The startup targeting stroke prediction and prevention

When someone is having a stroke, there’s no lead time for detection or treatment. Everything happens suddenly, and a second’s delay in treatment could mean life or death. According to the World Health Organization, strokes are the second leading cause of death and the third leading cause of disability worldwide.

While there’s no way to predict the occurrence and its severity, Singapore- and Melbourne-based startup See-Mode hopes to help clinicians predict the probability of their patients suffering a stroke through analyzing medical images.

The See-Mode team with Sadaf Monajemi (rightmost) and Milad Mohammadzadeh (second from right) / Photo credit: Handout

Just a few years ago, Sadaf Monajemi and Milad Mohammadzadeh were in different teams at Entrepreneur First, a startup incubation program.

Mohammadzadeh was building on his doctorate work in biomedical fluid dynamics, or the study of fluid flows in biological systems, while Monajemi was continuing her own doctorate efforts in AI, image analysis, and machine learning.

The latter soon met a well-known neurologist in Singapore who described a critical problem involving strokes: Even with multiple tests and medical images, doctors do not have enough information to make more accurate decisions about stroke prevention and treatment. If clinicians can better predict the potential for a stroke before it happens, it can be identified and prevented.

The married couple quickly figured out what this meant. “[She] realized that half of the words that this doctor was saying was related to her background … and the other half was related to my background,” Mohammadzadeh says.

They spoke with about 50 vascular surgeons and neurologists around the world to confirm if the problem was accurate and if the combination of AI and fluid mechanics could provide a solution.

In 2017, the two officially joined forces to work on what would become See-Mode.

Tech-enabled predictions

The current process of analyzing ultrasound scans is a manual one.

Sonographers and radiographers often analyze anywhere between 50 to 100 ultrasound scans per patient. It’s a time-consuming and error-prone process that can lead to incorrect or unnecessary surgeries and treatment plans.

See-Mode’s first product, Augmented Vascular Analysis or AVA, automates the analysis and reporting of blood vessel ultrasounds. With AVA, medical staff can scan the same 50 to 100 images with a push of a button.

The software applies deep learning during image analysis and optical character recognition to analyze scans. This data is combined into a more comprehensive and accurate report that can be reviewed easily.

According to the startup, its next products are still in the research stage. Instead of simply assessing the images, it hopes to use them to make stroke risk predictions.

Current mode