Skin lesion diagnostic device created with HPC

Medical device company Bdetect, which has specialized in the research and development of skin cancer diagnostic products for more than 10 years, used the resources of the High-Performance Computing (HPC) cluster or supercomputer graphical processors (GPU), to train the diagnostic device with deep learning and artificial intelligence (AI) techniques to recognize potential malignant skin formations at an early stage.

The technology uses light of different colors to determine whether skin damage is malignant or benign and aims to simplify skin cancer detection. With a diagnostic device, family doctors can do rapid screening in their practices, making the diagnostic more accessible and faster.

This medical device detects melanoma contactless early through the developed algorithm. Diagnostics are based on many photos, which by deep learning and AI methods, form a diagnostic formula – after diagnostics, the device colors green or blue, which means – skin health is okay, red – a need for advice from a skin doctor. During the test, the patient’s skin health is analyzed after the photos taken during the test. They are compared by the device using an AI algorithm, such results come within seconds.

“At Bdetect, our goal is to democratize early skin cancer detection by providing intuitive, affordable, and accurate diagnostic tools that require no medical expertise. We are committed to saving lives through innovation, making advanced healthcare accessible to all” , Ronalds Skulme, CEO @Bdetect

Challenges & Solution

  • Skin images, especially in the multispectral range, have valuable information to help detect malignant lesions at an early stage. Deep learning AI methods can help, but require a lot of resources to train.
  • HPC GPU resources allowed to fit high-resolution models in the memory, that would be impossible on a typical high-end GPU.

Benefits

  • Cloud-based AI model provides results in seconds
  • Clinically validated on more than 4000 lesions
  • Wireless and Ergonomic design for easy of usage
  • Multispectral imaging analyzes skin at a cellular level

NB! in 2018, the developers of this medical device, scientists from LU and RTU, received an assessment from the Latvian Academy of Sciences in the contest “The Most Significant Advances in Science.”

Foto: https://www.bdetect.com/