컴퓨터비전, 머신러닝, 딥러닝을 이용한 의료영상분석 1-3 Introduction to medical image analysis 2
컴퓨터비전, 머신러닝, 딥러닝을 이용한 의료영상분석 1-3 Introduction to medical image analysis 2
본 내용은 Edwith의 컴퓨터비전, 머신러닝, 딥러닝을 이용한 의료영상분석을 요약 정리한 내용으로 DGIST 박상현 교수님과 Edwith, STAR-MOOC에 그 저작권이 있음을 미리 공지합니다.
URL : https://www.edwith.org/medical-20200327/lecture/63119/
Artificial Intelligent
1950년도
Loop[Problem -> Rule -> Evaluation -> Error Analysis] -> (Evaluation) Launching
Machine Learning 1980년도
Rule 대신 Data 기반 Machine Learning의 Loop
Data Mining : Machine Learning Solution 분석 -> Discovery
Classification
In Medical : Normal vs Alzheimer's Disease
Segmentation
자율주행
Enhancement
해상도를 높게 바꿔줌 Super Resolution
의료 영상에서도 Quality를 높이기위한 기법 제안
Registration
뒤틀린 영상 사진(Warp Image)을 잘 맞게 맞춰줌
의료에서도
Reference Image
Target Image -> Co-Registered Image
Difference Image -> Difference Image
Imagenet : ILSVRC
DeepLearning의 메인 재료인 빅데이터 Image Database를 제공.
GPU : Parallel Computing 성능 향상, 처리속도 향상
BigData + GPU가 가져온 Breakthrough
A Roadmap for Foundational Research on Artificial Intelligence in Medical Imaging, Radiology 2019
MIA : Medical Image Analysis
앞으로 배울 내용 정리표
|
Conventional Methods |
Deep Learning Method |
Classification |
Logistic Regression Neural Network Support Vector Machine |
Deep Neural Network Convolutional Neural Network |
Segmentation |
Thresholding Region Growing Graph Cut Active Contour Model Active Shape Model |
FCN U-Net DeepLab |
Enhancement |
Normalization Histogram Equalization Filtering Dictionary Learning |
SRCNN GAN SRGAN |
Registration |
Transformation Matrix Iterative Closest Point(ICP) Non Rigid ICP Deformable Models |
FlowNet CNN for Registration |
