“In 1889 Joseph Pulitzer, publisher of the New York World, started to cruise around the world. His doctors had warned him against overexcitement. In Greece he received a cable from his brother-in-law, Colonel William L. Davis, which kept him in a rage until his ship reached Constantinople. As the ship was about to leave port, Mr. Pulitzer said to his secretary, Ponsonby:
“How suddenly it has gotten dark.”
“It’s not dark,” said Ponsonby.
“Well, it’s dark to me,” said Joseph Pulitzer.”
Pulitzer was forty-two years old during the voyage. It remained dark for the rest of his life.
He suffered from the detachment of the retina.
While Pulitzer did not let his ailment hinder his strong journalistic temperament, it did affect his effectiveness, and prevented him from witnessing global events until his death in 1911.
During his time, retinal detachment invariably led to blindness. Had he been born a hundred years later, modern medicine may have helped him keep his eyesight. Today, we have various surgical methods of alleviating retinal detachment – scleral buckling, retinopexy and vitrectomy, to name a few. Vitrectomy, in particular, is successful in around 90% of cases. However, these numbers mask a few hard realities –
- The high success rate is directly attributed to the experience of the surgeon. But experience comes with a price; namely sufficient time and potentially a few casualties.
- No amount of experience grants a surgeon immunity against the myriad complications that can arise from patient-to-patient, a consequence of the delicate and complex balance that every part of the human body maintains.
Can we share the burden of the surgeon? In particular, can we rely on computational prowess beyond intuition built off experience to predict anomalies? Searching for answers to such questions was what led Nihar, ED19, to take up this project under Prof. Krishnakumar of the ED department that eventually became his Master’s Thesis.
Vision of the Project
The retina is a thin layer of tissue located at the back of the eye that is responsible for receiving light and converting it into neural signals, which are then sent to the brain for visual recognition. The space between the lens and the retina is filled by a clear, gel-like, visco-elastic substance called the vitreous humour. It consists of 99% water, and some non-homogeneously distributed fibers that “hook” onto the retina.
Retinal detachment (RD) is a surgical emergency where the retina, situated at the back of the eye, separates from its underlying support tissue, including crucial blood vessels that supply oxygen and nutrients. This separation can lead to severe vision loss and blindness.
Vitrectomy is a popular micro-invasive procedure to correct RD. A device called a vitrector (think straw) is used to suck out the vitreous humour and is replaced with a saline solution with the help of an infusion cannula, which helps maintain turgidity within the eye. Except, it isn’t that simple. The fibers, which are hooked to the retina, have to be snipped before the vitreous humour can be removed.
As one would expect, there are a huge number of factors that can subtly or drastically affect such a procedure. The question (a pretty hard one at that) is – given many such factors and parameters, ranging from retina dimensions, myopia status, to even the diabetic status of the patient, can we computationally determine their effect on the retina and inform the corrective procedure to guarantee success?
Model
What do all good predictions need? An accurate model. Nihar’s task was to find a way to model everything in our setting – retina, vitreous, fibers, vitrector – the entire lot. With every component governed by their own set of daunting equations, how would one go about modelling their interactions? As many of you might correctly guess, the key is FEM.
The Finite Element Method (FEM) is a prevalent engineering technique to solve complex differential equations with equally complex boundary conditions. Essentially, it involves dividing the object of study into many many smaller pieces by creating a mesh on the object, then identifying simpler equations that govern these smaller pieces, and solving all these sets of equations simultaneously to arrive at a solution for the entire system.
However, a certain complication arises. The equations governing an element purely within the fluid are different from those governing the interactions between the solid and the fluid. All ocular biomechanics researchers so far tackled this problem in the traditional way – they perform CFD (computational fluid dynamics) simulations first, take the output, and use it to simulate the so-called Fluid-Structure Interaction. But the particular method Nihar chose to employ, called the Coupled Eulerian-Lagrangian (CEL) method, models fluid interactions using solid equations with a couple of modifications, allowing a single simulation to model both interactions.
Steps
“Modelling a complex system is like writing a long piece of code. If you write the entire code off the bat, and if you make an error (you’re bound to), the error will propagate throughout the code and you will be left unable to pinpoint the origin of the error. That’s why you always start simple.”
Armed with the requisite background, the first thing Nihar did was to tackle a simpler problem – given a cylindrical straw in a liquid contained in a rigid vessel, how does one model the liquid aspirating out? With CEL being unheard of in this domain, Nihar had to work things out from scratch. Successful simulation of this marked the first big step in the project.
All subsequent steps are essentially refining the model to include more and more complexities of the system the model is meant to simulate. For instance, there’s no “atmospheric pressure” in the eye; pressure is maintained by the influx of the saline solution. Modelling this was the next step.However, it turns out that even the outflow modelling needs fixing. The vitrector is not just a cylindrical straw, it has a complex structure with a valve for gating the flow. The outflow is thus regulated pressure-driven flow. Including this is the next task.
After all this hard work, our protagonist got to fixing the metaphoric elephant in the room of simplifications – the retina is not a rigid vessel, neither is the vitreous humour water. This puts us into the domain of soft tissue mechanics, which is notoriously non-linear and challenging to simulate. So far, Nihar has been able to get the dynamics of the vitreous humour up to a functional level, and is currently working on modelling the adhesion between the humour and the retina (via fibers).
Future work in this domain will be to include more and more patient parameters, inching closer to the goal of having complete Patient Specific Simulation, ultimately personalising every surgery for the particular patient.
Backstory
When doing the ED course Implantable and Surgical Devices taken by Prof. Krishnakumar, Nihar became really interested in the subject and decided to explore further. “It’s fascinating that Prof. Krishnakumar specialises in both tire mechanics and biomechanics. It makes you realise how versatile knowledge of the underlying mechanics is,” he says. When Nihar asked about a summer project, the prof offered him a “long-term problem requiring more than a summer’s work”. Nihar assented and opted to work on it instead of his 6-month corporate internship, and it eventually became his MTech project.
The problem statement actually originated from the acclaimed eye hospital Sankara Nethralaya, known to engage in clinical research of all things eye. Thus, Nihar also reports to the doctor who gave him his problem statements, discussing the clinical aspects of his work with him.
“It’s crucial not to delve so deeply into mechanics that the simulation loses sight of the bigger picture. Neglecting real world aspects might get you a “Nice colours and all, but how does this particular simulation help me?” in response.”
Recently, Nihar visited the hospital to spectate an actual vitrectomy surgery to understand the shortcomings of the procedure that he hopes to assist. “Straddling the boundary between the clinical world and the mechanical world, it is just as important to speak both languages as it is to understand considerations of both sides”, he says.
Nihar is admitted to a PhD programme where he’ll be jointly working with Johns Hopkins University, University of Delaware and University of Louisiana State. His work will be to model the mechanical effect of extreme stress on the brain during accidents, and form a corresponding microstructural theory. “The knowledge gained is not in the specifics. The engineering intuition gained by learning how to tweak such a model can be extrapolated to a bunch of other domains,” which he says helped him with his PhD applications.
Readers interested in delving more can contact Nihar through email.
Edited by Ishan Khurma
