: Hi everyone, my name is Leonardo Politi.

00:00:41: I'm a PhD student at the Department of Biomedical and Rheumatoid Sciences at The University of Poland.

00:00:48: My research focuses on brain-computer interfaces or BCIs specifically designed for motor rehabilitation.

00:00:54: And the core idea's quite simple after a neurological injury Or if you lose a limb... ...the brain still knows what it wants to do and still generates motor commands.

00:01:05: But this signal cannot reach the muscle effectively anymore.

00:01:09: So what we are trying to do is decoding, um... What the brain is trying to say and use this signal and this information To control prosthetics and communication devices.

00:01:20: And one major step of my project actually consists in using This signal for controlling an electrical stimulator Which activates a patient muscle directly from its own intention?

00:01:31: And this connects his brain and his muscles again In close loop which has been shown to be very promising for rehabilitation and for neural plasticity.

00:01:41: So that's a very fascinating, but how does the machine actually know what movement you're thinking about?

00:01:49: Every time you intend to move... For example if think at opening your hand Your brain produces specific electric patterns Which can be recorded using electrodes.

00:01:58: There are mainly two approaches for doing this.

00:02:00: You can go invasive, which means implanting tiny electrodes directly into your brain Which lets you pick a really good signal with eye precision But of course these require surgery and could be Really problematic for the patient.

00:02:15: or you can go non-invasive.

00:02:17: With is actually what we're trying to do using EEG Which is essentially cap with sensor placed on the sky And the signal is noisier.

00:02:25: but the system is safe and can be used by anyone, so it's much more practical for a clinical setting.

00:02:34: And from this point we implement partial learning algorithms which are trained to recognize these neural factors in real time.

00:02:43: In this way the computer learns how to distinguish between the intention of opening your hand versus resting or the direction that you want to perform.

00:02:55: Would also this technology be used beyond rehabilitation, for example in any individuals about

00:03:00: you?

00:03:00: Yes absolutely and I think that is one of the most exciting aspects.

00:03:04: The interaction with a machine actually has two-way adaptations.

00:03:08: so from one side it gets refined as they collect more data from specific patients but on other hand the brain itself learns to produce cleaner patterns which the system can decode easily.

00:03:25: And so yes, the brain also gets better at using the BCI.

00:03:29: And

00:03:30: final question... Can the brain adapt and get better at choosing a BCI over time?

00:03:36: Yes!

00:03:36: I think this potential use of these BCIs could go beyond rehabilitation especially for non-invasive recording systems which can be safely used by anyone.

00:03:46: So you can imagine some sort of video game that is controlled through these systems.

00:03:53: And maybe in the future, as the technology matures we will see some commercial devices that let you interact with for example your phone or computer just by thinking.

00:04:03: That's very fascinating.

00:04:04: thank-you very much so... We have seen how brain signals can be decoded but what if could go one step further not only reading the brain directly modulating its activity?

00:04:23: Dario, obstructive sleep apnea is a very common condition.

00:04:26: Can you briefly introduce yourself and explain us what happens in the brain during sleep in these patients?

00:04:32: Hi everyone!

00:04:33: I am Dario Goraci, a PhD student at the Department of Biomedical and Neuromotor Sciences at the University of Bologna.

00:04:41: During my research i have focused on studying sleep and respiratory phenotypes in marine moods.

00:04:47: Today we will talk about obstructive Sleep Apnea also known as OSA.

00:04:52: OSA is a pathological condition characterized by the complete or partial obstruction of upper airway during sleep.

00:05:00: OSA leads to frequent arousal, sleep fragmentation after hyperintention and increased morbidity and mortality primarily due to adverse cardiovascular events.

00:05:11: This condition is particularly severe during early sleep A state characterised by Atonia of voluntary mascours, including those involved in assuring upper airway patency such as tongue-mascours.

00:05:23: So your project proposes a very innovative solution?

00:05:26: How can chemogenetics help modulate this condition according to you?

00:05:32: Chemogenetics is the technology that can be used remotely to modulate motor neurons responsible for tongue contraction and therefore prevent airway obstruction during REM sleep.

00:05:43: Briefly, in our experiments retrograde viral vectors are injected into the dorsal collar to express chemogenetic receptors in the projecting neurons.

00:05:53: A few weeks later my standard go surgery for the implementation of electrodes to record sleep states and a diaphragmatic activity.

00:06:02: Few days later six recording sessions of similar average are conducted alternating tested and controlled conditions.

00:06:09: The conditions differ in terms of the interperitonal injection or with the saline, or the chemogenetic activator compound before recording of the biolatric signals concomitantly to non-invasive detection of the breathing pattern specifically of sleep apnea.

00:06:27: Wow that sounds very promising.

00:06:29: but can this approach be translated into humans?

00:06:32: Yes because it is minimally invasive so will easily translate.

00:06:37: And one critical point, is the viral vector safe?

00:06:41: We'll conduct an analysis and explore this possibility by sectioning the tongue to check whether inflammation is present both in acute and relic stages.

00:06:50: Moreover, a study of behavioral states will show if there would be differences on mice treated or not!

00:06:58: How do you ensure selectivity between CNO and virus?

00:07:02: This is quite an interesting question.

00:07:04: We are sure that you see the threads, which are designer receptors exclusively activated by designer drugs.

00:07:11: so the CNO can activate only neurons.

00:07:14: having that kind of receptor expressed

00:07:17: Okay!

00:07:17: So we could say that modulating neural circuits open new therapeutic possibilities for sure but to target brain effectively... ...we first need understand how its networks are organized.

00:07:36: So, it's my pleasure to introduce the next speaker.

00:07:39: Please go ahead!

00:07:40: Hello everyone I am Dr.

00:07:42: Vittoria Rosetti and I'm a neurosurgeon and PhD candidate in biomedical neuro-motor sciences.

00:07:49: After completing my neurosurgery residency at University of Bologna i started as second residency in radiation ecology To further explore how this surgery can be used to treat both functional and oncological brain disorders And that gives the context to my research which could be introduced with a question, why study pain?

00:08:14: The

00:08:17: guiding question is simple but fundamental.

00:08:19: Can we better understand how the brain constructs this experience of pain in order to treat it more effectively and with greater precision?

00:08:27: Today, we know that pain is not merely a signal originating from tissue... ...but rather an emergent product of complex brain networks integrating sensory-affected and cognitive components.

00:08:38: In other words, pain isn't just stimulus It's network.

00:08:42: And according to you, why is it so especially important to investigate these pain networks today?

00:08:49: Because chronic pain represents one of the most significant challenges in contemporary medicine.

00:08:54: A substantial proportion of patients continue to suffer despite currently available therapies.

00:09:00: This is particularly evident in conditions such as advanced cancer pain or neuropathic pain syndromes, or trigeminal neuralgia which is one the most excruciating pains that patient describe where conventional pharmacological and interventional approaches often provide only incomplete relief.

00:09:21: Basically your project aims to change the way we conceptualize and treat pain, right?

00:09:26: Exactly!

00:09:27: The central objective of the project is to map the brain circuits involved in the generation and maintenance of pain both in health individuals, and then patient with chronic pain condition.

00:09:38: To achieve this we employ advanced neuroimaging techniques including functional magnetic resonance imaging a diffusion-based trotography which allow us to investigate both their functional and structural connectivity.

00:09:55: According to you, what makes this project particularly innovative compared with the previous research for example?

00:10:02: There are

00:10:02: two main elements of innovation.

00:10:04: The first one is the integration of advanced neuroimaging specifically functional MRI and tractography with computational analysis.

00:10:13: using specialized software and artificial intelligence based models we aim to reconstruct the neural architecture.

00:10:23: The second innovation lies in the application of neuromodulation strategies, particularly functional red surgery which enables targeted modulation of specific brain circuits in a completely non-invasive manner.

00:10:39: The conceptual shift here is very important!

00:10:51: potential impact could this have for patients according to you?

00:10:55: If we can more precisely identify the brain circuits responsible to move toward a model of personalized pain medicine, in which treatment strategies are tailored

00:11:45: We can go forward, so understanding the brain also requires advanced imaging as we know.

00:11:50: But what if we could go beyond real patient data and actually generate new brain images?

00:11:57: That's why we have with us a WADD.

00:11:59: So let me start with a question that sounds like science fiction Can artificial intelligence really create brain images that have never existed?

00:12:07: Thank you for having me.

00:12:09: In our lab, we are working to generate synthetic brain images.

00:12:12: We use advanced model called VI Gain and monotraning on a real brain MRI scan from ATME.

00:12:20: These models learn not brain the curve of cortex shape of hippocampus texture of tissue.

00:12:26: then we inject control noise and AI generate completely new realistic brain images that doesn't belong to any actual patient.

00:12:35: Here is the wow moment.

00:12:37: So you are essentially creating synthetic brains, why it's this important?

00:12:41: In which

00:12:42: you're a researcher who needs thousands of brain scans but only have access to two hundred meters And those patients have privacy rights.

00:12:51: You cannot just share data.

00:12:53: synthetic images solve both problems.

00:12:55: they are automatically accurate but hundred percent anonymous.

00:13:00: no patient,no privacy violation.

00:13:03: So how do we evaluate whether these images are truly realistic?

00:13:08: We rate that to bring better diagnosed algorithm.

00:13:11: That is why the innovation happened each brain into a brain region, thinking of them as actual fingerprint of the brain.

00:13:23: We may have hundreds of features brightness and above-the-tissue.

00:13:27: had an us about size and texture reveal patient patterns human eye cannot see.

00:13:33: our radiologist may see this look normal but our AI detected textures differently model their distinguish between real and synthetic images with up to ninety eight percent accuracy.

00:13:47: Wow,

00:13:47: so you reported very high accuracy.

00:13:49: What did you actually discover?

00:13:51: Today we explore how AI is transforming the way we see from synthetic images to digital fingerprint that reveal digital.

00:14:00: but even after revolution its collaboration because the future of brain medical will be built.

00:14:07: neuroscience engineer and computer science world.

00:14:10: Oh, thank you very much.

00:14:11: So so far we have seen lots of interesting projects in our PhD program from digital models of the brain and now we move to the cellular level where understanding disease mechanisms may open actually new therapeutic strategies.

00:14:36: that's why we have Camilla.

00:14:39: your project connects actually a rare genetic disease with the brain tumors can explain this link?

00:14:45: Hi, I'm Camilla a PhD candidate at the Cellular Signaling Laboratory in Bologna.

00:14:52: My project concerns glioblastoma – the most common and aggressive brain tumor in adults originating from Neuroglia.

00:15:00: The need to find new therapeutic targets for this pathology is undeniable.

00:15:05: Just think that standard therapy has been the same since ten years and prognosis remains poor.

00:15:11: This is mainly due to therapy resistance and tumor recurrence, largely attributed a group of cancer cells named glioma stem cells.

00:15:21: So what's the role of Prelamin A in this context?

00:15:25: To do so, to find new therapeutic targets we took inspiration from progeroid laminopathies a group of rare genetic diseases that lead to early aging.

00:15:35: Progeroid Laminopathes are characterized by the accumulation of a protein named Prelamin A That is the precursor for another protein that is Lamin A and this accumulation increases cell sensitivity to oxidative stress.

00:15:50: And which drugs you currently testing?

00:15:53: Well, the interesting thing is that there are some drugs that lead to the accumulation of Prelamin A and any cells that express Lamin-A can acquire a progerod phenotypes when treated with these drugs causing their exit from cell cycle becoming senescent cells.

00:16:13: And guess what?

00:16:14: The Lyoma stem cells, one responsible for tumor relapses do express LAMIN-A.

00:16:19: So the idea is to use this type of drugs, to accumulate Prelamin A in glioma stem cells reproducing a progeroid phenotype.

00:16:29: In order to reduce their sameness and aggressiveness.

00:16:33: so we took inspiration from Laminopathies for treating glioblastoma.

00:16:40: Moreover, we want to investigate also the molecular mechanism under action of this drug in gliomas themselves hoping to spread light on complexity and find new therapeutic targets.

00:16:57: This

00:17:02: is because normally in healthy astrocytes there's a silencing mechanism by which lamin A is not expressed, so they cannot accumulate the precursor of this protein.

00:17:12: So the drug has no effects on the healthy astrosites cycle and vitality.

00:17:18: Wow super interesting!

00:17:20: All these advances from imaging to molecular biology ultimately converge into one goal improving how we treat patients.

00:17:28: and this is where surgery plays a crucial role.

00:17:40: And, This Is Where My PhD Project Actually Comes In.

00:17:43: So as a neurosurgeon I will explore the evolution of cranial neurosurgery from fully endoscopic techniques that are actually improving to robotic minimally invasive approaches.

00:17:53: Today in fact We can reach deep complex regions through keyhole corridors minimizing surgical trauma.

00:18:01: but these approaches are still not fully standardized and their potential remains largely untapped.

00:18:07: At the same time robotics is opening a new frontier, it's always so developed also in other surgeries... ...and it allows us to enhance precision scale down movements beyond human limits.. ..and potentially integrate real-time imaging artificial intelligence and advanced visualization just thinking about all the free-range movements that a robotic arm has.

00:18:33: The goal of my project is to create a translational platform where anatomy, technology and circle training actually converge together from cadaveric dissections in three D models to an artificial intelligence drive and trajectory analysis and augmented virtual reality simulation.

00:18:51: We would like to better understand and optimize of course, later teach all these new approaches.

00:19:00: Ultimately last but not least this pathway leads to clinical translation bringing robotic minimal invasive techniques into actual real surgical practice because in the end behind every new surgical technique there is anatomy.

00:19:19: So, today together we have seen from the coding thoughts to modulating neural circuits and generating synthetic brain images.

00:19:26: We are transforming actually the way we understand and treat the brain And this progress is driven by the convergence of neuroscience engineering and clinical practice.

00:19:37: Thank you for having been with us today For more details on each project an opportunities for collaboration.

00:19:44: You can find all information in that description.

00:19:47: Thank you again for having listened to us and we'd be happy to connect very soon.

00:19:52: And with that, I will wrap up today's episode!

00:19:56: Thanks so much for joining me on this journey.

00:19:58: Stay tuned for more episodes of PhD Dynamics beyond the London.

00:20:03: Until next time...