Deep Learning AI Helps Researchers Accurately Diagnose Dystonia

According to Neurology Today, researchers have discovered a new way to use artificial intelligence (AI) to enhance medical knowledge. Through the DystoniaNet platform, which uses deep learning AI, researchers targeted and diagnosed focal dystonia. Ultimately, many believe that DystoniaNet will assist doctors in better diagnosing dystonia and serving patients. Interested in the research? An article published in the Proceedings of the National Academy of Sciences (PNAS) of the United States of America explains how the DystoniaNet platform developed an understanding of a new biomarker for dystonia: a microstructure neural network.

Deep Learning & AI

What are deep learning and AI?

To first understand why DystoniaNet is so promising for the medical field, it is first important to know what artificial intelligence (AI) and deep learning are. Investopedia describes AI as:

the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. The term may also be applied to any machine that exhibits traits associated with a human mind such as learning and problem-solving.

According to MathWorks:

Deep learning is a machine learning technique that teaches computers to do what comes naturally to humans: learn by example. In deep learning, a computer model learns to perform classification tasks directly from images, text, or sound.

So, both artificial intelligence and deep learning allow researchers to utilize technology to come up with solutions to problems. In this case, researchers used an algorithm to analyze microstructural neural networks in brain MRIs. Through this, DystoniaNet identifies potential brain differences and biomarkers to achieve a dystonia diagnosis. Altogether, DystoniaNet identified dystonia from healthy control MRIs in 98.8% of cases.


As many people in the rare disease community know, it can be extremely time consuming to seek a diagnosis for your condition. This is no different with dystonia, as there are no targeted biomarkers or standard diagnostic criteria. However, DystoniaNet may soon change this. By using deep learning AI, researchers made dystonia diagnoses within 0.036 seconds.

The team developed DystoniaNet by teaching the AI platform through MRI images. Overall, the researchers sourced raw brain imagery from 612 people. 220 were healthy control subjects. 279 patients had laryngeal dystonia; 59 had cervical dystonia; and 54 had blepharospasm dystonia. Next, researchers ran these MRIs through DystoniaNet. Then, the program used convolution, a mathematical operation, to repeatedly filter, magnify, and analyze the images.

Interestingly, DystoniaNet is not told what to look for in these images. Instead, the platform analyzes patterns, images, or signals that separate patients with dystonia from healthy controls. However, DystoniaNet was effective in marking down affected areas of the brain, such as the temporal gyrus and corpus callosum, that showed signs of dystonia. Ultimately, white matter changes highlight potential dystonia. One researcher notes that this is important as many of the changes are too small to be identified by researchers alone.

Looking Forward

As of right now, DystoniaNet can identify three forms of dystonia from healthy controls. However, there are an estimated 100 forms of dystonia. So additional deep learning in the future can help hone this platform. Dr. Richard L. Barbano, MD, PhD, FAAN also notes the importance of further testing with DystoniaNet. All of the patients that took part in the study had diagnosed dystonia. Dr. Barbano questions whether DystoniaNet can identify mild forms of dystonia, rather than more severe or previously diagnosed forms.


There are three main forms of dystonia, a chronic movement disorder characterized by involuntary muscle contractions:

  • Focal: For patients with focal dystonia, only one localized area of the body is affected. As an example, cervical dystonia affects the neck; muscle contractions pull the head forward or backwards. In oromandibular dystonia, the jaw is affected, causing sometimes sustained and painful clenched muscles.
  • Generalized: Patients with generalized dystonia experience muscle contractions in most or all of the body. For example, DYTI dystonia may start in the limbs and progress throughout the body, causing worsening disability and loss of function.
  • Segmental: Finally, patients with segmental dystonia experience symptoms in two or more related or neighboring areas. Craniofacial dystonia may cause symptoms in the eyes, mouth, and tongue, for example.

An estimated 300,000 people in North America have dystonia. There are a number of causes for this condition, such as genetics or acquiring dystonia from other disorders. However, researchers are not sure exactly what causes the brain to become unable to handle movement-related messages. Dystonia symptoms vary in severity, affected area, and frequency. However, some symptoms include:

  • Difficulty speaking
  • Poor or worsening handwriting
  • Tremors
  • Uncontrolled blinking
  • Foot cramps
  • Eyelid spasms
  • Neck cramps

Learn about dystonia.

Jessica Lynn

Jessica Lynn

Jessica Lynn has an educational background in writing and marketing. She firmly believes in the power of writing in amplifying voices, and looks forward to doing so for the rare disease community.

Share this post

Share on facebook
Share on twitter
Share on linkedin
Share on pinterest
Share on print
Share on email