Skychain AI for Brain Glioma Recognition released

Skychain has recently launched the fourth Skychain AI, which is aimed to detect brain gliomas using MRI images.

The Problem

Glioma is a common type of tumor originating in the brain. About 33 percent of all brain tumors are gliomas, which originate in the glial cells that surround and support neurons in the brain, including astrocytes, oligodendrocytes and ependymal cells.

The symptoms, prognosis, and treatment of a glioma depend on the person’s age, the exact type of tumor, and the location of the tumor within the brain. These tumors tend to grow and infiltrate into the normal brain tissue, which makes surgical removal very difficult — or sometimes impossible.

Due to such variety of glioma manifestations, its segmentation is also quite a difficult task.


The Skychain Solution

The Skychain AI for brain glioma recognition available in Skychain Alpha detects gliomas by analyzing MRI images.

Magnetic resonance imaging (MRI) is a medical imaging technique used in radiology to form pictures of the anatomy and the physiological processes of the body in both health and disease. MRI scanners use strong magnetic fields, magnetic field gradients, and radio waves to generate images of the organs in the body.

For more complete and accurate segmentation, the Skychain neural network uses three-dimensional images of the brain MRI of the following four types:

  • MRI-sequence FLAIR (Fluid attenuation inversion recovery);
  • T1-weighted sequence;
  • T1c-weighted sequence with increased contrast (Contrast Enhanced);
  • T2-weighted sequence.

The U-Net is used as the AI architecture. It is built upon the Fully Convolutional Network and modified in a way that it yields better segmentation in medical imaging.

The neural network identifies the following areas of glioma:

  • Necrotic tumor core (Mark 1, red color);
  • Peritumoral edema (Mark 2, green color);
  • Enhancing core (Mark 3, purple color).
The result of segmentation is presented to a user as a three-dimensional brain model.

Maria Piliugina, who developed this neural network, reports that the current accuracy rate is about 87%. For estimating this she used dice coefficient which characterizes the similarity between two samples(true segmentation and the ANN segmentation).

Dice coefficient formula

Further improvement of network characteristics will be facilitated by the expansion of the base of training images and segmentations.

Watch the interview with Skychain developer Maria Piliugina on the AI for brain glioma recognition:

Try Skychain Alpha.

Join Skychain on social media: Twitter, Facebook, Telegram.

Iva Chernysheva, Marketing Manager