Machine Learning Algorithm That Detects Tinnitus Based on EEG Measures and Cognitive Performance

[Christiaan]

Cortical auditory evoked potentials, brain signal variability and cognition as biomarkers to detect the presence of chronic tinnitus (Cardon et al, 2022)

Here, we present the results of an observational study comparing potential biomarkers for tinnitus between tinnitus patients and matched controls. Our most important finding is the development of a logistic regression model that can accurately detect tinnitus cases based on three proposed biomarkers. These include cortical evoked activity corresponding to the P300 component, brain signal variability on fine time scales, and delayed memory cognitive performance. Learning which parameters can distinguish between tinnitus patients and controls could potentially provide insight into the underlying pathophysiological mechanisms of tinnitus.

The proposed model is based on parameters that can be measured objectively, including cortical auditory evoked potential (CAEP) activity and brain signal variability, as well as cognitive performance. This model had an excellent fit on the data used for training, and exhibited above chance level accuracy and excellent performance on a test dataset, as evidenced by the high area under the receiver operating characteristic curve. Although we used the results of the statistical analyzes performed on the training dataset to inform the model, its final performance was tested on an unrelated dataset not used for these analyzes. As such, we provide an unbiased evaluation of the final model fit. Strict measures were taken to ensure that this model would be tinnitus-specific. For instance, tinnitus and control subjects were accurately matched for hearing levels and all included subjects had normal hearing for their age. As hearing loss is known to have significant effects on both cognitive performance and CAEP (Alain et al., 2013; Claes et al., 2018; Rufener et al., 2014), limiting the confounding effect of hearing loss is crucial in analyzing these parameters. Furthermore, included subjects did not exhibit clinical signs of anxiety or depression, and both groups were matched for other important demographic factors including age, sex, education and handedness. Of course, it might be the case that different confounding factors, particularly hearing loss, age and signs of anxiety and depression, modulate the perception of tinnitus in different ways. In order to chart putative influences of these factors on tinnitus perception, the proposed model may be expanded in the future to include different patient groups. For instance, a grouping of participants based on age or the presence of hearing loss might be considered in future research.

Analyzing the cortical auditory evoked potentials in response to an infrequent target tone, we identified a cluster of significant differences in evoked activity between tinnitus patients and controls. Activity in this cluster was significantly lower in tinnitus patients. This cluster corresponded precisely to the P300 component, a response to unexpected but relevant stimuli. A recent systematic review and meta-analysis found that the amplitude of the P300 component was specifically decreased in tinnitus patients compared to controls, whereas earlier exogenous sensory components were similar in both groups (Cardon et al., 2020). Using a data-driven approach, we were able to confirm these findings in our matched subject groups. A diminished P300 response has been linked to an impairment in top-down information processing, as evidenced by abnormal cortical connectivity profiles, and cognitive efficacy (Hong et al., 2016; Polich, 2004). Specifically, the decreased P300 component in tinnitus might be related to an altered functionality of key structures in the so-called 'salience network', such as the anterior cingulate cortex and insula (Cardon et al., 2020; Menon and Uddin, 2010). In the absence of external stimuli, the increased activation in these regions in tinnitus patients has been linked to the salience attributed to the tinnitus sound (De Ridder et al., 2011). It is possible that, when presented with an auditory stimulus, this importance awarded to the tinnitus percept overrules the salience assigned to a target sound, thus resulting in a diminished P300 response. Indeed, a source estimation procedure revealed a decreased underlying source activity in the insula in response to the target tone, as well as the bilateral temporal cortices and parahippocampus. Previous EEG studies have shown hyperactivity under resting-state conditions in all of these areas in tinnitus patients (Adamchic et al., 2014; Moazami-Goudarzi et al., 2010; Vanneste and De Ridder, 2013). The observed decreased activity in these regions in response to a target tone might indicate that priority is given to the tinnitus percept at the expense of external stimuli, further confirming the importance of top-down attentional processing in tinnitus perception.

One of the researchers has recently published an article on this particular subject in a popular science magazine (EOS Wetenschap, Belgium). I think it's easier to comprehend than the original science paper.

Tinnitus caught in the brain (Cardon, 2022)

Photo: Bart Vandersnickt

No one else hears that beep, ringing bell or jackhammer in your ear. But for the first time I can track him down in your brain. With this new method I want to be able to determine later which treatment is the best choice for you.

You are constantly harassed by a sound that seems to come from nowhere. You hear it every moment of the day, but putting into words exactly what you hear is not so easy. Does it sound like a high-pitched beep, or rather a low hum? You may still hear the peaceful murmur of the sea in the morning, but in the evening it sounds more like a jackhammer working on the inside of your skull pan.

Everyone experiences tinnitus differently

What you experience is called tinnitus, and you are not alone: worldwide about one in six suffers from the same problem. It also doesn't stop at that annoying sound. Often people with tinnitus have difficulty falling asleep or concentrating. Tinnitus can also make it difficult for you to understand others or follow conversations. In severe cases, it can even cause symptoms of anxiety or depression. Everyone experiences tinnitus differently. All these differences make it difficult to diagnose, let alone treat, tinnitus.

Surprisingly, the key to understanding these various complaints is not in the ear, but in the brain. After all, in my doctoral research I discovered that tinnitus leaves a clear signature in the brain. Some brain regions are so captivated by that tinnitus that they respond less well to sounds from the outside world . Thanks to my research , we can now even objectively detect tinnitus in the brain. Equipped with that information, I want to make it easier in the future to determine which treatment is the best option for you, so that the volume knob on that jackhammer can finally go down.

Brain under the spell of tinnitus

During my PhD I discovered that tinnitus causes the brain to respond less well to external stimuli. I had subjects with and without tinnitus listen to different tones, while recording which electrical activity this triggered in the brain. Your brain does give some response to any environmental noise. But when you hear an important and striking sound (say, a fire alarm, your alarm clock, or Lizzo's new single), your brain will react much more strongly. And in this response, tinnitus clearly plays the role of a jammer. I found that, in individuals with tinnitus, certain brain regions respond less well to those important sound stimuli.

With a simple measurement I look for which regions in the brain are occupied by tinnitus.

It seems that these brain regions favor tinnitus. For example, our hearing center appears to be so captivated by the tinnitus sound that it responds less adequately to important outside information. But other regions of the brain are also taken up by the tinnitus. I found the same pattern in the insula, an important emotional core, and the parahippocampus, a region that plays a crucial role in storing and recalling memories. It is therefore not surprising that tinnitus often evokes such an emotional response, or that people with tinnitus sometimes experience concentration or memory problems: all these different complaints can be explained by the influence that tinnitus exerts on the brain.

Surprisingly, not only the hearing center, but also other crucial brain regions are under the spell of tinnitus.

This brain activity is indeed a 'signature' of tinnitus: you can detect tinnitus effectively with it. After all, at the end of my PhD I developed a model that is able to recognize the tinnitus brain. You could call it a sorting model, because its function is to sort a large collection of data into two groups: a tinnitus group and a tinnitus-free group. Based solely on the simple brain measurement we took, in which we recorded how well the brain responds to external stimuli, the sorting model succeeded in successfully completing that task. In other words, tinnitus can be accurately detected in the brain.

Towards a quieter future

For the first time, thanks to my PhD research, we can objectively detect tinnitus in the brain. That is already an important step forward. Although everyone experiences tinnitus differently, there appears to be a common basis: the signature of tinnitus in the brain. This could be an important starting point for further research.

Finding the right treatment for tinnitus isn't always easy

For example, information from the brain can help us better manage tinnitus. Although people are often told to 'learn to live with that tinnitus', there are treatments that can significantly reduce the impact of tinnitus. But finding the right therapy is not always easy. In the future, I want to use the same brain measurement to determine whether I can predict how much you will benefit from a certain treatment. That way we could save a lot of time and effort, and put you on the road to a quieter future as quickly as possible.

 

[Nick47]

A difficult read. Where was this study done? Is it the same/similar to the research at the Bionics Institute?

 

[transient18]

I also read it today! Though it does not yet point to a solution, did I understand that right?

 

[Christiaan]

A difficult read. Where was this study done? Is it the same/similar to the research at the Bionics Institute?

The study was conducted in Belgium.

I think that the Belgium study differs from the Bionics Institute in terms of recording information/data.

The Bionics Institute is using a special kind of imaging technology (functional Near-Infrared Spectroscopy) in the brain cap that measures blood oxygen levels when a patient is exposed to sound.

Cardon and her colleagues at the University of Antwerp have used a certain kind of EEG instrument (Micromed™ SD LTM 64 Express system) that measures electrical activity in the brain when a patient is exposed to sound by using small electrodes attached to the scalp.

 

[Padraigh Griffin]

One day 'machine learning algorithms' will solve tinnitus.

I got excited when I saw the title.

 

[KoolKat]

Cool. Hopefully it won't take 200,000,000 years I just know there's something going on up in our brains. We just have to find a way to discover it. And please for the love of god let our brain have a mechanism switch that we can turn the tinnitus on & off.

Seriously, how in the world wasn't this nobody's objective until recently, meaning why did all these researchers and scientists, who were formulating cures for tinnitus, not think about creating ways to map tinnitus out?

I just had an EKG and was wishing they had something like that for the brain to detect tinnitus.

Hopefully with the way things are moving, it won't be long. All we need is Dr. Susan Shore or someone to make something that can at least diminish our tinnitus perception enough to where it doesn't bother us, then wait for things like this to come into reality.

 

[IndyMLVC]

I have many medical "issues" and tinnitus is just one of them. I have lots of problems sleeping (can you guess why?) and went to have a sleep study done about two years ago. For those that don't know, you stay overnight at the study location, they hook up a TON of wires to you to monitor your brain and body during sleep.

At the time, my tinnitus wasn't my primary concern. However, after getting the results, I'm wondering how much it plays a part in it or if my results could somehow be used to measure it.

My doctor told me afterwards that I sleep a lot... but little of it is REM/deep sleep. I was coming out of that sleep 40+ times an hour, in addition to grinding my teeth and the apnea, etc, etc, etc. His words to me were "you know that your anxiety is ruining your life while you're awake. You now have proof that it's doing the same while you sleep."

I really do wonder how much the tinnitus is playing a role in my inability to sleep properly.

I don't mean to hijack this thread. ust adding my 2 cents...

 

[Christiaan]

I also read it today! Though it does not yet point to a solution, did I understand that right?

Ey @transient18, I think it's part of the solution, as it may help researchers to assess measurable differences in patients with tinnitus who have received treatment X vs. those who haven't received it by analysing electrical activity in parts of the brain that are involved in the manifestation of this horrible affliction.

 

[Nobody19]

Btw, the main author of this research is a Belgian PhD student and finalist in the Belgian PhD Cup. I hope she wins, it would bring quite some attention to tinnitus here.

 

[Nobody19]

Btw, the main author of this research is a Belgian PhD student and finalist in the Belgian PhD Cup. I hope she wins, it would bring quite some attention to tinnitus here.

She finished second.

 

[lusiol]

It's very, very good news, right?

 

[CRGC]

I've noticed that tinnitus appears to be quite a talked-about subject in Belgium. Many young music artists from there (Angèle, Roméo Elvis, Veence Hanao) have been very vocal about their struggles with tinnitus and hyperacusis.

 

[Nobody19]

I've noticed that tinnitus appears to be quite a talked-about subject in Belgium. Many young music artists from there (Angèle, Roméo Elvis, Veence Hanao) have been very vocal about their struggles with tinnitus and hyperacusis.

Which one of those have hyperacusis?

 

[CRGC]

Which one of those have hyperacusis?

Veence Hanao is the artist struggling with hyperacusis. Here's a text he wrote about it in French: https://www.oreille-malade.com/grou.../temoignage-musicien-acouphenes-hyperacousie/

 

[lusiol]

I've noticed that tinnitus appears to be quite a talked-about subject in Belgium. Many young music artists from there (Angèle, Roméo Elvis, Veence Hanao) have been very vocal about their struggles with tinnitus and hyperacusis.

Yeah definitely!

I wish more artists would talk about this and raise awareness.

 

[Nobody19]

This video talks more about her research and how exactly they can detect tinnitus in our brain.

 

[Nick47]

A small study out of China comparing EEG findings in tinnitus patients versus healthy controls.

Difficult read. Small population. Not much learned really. Feeds into the thalamocortical dysrhythmia model.

EEG signal classification of tinnitus based on SVM and sample entropy

This study found a general increase in resting state EEG signal entropy compared to healthy subjects with tinnitus, reflecting the confusing behaviour of the brain in this population. Non-linear kinetic analysis provides a potentially valuable method to study brain networks, facilitating the investigation of the neurophysiological mechanisms behind the occurrence of tinnitus and hopefully providing a criterion for future clinical diagnosis of tinnitus.