Artificial Intelligence Helps Doctors Read Brain Scans

Home » Artificial Intelligence Helps Doctors Read Brain Scans
December 03, 2019
Edward Smith

Reading Brain Scans with Artificial Intelligence

A new study showed that artificial intelligence (AI) can help doctors diagnose and treat brain injuries. This study was published by the research team at the University of California at San Francisco (UCSF), who worked with scientists from the University of California at Berkeley (UCB). Every day, radiologists review countless images that are taken of all parts of the body. Doctors are under a lot of pressure to not miss a single abnormal finding on these scans. Some injuries, such as brain hemorrhages, can be incredibly challenging to spot.

To take some of the stress away from the medical team, scientists from UCSF and UCB developed a computer algorithm to help doctors read brain scans. The goal is to help doctors more quickly diagnose traumatic brain injuries. The hope is that this will lead to faster treatment times and improved patient outcomes.

The Construction of a Brain Scan

X-rays, MRIs, and CT scans show doctors what might be happening in the human body. These scans generate images differently, ranging from magnetic to density differences. Doctors are trained to interpret them and make decisions using these images regarding necessary diagnostic and treatment plans.

Three-dimensional (3D) scans, such as CTs and MRIs, make their images by stacking dozens of images on each other. A radiologist interprets these images by scrolling through them using a mouse wheel (or something similar). In many cases, the abnormalities on these images are minuscule and difficult to spot. Abnormalities might be missed if a doctor is on call, tired, and not thinking clearly. The goal is to reduce this error rate.

The Design of the Artificial Intelligence Program

Researchers from UCSF and UCB designed a computer algorithm to use artificial intelligence that could accurately pick out abnormalities in medical images. This program aims to help radiologists focus on the images that the computer deems most important. An image is considered significant if the program believes there might be an abnormality in the scan.

The researchers know that the accuracy of this program must be perfect. If even a single abnormality is missed, it might lead to an adverse patient outcome. Therefore, the researchers meticulously designed the program to be nearly flawless. The research team found that the algorithm took about one second to look through an entire head imaging series and identify that a brain bleed was present. Furthermore, the computer program even outlined the abnormality in the image, making it easier for the physician to see. This can save radiologists countless hours of looking through standard images, finding one abnormality.

Moving Forward with an AI Program

Undoubtedly, the research team knows that their program needs to be fine-tuned. There are a lot of hurdles to clear before this program is rolled out, analyzing real cases next to a doctor.

A traumatic brain injury can lead to long-term complications; this program might help doctors prevent these. With more research being done in this field, it will be exciting to watch this AI program evolve.

Watch YouTube Video: Neuroscience and Artificial Intelligence Need Each Other. In this TEDx Talk segment, Marvin Chun discusses how using artificial intelligence can help with the efficiency of brain scans.

Brain Injury Lawyers in Sacramento

I’m Ed Smith, a Sacramento Brain Injury Lawyer. Artificial intelligence is helping doctors read brain scans. Individuals who have been diagnosed with a brain injury due to the negligence of another person or entity should call me at (800) 404-5400 and/or (916) 921-6400 for free, friendly legal guidance and advice.

I’m humbled to be a member of the Million Dollar Advocates Forum and the Top One Percent.

See our client reviews on Avvo and Yelp, Google, and past verdicts and/or settlements.

Citation of Photo: The picture used at the beginning of this post was located first on Pixabay & has been shown here with the guidance of the Creative Commons License.

:dr 0p cha [cs 693]