TECHNOLOGY

NEUROPHET SegEngine

AI-POWERED BRAIN SEGMENTATION TECHNOLOGY

 

Neurophet-Invented Core Technology Overcomes the Technical Algorithmic Limitation

 

뇌 MR 영상을 이용한 딥러닝 엔진을 기반으로 뇌 영역별 해부학적 특성을 학습시켜 초고속 뇌분할 및 구조분석이 가능하며, 디지털 정보로 변환합니다. 

MRI data

Deep learining engine

Brain model

Stimulation parameter guidance to concentrate the stimulation effect

on the target region using numerical optimization techniques.

​Conventional methods

Target optimization

Rearranging electrode placement

Target area

Stimulated area

Self-simulation engine to predict the stimulus effect on the brain

and visualize for various view

Target area

Stimulated area

Self-simulation engine to predict the stimulus effect on the brain

and visualize for various view

The brain MR Image based deep learning engine segments and analyzes structural information of each brain regions by anatomical characteristics, and it converts them into digital information.

MRI data

Deep learining engine

Brain model

Quantitatively analyzed information can be utilized to inspect neurodegeneration and abnormal structural changes in the brain.

FEATURES

Based on our core technologies, we are developing various medical application solutions for brain disease treatment/surgery guide, brain disease diagnosis, and brain image analysis.

 

Expertise in Neuroscience

neurology

The core engine was developed after more than 10 years of R&D by Neurophet's neural network research team composed of engineers with PhD in brain engineering.

Deep Learning

More than 4000 normative MR data has been trained

4000 DATA

Image Visualization

Automated image preprocessing eliminates ​brain MR image quality deviations

high quality

Quantitative Analysis

Segments 107 brain areas and extracts quantitative information of the brain structure in a minute

super fast

The Only One in the World

Reconstruct brain into 3D model including skull without the CT images and irregardless of ethnicity and race

special

Near-zero operation failure rate and error rate

ZERO FAILURE

  • Golden Standard as a brain segmentation tool

  • Outstanding performance than 'FreeSurfer'

Perfermance Comparison 

NEUROPHET

SegEngine

FreeSurfer*

Operation exceeding 24 hours

0%

​Average 1 minute

50% or more

Operation failure rate

0%

0/2000​

8.9%

178/2000​

Brain area error rate

1.3%

26/2000​

22.8%

456/2000​

Whether the skull is split

possible

Whole head

impossible

​Brain only​

User interface

Easy GUI

Difficulty

Script Code​

Setting parameters

​Auto

​Manual

Comparison of results with Freesurfer: Dice-coefficient 91.6%

​Compatibility verification of the world's top 3 MRI devices

INTER-SCANNER

& CENTER VARIABILITY

  • Perfectly compatible to the GE, Philips, Siemens's representative 12 MRI models without special parameter setting 

  • Compatibility with the 12 models is verified with 2000 MRI data 

Verified with three major companies' MRI machines 

manufacturer

MRI machine

Verification count

GE

Discovery MR 
Genesis Signa
Signa Excite
Signa HDx
Signa HDxT

200
41
79
89
200

Philips

Achieva
Ingenia
Intera

200
200
200

Siemens

Prisma fit
Skyra
Trio TIM
Verio

200
200
200
200