The Role of Computer Vision in the Medical Field
We, as humans, have eyes to see and recognize objects in real life. However, computers can only work with 0s and 1s. How can a computer recognize different objects?
The world has always been a mystery to humans, but understanding this mystery has been made easier by our sensory organs, including our eyes.
The mechanism behind eyesight is incredibly complex, so it has always seemed impossible to replicate this wondrous ability. In recent decades, however, scientists and engineers have developed ways to give the gift of sight to machines! This is what is known as computer vision (CV), which has become one of the most popular fields of research in computer science. These programs have performed relatively simple tasks like classifying an image as a dog or cat or detecting an object, but now, people are finding more complex uses of this technology.
One of the most important uses of CV is in medical technology, from medical image analysis to surgery! Before we get into these uses, however, let’s first discuss what CV is and how it works.
How Does Computer Vision Work?
According to Adobe, computer vision is “the field of computer science that focuses on creating digital systems that can process, analyze, and make sense of visual data (images or videos) in the same way that humans do.” Essentially, a computer can analyze an image — at the pixel level — and understand it as humans might. A human can capture their surroundings using their eyes and analyze them in the brain. Computer vision works by using a camera to capture an image (like an eye) and a program to analyze it — very similar to the human visual processing system.
To identify a tree, a human outlines the features of the tree (e.g. trunk, leaves) and identifies the object as a tree. All of this processing is done by the brain and its large networks of neurons. With enough representations of trees, a person can infer a new tree they encounter to be a real tree. Similar to how humans store images through visual memory, computers use pixels to represent images. Additionally, using machine learning, computers can analyze new images of a certain object (for example, a car or an animal) and make predictions after being trained on many other images of that object.
Computer vision has been used frequently in manufacturing and engineering fields. However, it can also be used in the medical field, especially since computer vision can identify objects, classify them, and track objects for a specific amount of time, which is incredibly helpful to medical professionals like radiologists, who deal with images of medical scans on a daily basis.
Computer Vision Applications in Medicine
Since it enables computers to recognize objects, computer vision plays a key role in the ongoing automation of the medical field. Let’s take a look at some examples.
Precise Detection of Diseases
Computer vision can be used to detect diseases by analyzing visual symptoms; for example, the detection of various suspicious pigmented lesions (SPLs) on the skin can help a computer diagnose melanoma, a type of skin cancer. Identifying each lesion and coming up with a diagnosis manually is very time-consuming. However, there is an easier method, one that uses a computer vision system to detect melanoma. To understand what melanoma-related SPLs look like, a predefined dataset of images with and without melanoma-related SPLs was given to the system so that it could learn to distinguish the two categories. According to MIT News, researchers “trained the system using 20,388 wide-field images from 133 patients at the Hospital Gregorio Marañón in Madrid, as well as publicly available images.” Using this trained system, a device such as a phone could then be used to classify SPLs on the skin and diagnose melanoma. This system has “achieved more than 90.3 percent sensitivity in distinguishing SPLs [suspicious pigmented lesions] from non-suspicious lesions, skin, and complex backgrounds’’.
Given that melanoma is responsible for over 70% of skin cancer deaths, such a computer vision application could have a useful and potentially life-saving role in healthcare. By combining computer vision with deep neural networks, devices could detect common signs of harmful diseases quickly and accurately. If this concept was extended to diagnosing other diseases, it could have a profound effect on the healthcare industry.
Surgical Procedures
Humans performing surgery without medical robots will soon become a thing of the past. Today, many medical robots can perform many medical procedures with great precision. According to Addepto, “[t]he 3D, high-definition imaging that medical robots use, increases the vision of the operation field and makes depth perception accessible. As a result, surgical operations are more accurate and take less time.”
One such company that applies computer vision in the healthcare field is RSIP Vision. Their systems could “provide the surgeon with a highly accurate and effective real-time in-op view of the surgery environment.” By using solutions that enable correct and helpful navigation in surgery, surgeons can give patients better treatment when surgery is needed.
The Future
Computer vision is already starting to revolutionize the medical field for the better. By automating medical procedures and detection, we could diagnose diseases and perform surgery with less difficulty. In the coming future, such computer vision robots can be improved. One such improvement is training the robot to recognize brain disorders, which will be a more complex task than recognizing skin diseases such as melanoma. By combining multiple artificial intelligence techniques, we can improve the medical field, one line of code at a time.
Like what you read? Sign up for our newsletter below.
