Sonio: Deep-Learning for Detection and Diagnostic of Prenatal Malformations

Introduction

1 child out of 33 is born with a congenital malformation in developed countries, leading to mortality and disability which impact the children, the families, and the healthcare system.

Challenges in Detection

Fetal ultrasound is the standard non-invasive examination to screen for malformations, but 50% of malformations are not detected at routine exams. This is due to several factors:

• Fetal ultrasound is very complex.
• It is time-consuming.
• It is highly operator-dependent.

To achieve an accurate diagnosis, one needs to:

1. Acquire the right images.
2. Interpret them correctly.
3. Combine them with blood and/or genetic tests.

Solution: Sonio

We created Sonio, an AI one-stop modular software platform to guide OBGYNs and sonographers during fetal ultrasound.

Core Features

The core of Sonio is the Clinical Brain, which is a unique mix of fetal medicine and AI. It is aware of:

• 1.6k anomalies
• 450 syndromes

The Clinical Brain can prioritize anomalies to identify the most probable diagnoses based on medical history and observed phenotype.

Future Developments

With EIC support, we will fully build image recognition and genomics into our platform to revolutionize prenatal diagnosis.