A research team of Samsung Medical Center has developed an artificial intelligence model for diagnosing Meniere's disease by analyzing images obtained by magnetic resonance imaging (MRI).

The newly developed technology is the first deep learning-based AI model to diagnose Meniere's disease. Meniere's disease is one of the common diseases that repeat symptoms such as severe dizziness, hearing loss, tinnitus, and fullness, deteriorating the quality of life.

The exact cause of the illness is unknown. Still, endolymphatic hydrops due to lymphatic circulation problems is believed to be the leading cause.

Until now, the only diagnostic criteria for Meniere's disease were subjective symptoms. Recently, however, the development of contrast-enhanced inner ear MRI has increased the significance of imaging tests. Major hospitals in Korea, including Samsung Medical Center, use this inner ear MRIs. However, it has a disadvantage because the images are complicated and time-consuming to calculate the degree of endolymphatic hydrops by hand.

To solve these problems, the research team of three experts – Professors Chung Won-ho, Cho Young-sang, and Cho Baek-hwan -- created an "Inner Ear Hydrops Estimation via Artificial Intelligence" (INHEARIT) model by using a convolutional neural network algorithm that is useful for image learning and pattern processing.

Researchers compared the results calculated by the radiologists and otolaryngologists to those computed by the AI model in MRIs of 124 real patients. The intraclass correlation coefficient of the results calculated by experts was 0,971, which was very high, the research team said in a press release Monday.

"Meniere's disease is still uncertain and takes a lot of time and effort to diagnose, although doctors have been listening to the patient's subjective medical history and some recent MRIs," the research team said.