Engineering the brain

If you’ve stayed up to date with all the technical buzz in the college, you’re probably aware of the interdisciplinary ‘mega project’ that four of our revered faculty have undertaken. The project essentially consists of two parts: interactive moulding of 3D models with gestures (think Iron Man), and meshing of a brain to evaluate its stress distribution.

The latter part of the project is what we’re interested in – despite the Iron Man reference. Yes, we’re nerds – so let us give you a little context: there are multiple brain injuries – called mild traumatic brain injuries – which can easily go undetected in an MRI, but give you many problems a few months later, as the brain degenerates. The idea is to create a 3D model of a patient’s brain and mesh it (i.e., define it as a discretized mathematical surface), and perform finite element analysis to simulate the stress distribution that occurred during the accident. Much of the parameters regarding the impact (we’re assuming an accident, here) will have to come from the patient, of course, but this is certainly a step in the right direction. For the stress plot allows a doctor to be aware of areas – seemingly undisturbed – that have been subjected to a dangerous level of stress, and they can take precautionary measures accordingly.

Spanning multiple branches and involving a whopping 22 students – this project group has its own meet on every Monday at 4.00 pm, in B308, where students working on the project regularly present their work. From CS to Biology to Civil to Mechanical to EEE – absolutely anyone (students and teachers alike) is welcome to walk in and attend these presentations.

For those that cannot wait till Monday for more information, however, we at the The Daily BITSian caught up with Dr. Chandu Parimi, one of the professors involved in the project, to discuss the ponderings – both technical and philosophical – involved in all that research.

We don’t normally see such large scale projects on campus, especially involving so many different students and branches. What exactly started this?

We didn’t think of it in that way. The way we saw it, we wanted students of different branches to come together and start working on a practical problem. Obviously, the faculty who started this project were from different branches so it came naturally. There was a lot of work involved, so we thought let’s advertise it, see the response and see who is actually interested based on that.

For a long scale project like this, is it difficult working with UG students considering that many students will pass out before the project is completed?

Yes, in the sense, continuity is harder. Especially since coders are usually not good at documenting code, which makes passing on knowledge to new members more difficult. Even PhD students have issues but at least PhD students are there for a longer time. But that said, the work we get out of a truly interested undergraduate makes it worthwhile. We just hope that there’ll be as much interest going forward.

How exactly are the various branches necessary FoR the project?

Well, there are a lot of parts to the project. There’s a part which is Mechanical Engineering – I mean, I wouldn’t say Mechanical Engineering, but solid mechanics, and some Biology – on how a brain actually deforms. And there’s a part where you want to mesh it – the Computer Graphics part. And there’s a part where you’re trying to process the digital image, and that’s more like digital signal processing type work. And then there could be noise in the data, so that might involve Machine Learning. So, we needed people from all over.

What has been the response of students in taking up this project in a field new to most engineers?

We have had a very good response from students of different branches. See, most of the faculty on campus, and in most major universities in India and abroad, are involved in very advanced fundamental research. This means that its effect can only be seen maybe 20-30 years from now. I think students have been attracted by the very practical nature of this project, where they can directly see their work being used in some setting. The way I see it, everyone wants to be novel but any new idea you get, someone has already tried it in some part of the world. So I just find niche problems with real life applications and try to solve it.

Do you have any suggestions to promote interdisciplinary research on campus?

For interdisciplinary work, what we can do is have more forums where we can come out and discuss our work – like this interdisciplinary workshop we just had at BITS Goa. Or the idea someone had about putting up our ongoing work on our websites, or on posters in the department, so that other faculty passing by can see it and maybe feel interested. So it’s really just a matter of interacting and working together – and especially with the teaching roles we all have, working together is really vital.

How does working with a biological system vary from the traditional mechanical systems we deal with? Are there any added difficulties?

Biological or not, it is still a material. I’m an engineer by profession and a coder by research. We are not ever limited to a single branch. The human body fascinates me, and the project seemed a natural choice for me. With a fundamental understanding how materials act and behave, we can apply this same knowledge in biology. Maybe biological systems are more complex because they are not man made, but fundamentally they are the same.

Considering the nature of the project, are any of the Biology faculty involved?

No one right now. At this point, we just consult them when we have doubts. In the past I have talked to Dr. Debashree Bandyopadhyay about a tissue mechanics problem. During a workshop in BITS Goa, I found a few biology faculty from Goa and Pilani who were interested in the project. I’ve also been in touch with Dr. Ramakrishna Vadrevu a couple of times. That being said, we are always open. If anyone genuinely wants to contribute, we are willing to work with them irrespective of branch. We are also in touch with a doctor from Kerala who will help us with medical aspects.

How will your model represent the various different parts and layers of the brain, some of which are not even solid?

MRI image give us detailed information from which we can define surfaces and the various segments of the human brain. Different researchers in the past have already studied the mechanical properties of the various parts after dissection. We are inputting the information in our model to represent different ‘materials’ found in the brain. Each ‘material’ is very different, some more jelly-like, some a little harder. The subsequent mesh for the model will be quite complicated and that is why we have CSE faculty helping us out.

If we consider a stress distribution throughout the brain, how will you deal with the fact that some tissues are more sensitive than others?

Any material has two types of properties we would need to deal with: stiffness and strength. Stiff parts draw more loads or stress. Strength determined the breaking properties of  a part. So we need to study the different materials in the brain and how they react. Various studies already discuss this. So we won’t work with stress plots but a plot representing percentage of breaking stress of various parts of the brain.

What fixed supports would you input in the software to carry out the analysis?

Finite element techniques allow us to set different boundary conditions. Now, the brain is contained in the cranium. Most of it is not directly attached to the cranium. It’s just lying inside. It’s momentum causes it to travel to one side on collision but its movement is restricted by the cranium itself. After that, it moves back to its original position. So these are the conditions we need to input. A fixed support at the brainstem and maybe a system of one way springs to controlling its displacement. 

Is there any sort of experimental validation you plan to carry out to correlate your findings with actual data?

I wouldn’t call it experimental correlation, but some form of correlation. We have MRI Images of a brain just after an injury, 3 months after the injury, and 6 months after the injury.So we can look at the 3 month or 6 month results and see if they make sense based on our model and that is a form of corroboration. Also, we’re also considering experimenting on animals by subjecting them to impact and analyzing its effect. We are in talks with Prof. Yogeshwari regarding setting up a system.

So is the idea to have a specific model for each individual patient, or to have a general model representing the human brain?

Every brain is different, every skull is different, every injury is different. There are already people who have carried out finite element models of brains, and people have subjected it to impacts already. The way we want to make a difference is the heterogeneity of the problem. As soon as our patient walks in, we want the patient’s MRI to be taken. From that MRI we build a model. We use a brain atlas of some sort to check for any lesions in the brain, perhaps, in order to reconstruct the healthy brain of that person – and then we do the simulation.

Is there any way to confirm the injury to specific parts of the brain?

It cannot be confirmed per se, but if we show that our system is reliable in predicting the regions where an injury is likely over a large range of experiments, we may see the system being adopted. It is similar to the way MRI scans are considered full proof nowadays. 

So how does a doctor use this system?

Well, if it becomes a proven system, a doctor can have an idea of what type of problems may arise in the future. Different parts of the brain are associated with different functions like memory. If suspected of damage, a doctor can take preemptive measures by starting pre-treatment depending on the type of injury sustained. At the very least, the patient can be told to keep aware of certain problems that may arise in the future so he/she knows to get medical consultation for it.

If you are working with a 3D distribution, how will the data be represented for doctors to use?

Visualization is going to be a tricky problem. We could show the output in multiple cross sections like you see for an MRI. I personally feel that we will have to give the output as a video. We could also create our system in such a way to find and highlight the areas of maximum stress.

How does your simulation work vary from previous studies?

Various studies have already been undertaken but researchers haven’t carried out their analysis directly on real patient data. Rather than a theoretical study, we are looking for a comprehensive end to end solution from MRI to processing to output. We want to provide unique outputs based on each patient. Right now, we are just trying to get the program to work but, later, we will have to get into the nitty gritty details. One issue would be the anisotropy in properties caused by a particular directional arrangement of  fibres.

Do you have plans for any other projects in similar vein?

Depends. My main focus remains on teaching, which I enjoy. That said, if something interesting comes up I will see. We are also working on 3D sculpting project right now. In fact, I would like to work on something even more grounded than this project. Something like an App. I’m honestly surprised by how few civil engineering apps are out there.

How long do you expect the project to take to complete?

Realistically, we cannot really set a date on when we’ll come up with a tangible product. We have set a deadline of 3 years for a 2D model. It may take upto 5 years for a 3D model. We have already submitted details of the project for funding. 

Karthik Vickraman
Reeti Sarkar

Note: We would like to sincerely thank Dr. Chandu Parimi for sparing quite a bit of his valuable time to answer our incessant queries on the subject.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s