​Samsung applies deep learning to ultrasound device

Samsung's S-Detect applied to its ultrasound examination device RS80A uses big data to assist doctors in the diagnosis of breast tumours.

Samsung has applied deep learning technology in its ultrasound examination device to help doctors correctly diagnose breast tumours, the company has announced.

The deep learning algorithm, S-Detect, applied on the RS80A, uses around 10,000 cases of breast tumour examinations to assist doctors in lesion diagnosis.

It analyses the size of the tumour, whether it is benign or malevolent, and the type of cells where the sonogram is reading, and allows for more precise diagnosis thanks to it being optimized for image cognition, Samsung said.

Deep learning is a new area of machine learning research where a computer emulates the way the human brain draws out patterns from data and processes it.

At trial, a diagnosis from S-Detect showed a "high-level of conformity" to those from biopsy, said Han Boo Kyung, a doctor at Samsung Medical Center, in a statement. Han said he hoped that application of S-Detect will reduce unnecessary biopsies and correctly find malevolent tumours.

On the RS80A, new software called S-Fusion was added that compares in real-time sonogram images with those from CT or MRI. It reduces the image changes caused by the patients' breathing, Samsung said.

The upgraded RS80A will began sales in South Korea, Europe, and the Middle East and will later be commercially released in the Americas, China, and Russia, depending on when it gets certified there.

It is currently being used for testing at Mayo Clinic in the US, Charité Hospital in Germany, and Seoul National University Hospital in South Korea.

The product is a co-development of Samsung Electronics and its subsidiary Samsung Medison, which makes medical devices.