Vision Correction Surgery X AI

Can I receive refractive surgery?

Vision Correction Triage
Vision Correction Triage

The cornea is the main element of the eye to be analyzed when determining the viability of refractive surgery. Various measurements regarding the cornea have to be taken to decide if someone can have refractive surgery. This includes: thickness and strength of the cornea, flatness and the overall topography of the cornea.

After examining the data produced at the end of several examinations the doctor will ask 3 important questions:

  • Can surgery be conducted?
  • Which surgery is most appropriate for the patient?
  • Is additional surgery to strengthen cornea required?

The first question is widely regarded as the most important as this question looks at the corneal safety in order to draw a conclusion on whether the doctor will proceed with the surgery or not. If the doctor chooses to proceed with a surgery in spite of evidence that suggests a patient is not suited for it, the patient will suffer severe post-surgical complications which will affect them for the rest of their life.

Corneal thickness is the main factor when considering corneas safety. The residual cornea (after surgery) must be at least 380μm. This means that the thickness prediction has to be accurate.

Another important factor is the cornea curvature. The minimum diopters of corneal curvature to be secured for a safe refractive surgery is 34. It is important to note that myopia correction by 1 diopter changes the corneal curvature by 0.75 diopter.

Corneal topography is another aspect that has to be inspected. Corneal maps are used to analyze the differences between normal corneas and potentially abnormal corneas. 

What is the best corrective surgery?

An architecture of B&VIIT's proposed machine learning system to predict candidates for corneal refractive surgery

The B&VIIT AI has learned up to 230,000 patients’ data sets through various existing assessment standards. After intense scrutiny of this data it was established that certain test results will be associated with an appropriate surgical method.

When determining the possibility of surgery, the AI will look at the state of myopia, astigmatism and corneal thickness.  Once the possibility of surgery has been confirmed the next step is to determine whether laser surgery is possible. In the case of poor vision or uneven or asymmetric cornea surface area, the doctor might recommend PIOL or customized laser ablation surgery that can increase safety.

One thing needs to be kept in mind when considering the role of AI in the decision-making process of vision correction surgery. That is the AI does not operate on a superior level to the doctor, it merely assists the doctor’s role during simple, time consuming examinations. 

Another limitation might be the problem of data. Obviously without data the development of AI becomes impossible. As all machine learning is based on the analysis of mountains of data. 

Vision prediction after surgery

Test result of corneal thickness

Is it possible for a machine to predict what the results of your vision correction procedure will be before getting the surgery? In order to answer this question, we will take a look at a Google led AI project. In 2018, Google had an AI learn the medical records of patients hospitalized for 24 hours or more. Doing this the AI could predict, with amazing precision, the risk of death, re-hospitalization ratio within 30 days after release, and the diagnosis name when being discharged. 

This research is interesting because the machine can look at various, unsorted data, that contains arcane medical definitions, and then decide which information is important in order to make its prediction, and which information is not. 

The prediction of results was already possible before AI, but what AI has made possible, is the explanation of why the predicted results are what they are. And also, how those predictions were made, meaning which factors were analyzed to make that conclusion. 

The optimal refraction value

The AI demonstrated the differences of surgical factors between surgeon A and B.

A nomogram is a database that the doctor uses to determine the values with regards to the results of surgery. The values in the database are entered by measuring the value of certain factors such as myopia correction value, astigmatism correction etc. 

The last function of the B&VIIT AI is to automatically calculate the nomogram that was usually calculated by the doctor performing the laser eye surgery.

In order to make this AI, the following data needed to be analyzed: individual diagnosis value of patient receiving surgery, correction value decided by doctor, and result of surgery. 

The practical use of this AI is that it can be used as a confirmatory tool by the doctors in order to insure correct diagnoses and also to expedite recommendations. 

Published by B&VIIT EYE CENTER

B&VIIT Eye Center is an ultra-modern, Korea's representative ophthalmic clinic that has successfully accomplished vision correction surgeries on over 500,000 eyes for 27 years.

One thought on “Vision Correction Surgery X AI

Leave a Reply

%d bloggers like this: