Recent advances of AI for medical imaging : Engineering perspectives

Namkug Kim
Namkug KimUniversity of Ulsan College of Medicine/Asan Medical Center um University of Ulsan College of Medicine/Asan Medical Center
Deep learning and its application for radiologists
Namkug Kim, PhD namkugkim@gmail.com
Medical Imaging N Robotics Lab. http://mirl.ulsan.ac.kr
Convergence Medicine/Radiology
Biomedical Engineering Center
Asan Medical Center/Univ. of Ulsan College of Medicine
Researches with
Hyundai Heavy Industries Co. Ltd.
LG Electronics
Coreline Soft Inc.
Osstem Implant
CGBio
VUNO
Kakaobrain
이해상충
Stockholder
Coreline Soft, Inc.
AnyMedi
Co-Founder
Somansa Inc.
Cybermed Inc.
Clinical Imaging Solution, Inc
AnyMedi, Inc.
Selected Grants as PI
한국연구재단, NRF, South Korea
7T용 4D 자기공명유속영상을 이용한 심뇌혈관 질환의
in-vivo 유동 정량화 SW개발, 2016
4D flow MRI을 이용한 심혈관 질환의 in-vivo 유동 연구,
2015-7
자기공명분광영상 및 MRI의 통합 분석 소프트웨어
개발
산업부, KEIT, South Korea
의료영상 인공지능 과제, 2016-20
3DP 척추 맞춤형 임플란트, 2016-20
3D 프린터 기반 무치악 및 두개악안면결손 환자용 수복
보철물 제작, 재건 시스템 개발, 2015-9
근골격계 복구 수술 로봇 개발, 2012-7
영상중재시술 로봇시스템 개발, 2012-7
보건복지부, KHIDI, South Korea
영상 뇌졸중 예후 예측 및 치료방침 결정 시스템 개발,
2012-8
관동맥 관류 CT 의 자동 진단 프로그램을 활용한
허혈성 질환의 진단과 치료, 2013-6
산학협력
Hyundai Heavy Industry, Osstem Implant, S&G Biotech,
Coreline soft, Midas IT, AnyMedi, Hitachi Medical, Japan,
5
MBC 다큐 스페셜 ‘미래인간 AI’, 2016.12.05
영상유도 중재 수술로봇 ; 현대중공업 폐영상 분석 SW ; 코어라인 소프트
3D 프린터 응용 ; 서울아산병원
Movie Clips
Major Breakthroughs in Feedforward NN
K. Fukushima Yann Lecun G. Hinton, S. Ruslan
Neocognitron
Neocognitron (1979)
• By Kunihiko Fukushima
• First proposed CNN
Convolutional Neural Networks (1989)
• Yann Lecun et.al
• Back propagation for CNN
• Theoretically learn any function
LeNet-5 architecture
Alex krizhevsky ,
Hinton
LeNet-5 (1998)
• Convolutional networks
Improved by Yann Lecun
et.al
• Classify handwritten digits
D. Rumelhart, G. Hinton, R.
Wiliams
1960 1970 1980 1990 2000 2010 2012
Perceptron
XOR
Problem
Golden
Age
1957 1969 1986
F. Rosenblatt
M. Minsky, S. Papert
• Adjustable weights
• Weights are not
learned
• XOR problem is not linearly
separble
• Solution to nonlinearity
separable problems
• Big computation, local
optima and overfiting
CNN Breakthrough (2012)
• By Alex Krizhevsky et al.
• Winner of ILSVRC2012 by
large marginDark Age (AI
winter)
Back propagation
(1981)
• Train multiple layers
Multi-layer
Perceptron (1986)
Why Deep Learning ?
From MS speech group
Speech Recognition
6.66%
5.98%
5.10% 4.94% 4.80%
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
7.00%
Google 2014
Baidu 2015
Human Level
MS 2015
Google 2015
Object (Image) Recognition (ImageNet)
Face Verification Gene Network Structure Inference
97.35% 97.45% 97.53%
98.52%
99.15%
99.47%
99.63%
96.00%
96.50%
97.00%
97.50%
98.00%
98.50%
99.00%
99.50%
100.00%
←Using DL
Major Components in Deep Learning
Breakthroughs
Algorithms
Parallel
computing
Big Data
Unsupervised pre-training
Supervised training for deeper models
NVIDIA® CUDA® 코어 5760
메모리 클럭 7.0 Gbp
표준 메모리 설정 12288 MB
Where do we use Deep Learning ?
Auto drive (pedestrian/traffic sign recognition)
• Netflix movie recommendation
• Language translation
• Breast cancer detection
• Skype translator
• Gesture/pose detection*
*Neverova, Natalia, et al. "Hand Pose Estimation through Weakly-Supervised Learning of a Rich
Intermediate Representation." arXiv preprint arXiv:1511.06728 (2015)
영상 인식
Image tagging,
retrievalObject recognition Scene segmentation
비디오 인식
Video understanding (Google, 2014)
Scene parsing (NYU/Facebook , 2014) NVIDIA DRIVE PX, 2015
음성 및 번역
Speech
Recognition
Machine Translation
Speech Recognition + Machine
Translation
영상 해석/자막 생성
Image Caption
Generation
Video Caption
Generation
질의 문답
Image Question
Answering
Speech Recognition + Image Questing
AnsweringVideo Question
Answering
Text Question
Answering
생성 모델 (아트)
Artistic neural style
생성 모델 (풍경)
Eyescream project in Facebook AI Research
(using Laplacian pyramid Generative adversarial
network; LAPGAN)
http://soumith.ch/eyescream/
생성 모델 (얼굴, 거실)
Deep Convolutional Generative Adversarial Networks (DCGAN)
Rotations are linear in latent space
Bedroom generation
Arithmetic on faces
잡기 로봇화 (구글)
Comparison btw Brain and NN
36
Comparison btw Brain and NN
37
1. 10 billion neurons
2. 60 trillion synapses
3. Distributed processing
4. Nonlinear processing
5. Parallel processing
6. Efficiency
(20~25W, 하루섭취량의 20~25%)
1. Faster than neuron (10-9 sec)
cf. neuron: 10-3 sec
3. Central processing
4. Arithmetic operation (linearity)
5. Relatively Sequential processing
6. Efficiency (Titan X : 250W)
cf. 1[kcal] = 1.16[Wh], 1W=1J=1Nm/s,
1cal=4.2J=1.163mWh
Brain Computer
Bio Plausible Neural Network
Mimic human visual recognition system
Each unit connected to a small subset of other units
Based on what it sees, it decides what it wants to say
Units must learn to cooperate to accomplish the task
38
Hierarchical Representations
pixels
edges
object parts
(combination
of edges)
object models
H. Lee, R. Grosse, R. Ranganath, A. Y. Ng, “Convolutional Deep Belief Networks for Scalable
Unsupervised Learning of Hierachical Representations," in ICML 2009.
From Shallow to Deep Learning
From Shallow to Deep Learning
Shallow learning
SVM
Linear & Kernel Regression
Hidden Markov Models (HMM)
Gaussian Mixture Models (GMM)
Single hidden layer MLP
Artificial Neural Net (ANN)
...
Limited modeling capability of concepts
Cannot make use of unlabeled data
특징 추출
42
Neural Networks
• Machine Learning
• Knowledge from high dimensional data
• Classification
• Input: features of data
• supervised vs unsupervised
• labeled data
• Neurons
Neural Networks (NN); 반복적인 에러교정
Forward propagation
Sum inputs, produce activation, feed-forward
Input 𝑋
Hidden
𝑊1
𝑊2 Neuron
y
𝑥1
𝑥2
𝑥3
𝑥 𝑛−1
𝑥 𝑛
…
Inputs
Output
𝑧 = 𝑏 +
𝑖
𝑥𝑖 𝑤𝑖
𝑦 = 𝐻(𝑧)
Output 𝑌 =f(x)
Scaling function Activation function Activation function
Information Propagation
Error BackPropagation
Output Comparison
Weights
Simple Perceptron
45
Simple Perceptron
46
Simple Perceptron; Non-linear
47
Need for Multiple Units and Multiple Layers
Multiple boundaries
are needed (e.g.
XOR problem) ->
Multiple Units
More complex
regions are needed
(e.g. Polygons) ->
Multiple Layers
48
From Shallow to Deep Learning
49
Best Practice
Normalization
Prevent very high weights, Oscillation
Overfitting/Generalisation
Validation Set, Early Stopping
Mini-Batch Learning
update weights with multiple input vectors combined
Challenges in Training Feedforward NN
Training multiple layers
Vanishing gradient problem
Gradients are diluted as layers go deep
Only use labeled data
most data is unlabeled
Stuck in local minima
Over-fitting
Too many hyper parameters
Problems with Backpropagation
Limitations
Get stuck in local optima
start weights from random positions
Error attenuation, long fruitless training
Slow convergence to optimum
large training set needed
Only use labeled data
most data is unlabeled
Backpropagation (BP) barely changes lower-layer
parameters (vanishing gradient)
Therefore, deep networks cannot be fully (effectively) trained
with backpropagation
52
Breakthrough with Backpropagation
Breakthrough
Recent – Long patient training with GPUs and special hardware
Deep belief networks (unsupervised pre-training)
Convolutional neural networks (reducing redundant parameters)
Rectified linear unit (constant gradient propagation)
53
Rectified Linear Units
More efficient gradient propagation, derivative is 0 or constant,
just fold into learning rate
More efficient computation: Only comparison, addition and mul
tiplication.
Leaky ReLU f(x) = x if > 0 else ax where 0 ≤ a <= 1, so that derivat
e is not 0 and can do some learning for that case.
Lots of other variations
Sparse activation: For example, in a randomly initialized network
s, only about 50% of hidden units are activated (having a non-z
ero output)
CS 678 – Deep Learning 54
Convolutional Neural Networks (CNN)
A type of feed-forward neural network
Inspired by biological process
Weight sharing (convolution) +
Subsampling (pooling)
Reducing the number of parameters (Reduce over-
fitting)
Translation invariance
Input
28 × 28
Feature maps
4@24 × 24
Feature maps
4@8 × 8
Feature maps
8@4 × 4
Feature maps
8@2 × 2
Feature maps
8 ⋅ 2 ⋅ 2 × 1
Output
10 × 1
Convolution
layer
Max-pooling
layer
Convolution
layer
Max-pooling
layer
Reshape Linear layer
[LeCun, 1998]
Convolution and pooling
57
Convolutional Neural Networks (CNN)
Convolution과 Pooling (Subsampling)을 반복하여 상위 Feature
를 구성
Convolution은 Local영역에서의 특정 Feature를 얻는 과정
Pooling은 Dimension을 줄이면서도, Translation-invariant한
Feature를 얻는 과정
Convolutional Neural Networks (CNN)
Neural network with sparse connections
Learning algorithm:
Backpropagation on convolution layers and fully-connected layers
Behavior of CNN
60
Visualization of FilterBank
VGG16 architecture
ImageNet
most filters
identical,
rotated by some non-random
factor (typically 90 degrees).
– rotation-invariant.
The rotation observation
holds true in block4_conv1.
Textures similar to that found
in the objects in
block5_conv2
61
Recurrent Neural Networks (RNNs)
26.2%
16.4%
13.5%
12.9%
11.8%
7.3%
6.7%
4.9% 4.8%
3.6%
Breakthroughs in CNN
2012 2013 2014 2015
Alexnet (2012)
• 1st place in 2012
• 5 conv layers + 3
fully connected layers
• Dropout & ReLU
ImageNet Large Scale Visual Recognition Challenge
Results
SIFT + FVs (2012)
• 2nd place in 2012
• SIFT + fisher vectors
• No CNNs
ReLU (Rectified Linear Units)
Alexnet
architecture
Data augmentation (flip,
random crop)
Dropout
26.2%
16.4%
13.5%
12.9%
11.8%
7.3%
6.7%
4.9% 4.8%
3.6%
Recent Breakthroughs in CNN
2012 2013 2014 2015
ZF Net (2013)
• 3rd place in 2013
• By Matthiew
Zeiler & Rob
Fergus
• Variant of
Alexnet
Alexnet (2012)
• 1st place in 2012
• 5 conv layers + 3
fully connected
layers
• Dropout & ReLU
Clarifai (2013)
• 1st place in 2013
• Deep Learning startup
• Founded by Matthiew
Zeiler
• Variant of Alexnet
VGG Networks (2014)
• 2nd place in 2014
• By Oxford computer vision
group
• 19 layers deep
GoogLeNet (2014)
• 1st place in 2014
• 24 layers of convolution
• Memory efficient network
ImageNet Large Scale Visual Recognition Challenge
Results
SIFT + FVs (2012)
• 2nd place in 2012
• SIFT + fisher vectors
• No CNNs
Overfeat (2013)
• 2nd place in 2013
• By NYU
• Variant of Alexnet
GoogLeNet
architecture
26.2%
16.4%
13.5%
12.9%
11.8%
7.3%
6.7%
4.9% 4.8%
3.6%
Recent Breakthroughs in CNN
2012 2013 2014 2015
ZF Net (2013)
• 3rd place in 2013
• By Matthiew
Zeiler & Rob
Fergus
• Variant of
Alexnet
Alexnet (2012)
• 1st place in 2012
• 5 conv layers + 3
fully connected
layers
• Dropout & ReLU
Clarifai (2013)
• 1st place in 2013
• Deep Learning startup
• Founded by Matthiew
Zeiler
• Variant of Alexnet
VGG Networks (2014)
• 2nd place in 2014
• By Oxford computer vision
group
• 19 layers deep
GoogLeNet (2014)
• 1st place in 2014
• 24 layers of convolution
• Memory efficient network
Batch normalization (2015)
• By Google
• Simple but powerful
normalization algorithm
Parametric ReLU (2015)
• By Facebook
ImageNet Large Scale Visual Recognition Challenge
Results
SIFT + FVs (2012)
• 2nd place in 2012
• SIFT + fisher vectors
• No CNNs
Overfeat (2013)
• 2nd place in 2013
• By NYU
• Variant of Alexnet
Parametric ReLU
Residual Learning (add skip connections)
Deep Residual Network
(2016)
• Winner of ILSVRC 2015
• By MSRA
• More than 100 layers deep
with skip connections
Recent advances of AI for medical imaging : Engineering perspectives
제안서 page 5,6
69/
61
제안서 page 1-6
70/
61
동영상 추가해야 함
제안서 page 61,62
71/6
1
제안서 page 8,93
72/6
1
제안서 page 6,101
73/6
1
Chest X-ray
75
Data cleansing SW
with AI
Gold Standard
(Generate 1000 data per class)
Manual Drawing SW
Nodule Interstitial Opacity
Consolidation Pleural Effusion
Anonymization
77
Data 수집
- 정상 1,821,455
- 비정상 287,626
DICOM header
Meta inf.
Chest (AP,Lateral)
Study description
filtering
Anonymizer
1. Batch conversion
2. 익명화
1. DICOM patient info 0008 tag
2. PatientID, SEX, AGE, NAME, Birth date, etc
3. Research ID generation
3. Efficient Error Handling
MIRL@AMC ananoymizer
Anonymizer@MIRL*
Cleansing
Data Description
- # 9,589 from 500,000
- Resize into 100x100
- Standardization
CNN
Training Data
(3,000/3,000 )
Validation Data
(1,000/1,000)
Test Data
(1000/589)
Normal Class
Abnormal Class
Model Precision Sensitivity Specificity
CNN 99.3% 99.8% 98.4%
CNNBN 100% 99.9% 100%
H. C. Shin,“Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset
Characteristics and Transfer Learning, “IEEE Transactions on Medical Imaging, 2016, pp. 1285-1298.
Chest X-ray CAM
Weakly Supervised Learning + Class Activation Map
- Resnet50 ILSVRC Pre-trained Model
Results
Interesting Case
85
Normal Cardiomegaly
Ground Truth Prediction
수술자국
Chest PA YOLO
• Fine-Tuned You Only Look Once(YOLO) model
• YOLO model-26 layers including 24 convolution layers and 2 fully
connected layers
• Only final layer training
• Only 6 classes of abnormal lesion, our last layer requires C = 6
Predicted Con.
True
Con.
1 0
1 164 56
0 50 941
1. nodule 2. consolidation 3. interstitial opacity
4. cardiomegaly 5. pleural effusion 6. pneumothorax
Predicted Con.
True
Con.
1 0
1 123 41
0 32 982
Predicted Con.
True
Con.
1 0
1 82 14
0 25 1023
Predicted Con.
True
Con.
1 0
1 414 9
0 80 691
Predicted Con.
True
Con.
1 0
1 267 32
0 43 838
Predicted Con.
True
Con.
1 0
1 55 19
0 22 1050
Results
• Nodule result ROI
1. Two nodules are detected in in Chest X-ray image
2. Nodule are only detected in Chest X-ray image
• Consolidation result ROI
1. Consolidation and pleural effusion are simultaneously detected in Chest X-ray image
2. Consolidation is only detected in Chest X-ray image
① ②
① ②
Examples
• Interstitial Opacity and Cardiomegaly result ROI
1. Two interstitial opacity are detected in Chest X-ray image.
2. Cardiomegaly is detected in Chest X-ray image .
3. Cardiomegaly and two pleural effusion are detected in Chest X-ray image.
②①
③
Examples
• Pleural Effusion result ROI
1. Cardiomegaly and Pleural Effusion are detected in Chest X-ray image
• Pneumothorax result ROI
1. Pneumothorax are detected in Chest X-ray image
2. Pneumothorax and pleural effusion are simultaneously detected in Chest X-ray image
①
②①
Examples
AI in Radiology
CAD for COPD
Compare various classifiers for OLD classification
Lee Y., et al, CMPB, 2009. 93(2): p. 206-15.
Adding shape features for accuracy enhancement of emphysema
quantification., J Digit Imaging, 22:136-148, 2009,
SPIE 2007 Honorable Mention Poster Award / JDI Best paper 2009
Compare texture-based quantification vs density-based quantification for
emphysema quantification, Investigative Radiology 43:395-402 , 2008
CAD for DILD
Study feasibility for DILD, Korean J Radiol 10:455-463, 2009
Context sensitive SVM for whole lung quantification, IFMIA 2011, IWPFI
2011, JDI 2011
CrossVendor study for ILD (GE vs Siemens), Kim N , et al, RSNA 2010, Med
Phys 2013
Texture based segmentation for iodine quantification in DECT of ILD, RSNA
2013, ER 2015
3D extension of texture analysis in DILD, RSNA 2014
CADD for DILD with regional parenchyma classification, WIP in Med Phys
Deep learning for DILD, WIP in JDI
DILD lung segmentation using deep learning, WIP in Med Phys
Noise reduction of HRCT using auto-encoder, WIP in Med Phys
Ensemble method for ILD classifier, JDI accepted
CAD for AVUS (Ultrasounds), Thyroid Nodule (Med Phys), Lymph node meta (Acta
Radiol), Pulmonary Embolism (Eur Radiol), …
Semantic Segmentation on DILD HRCT
92
Image Pattern Classification on DILD HRCT
(a) Consolidation (b) Emphysema (c) Normal
(d) Ground-glass opacity (e) Honeycombing (f) Reticular opacity
DEEP
Learning
Training dataset Testing set
Accuracy (%)
Ensemble Classifiers
GE GE 96.06 ± 1.61
Siemens Siemens 96.11 ± 1.19
GE+Siemens GE+Siemens 85.12 ± 1.91
+5~10% improvement
(breakthrough in 5 years!)
O = GE; X = SiemensSVM CNN
3D
DILD
2
Class : Normal
Raw Data ROI Extraction
3D
3 Channel
2D
2D
Results
2D (82.7%) • 3 X 2D (81.7%) • 3D (84.7%)
• 비교를 위해 augmentation 적용 안됨.
따라서 3D의 경우 variation이 큼. (많은 parameter)
CNN Airway Segmentation
80 COPD Patients’ Inspiration CT
69 CT volumes are included in training
11 CT volumes are NOT included in training
GS : Manual segmentation
107
Axial 3 slices, Sagittal 3 slices, Coronal 3 slices
32 x 32 x 3 x 3
Weights are shared
2-class
classification
CT
Volume
Probabili
ty
Volume
CNN
For each voxels inside lungs
Segment
ed
Airway
Hard thresholding (0.51) and
Select the connected component
CNN Airway Segmentation
Initial Manual
내시경 영상 판별 (대장)
NICEI실제 클래스 예측 클래스 클래스별 확률
Patch Image 결과
Whole Image 결과
결과 검토
(-)
(+)
(0)
*0의 위치는 다를 수 있음.
정답 NICEIII
NICEI CAM NICEII CAM NICEIII CAM
예측 NICEIII 이상 영역 marking
112
CBIR for Medical Images
113
The Previous Research
for Quantification Definition of Similar Lung Images
Extraction of Distribution Features Extraction of Distribution Features
* https://en.wikipedia.org/wiki/Large_margin_nearest_neighbor
* Y.J.Chang, et al,. “A support vector machine classifier reduces interscanner variation in the HRCT classification of regional
disease pattern in diffuse lung disease: Comparison to a Baysian classifier”, Medical Physics 40 (5), 051912 (2013)
114
Evaluation: Statistics
( * p-value < 0.01, ** p-value
< 0.001 )
Evaluation: Recall Accuracy
Fitting Test Data
Acknowledgement
This work was supported by the Industrial
Strategic technology development program
(10072064, Development of Novel Artificial
Intelligence Technologies To Assist Imaging
Diagnosis of Pulmonary, Hepatic, and Cardiac
Diseases and Their Integration into
Commercial Clinical PACS Platforms)
funded by the Ministry of Trade Industry and
Energy (MI, Korea)
CHALLENGES & DISCUSSIONS
115
116
AP 통신 : 인공지능(로봇)이 기사 작성, 기술 공개
초당 2,000개 기사 작성 가능
기존의 300개 기업 실적 -> 3000개 기업 실적 커버
117
Right now, about 80% of Americans who need a lawyer can't afford one
"With ROSS, lawyers can scale their abilities and start to service this very
large untapped market of Americans in need,"
GOLDMAN SACHS
118
기계파괴운동 (1811-7)
119
정밀의료(Precision Medicine) Initiative
“And that’s why we’re here today. Because something called precision
medicine … gives us one of the greatest opportunities for new medical
breakthroughs that we have ever seen.”
President Barack Obama
January 30, 2015
정밀의료 치료기기 핵심 : 빅데이터 + 인공지능
빅데이터
▶ 인공지능
▶ 환자 맞춤형 진단/치료/공공
120/27
의료 빅데이터 + 인공지능
의료 빅데이터를 이용하는
정밀의료 실현
사물인터넷 유전자검사
의료영상
환자 모니터링
하루 37만건의 의무기록 (연
간 약 1TB)
연간 약 2백만 영상데이터
(30TB)
3.8×109 염기쌍을 가진 DNA
정보
대형 종합 병원
의료비 절감의 필요성
의료효율성
국내현황
Efficacy (효용성)
Cost (비용)
Equation = Efficacy / Cost
• 정밀치료를 통한 효용성 극대화
• 효율적 의료시스템 구축을 통한 비
용 감소
• 고령화 사회 등으로 GDP대비 의료비 비율 빠르게 증가하고 있으며 상승폭 또한 커지고 있
음
• 국민 소득 및 소비수준을 고려하여 의료비는 변화에 반비례하여 감소되어야 할 필요성 존재
인공지능 관련 국내외 시장 현황
사회 및 산업 각 분야시장의 핵으
로 부상중인 인공지능
2020년까지 연평균 53.65% 성장
률을 기록할 것으로 예상1)
국내의 인공지능 시장 규모 예상
2020년 2조2천억원
2025년 11조원
2030년 27.5조원2)
1) 2016 market and market 2) KT경제경영연구소
123
Global Trends Drive Momentum in health care industry
Data Explosion 150+ exabytes Amount of healthcare data today1
Over 230K Active clinical trials2
80% Healthcare data that comes from unstructured data sources3
Dynamic
Delivery
Environment
50% Expected alternative payments form the Centers of Medicare and Medicaid by
2018 4
75%+ Percentage of patients expected to use digital health services in the future5
90K Expected shortage of physicians by 2020 6
Value vs Volume 4.7 trillion Estimated global economic impact of chronic disease by 2030 7
3 trillion Estimated US healthcare spending 8
100’s Approx. amount of decisions a person living with Type 1 Diabetes makes a day9
Efficient and
effective R&D
1 in 10 Clinical trials in cancer that are shut down due to lack of participation10
2.6B Average costs to develop a new pharma drug11
<10% Amount of drug currently in development that make it market12
124
1: NCBI: big data analytics in healthcare: promise and potential 2: ClinicalTrials.gov, 3: NIH 4 CMS 5: McKinsey Healthcare's Digital Future July 2014,
6: AAMC Report The complexities from 2014 ro 2025 7: WEF global economic burden non-communication diseases 8: Health affaires. Team analysis
9: OpenAps.org 10: Bio-clinical development success rates 20, Health Economic volume 47, May 2016
1. Life expectancy data, WHO, 2012
2. 2015 Global life sciences outlook: adapting in an era of transformation, Deloitte DTTL, 2014
3. Informa Pic Market Line Extracted Oct 2014
Opportunity
126
8 trillion : Industry
Size
2 trillion : waste in industry
Better experience
Imaging :
Unnecessary tests
Lower cost
Oncology:
Variability of Care
Better outcomes
Life sciences:
Failed clinical trials
Government:
Fraud, Waste and Abuse
Value Based Care:
Cost of chronic disease
360 billion : total IT and
healthcare market
opportunity
*IBM Watson
국제적 동향
인공지능 의료적용 현황 (해외)
1) marketandmarkets 2016.02
세계 시장 규모는 5.05조원 (2020년 예측치), 향후 5년 성장률 53.65%으로 新성장 사업분야1)
의료 문서, 영상 빅 데이터를 활용한
진단 및 처방이 가능한 기술을 개발
헬스케어 분야부터 정밀의료 산
업까지 폭넓은 기술 로드맵 보유
폐암진단을 위한 딥러닝
기반 분석 시스템
131
133
AI in healthcare
134
2011년부터 해외 인공지능 기반 의료 스타트 업 기업들은 8억7천만 달러(1조
원) 규모의 투자를 유치
인공지능 의료적용 현황 (국내)
국내 동향
• 삼성과 LG를 비롯한 IT 기업의 높은 인공지능 분야 기술력
• 세계적 수준의 임상의료 및 임상시험 기술
• 최근 뷰노코리아, 루닛 등 인공지능기반 의료분야 스타트업들이 조기성과를 보이는 등 의료분야
인공지능 산업 기반 확대
미만성폐질환에서 CT로부터 질병을 판별. 질병
판독. 질병의 진행상황, 치료법 등 전문의 판단
을 도움을 주는 시스템 개발
소아 골연령 판정 보조 시스템 개발
흉부단순촬영 등 영상 데이터를 딥러닝 기술로
학습, 결핵, 유방암등 검출하는 기술을 개발
시스템생물학에 인공지능을 결합시
켜 기존 약물 개발 과정을 개선
135
인공지능 의료적용 분야
인공지능 분야
시각지능
언어지능
판단지능
자동분류
요약/창작
공간지능
임상시험
케이스선정
신약개발프로세스
진료보조
비서서비스
음성인식 의무기록
데이터기반 정밀의료
유전체분석
약혼합사용 및 합병증 예측
진단검사추천
판독보조
정상유무판정
유사증례검색
판독문 생성
병리분야 판독 보조
물류, 수술실, 병실 운영
로봇수술
의료 인공지능
인공지능 의료적용 분야
136
인공지능 의료적용 분야 (임상 중심 분류)
정상 유무 판정
초기진단 빅 데이터와 차세대
인공지능 기법을 통해 정밀판독
이 없이도 초기 감별진단
유사증례검색
DB의 수많은 증례로부터 유사
한 증례를 검색, 시각화하여 진
단에 도움
예비 판독문 생성
인공지능 기반 의료영상 분석기
법과 자연어 처리기술을 융합,
영상전문의의 판독을 보조할 수
있는 수준의 판독문 자동생성
병리분야 판독 보조
병리영상 빅데이터에 인공지능
기법의 적용, 진단, 발병기전 분
석, 예후 예측 등에 활용
임상시험
인공지능 기술을 기반으로 단일
병원 및 병원 클러스터간 물류, 수
술실, 병실의 효율적 운영
물류, 수술실, 병실 운영
로봇수술
인공지능 기술을 통해 의료 로봇
의 수술을 계획, 위험도 예측, 침
습부위 최소화
진료보조
데이터기반 정밀의료
판독보조
신약
약물 개발과정에 인공지능 기법을 적용, 질병 치
료에 더욱 효과적인 약물의 조합과 용도 변경 탐
색, 약물 후보군 및 임상환자 군의 최적화
케이스 선정
인공지능 기반 검색기법을
통해 적합한 질병 및 환자
를 탐색, 임상 시험의 준비
기간을 단축, 객관성 향상
비서서비스
IoT기술, 음성 인식기술 및 인공지능 기술을 융합,
효율적인 예약, 진단 및 진료 프로세스, 업무 정보
업데이트 및 맞춤형 큐레이션
음성인식 의무기록
진단 및 판독 내용의 기록을 자동화,
의료분야 전문용어를 판독 및 구조
화할 수 있는 수준의 음성 인식 및
문서 생성 기술
진단 검사 추천
정밀 진단 및 치료를 위해 인공지능
및 빅데이터를 활용, 정확도를 높이
고 위험도를 낮추기 위한 추가 진단
검사 프로세스를 추천
유전체
맞춤 의료를 위해 유전체, 멀티모달 의료 영
상 및 임상병리 빅데이터를 바탕으로 연관
성을 분석, 모델링하여 예후 예측, 진단 및
치료
약 혼합 사용 및 합병증 예측
치료 및 약물 사용 시 사례기반 위험성 또는 합병
증의 위험을 알려 주어 의사의 최종 결정을 보조
137
인공지능 의료적용의 어려움과 대책
139
의료환경의
표준화의
어려움으로
인공지능 훈련을
위한 라벨링 어려움
개인정보보호에
따른 데이터의
비개방성
의료기기 인허가,
신의료기술인정 및
수가 장벽
인공지능 전문가와
의료전문가의 단절
의료빅데이터의
표준화와 연계 연구
필요
인공지능기술개발을
위한 공개 데이터
개발과 이를 이용한
그랜드 챌린지
방식의 기술 진흥
필요
연구기획 초기부터
관련 종합 대책
정책적 지원 방안
마련
인공지능 교육,
연구개발 산업화를
망라하는 생태계
구축 필요
어려움
대책
Strategy
Proof of principal
Google
Retinopathy, JAMA 2016
– 130k images, 54 ophthalmologists
– Reference standard
– Severity, image quality, L/R, FoV
Ophthalmology, Pathology, Radiology,
Dermatology
Cloud/Vision/Speech API
AI make AI
Amazon, MS
Naver
J Proj. Clova, 오감 AI
Line chat bot since 2014
141
Technical Issues on Medical Imaging
Data collection
More (clean) data
Accurate annotation
Legal issues
Models selection
Deeper network
Off the shelf models
Result Interpretations
Neural network visualization
Human-friendly interpretation
142
Machine Operable, Human Readable
Visual attention
Category – feature mapping
Sparsity and diversity
150
Machine Operable, Human Readable
Evidence hotpot for lesion visualization
“SpineNet: Automatically Pinpointing Classification
Evidence in Spinal MRIs”
151
Machine Operable, Human Readable
Visualization of salient region in bone x-ray
152
MD-friendly Interpretation
Breast cancer risk prediction through BI-RADS categorization of
mammography
Analysis and visualization for breast density prediction
Mapping and visualization of patient by predicted breast cancer risk
score
153
MD-friendly Interpretation
154
Learning to read chest x-ray
Automated annotation with attention and visual-
sentence mapping
MD-friendly Interpretation
155
Contents-based case retrieval
Suggest similar cases with the clinically matching
context
백문이 불여일타(百聞而不如一打)
http://playground.tensorflow.org/
165
167
Collaborators
• MIRL
• Clinical Collaborators@Asan Medical Center, SNU BH
– Radiology : Chest, Cardiac, Abdomen, Neuro/Brain
– Neurology :Dongwha Kang, Chongsik Lee, Jaehong Lee, Sangbeom
Jun, Misun Kwon, Beomjun Kim
– Cardiology ; Jaekwan Song, Jongmin Song, Younghak Kim
– Internal Medicine : Jeongsik Byeon
– Pathology : Hyunhee Go
– Surgery : Bumsuk Go, JongHun Jeong, Songchuk Kim
• Academy
• KAIST EE, CS, Math
• Related Companies
• LG, VUNO, Coreline Soft, MIDAS IT, KakaoBrain
1 von 103

Recomendados

AI in Quality Control: How to do visual inspection with AI von
AI in Quality Control: How to do visual inspection with AIAI in Quality Control: How to do visual inspection with AI
AI in Quality Control: How to do visual inspection with AISkyl.ai
455 views33 Folien
Deep learning von
Deep learningDeep learning
Deep learningHatim EL-QADDOURY
346 views12 Folien
Autoencoders Tutorial | Autoencoders In Deep Learning | Tensorflow Training |... von
Autoencoders Tutorial | Autoencoders In Deep Learning | Tensorflow Training |...Autoencoders Tutorial | Autoencoders In Deep Learning | Tensorflow Training |...
Autoencoders Tutorial | Autoencoders In Deep Learning | Tensorflow Training |...Edureka!
1.2K views50 Folien
Liver segmentation using U-net: Practical issues @ SNU-TF von
Liver segmentation using U-net: Practical issues @ SNU-TFLiver segmentation using U-net: Practical issues @ SNU-TF
Liver segmentation using U-net: Practical issues @ SNU-TFWonjoongCheon
973 views41 Folien
Deep learning von
Deep learningDeep learning
Deep learningBenha University
8.4K views36 Folien
Intro to Deep learning - Autoencoders von
Intro to Deep learning - Autoencoders Intro to Deep learning - Autoencoders
Intro to Deep learning - Autoencoders Akash Goel
5.8K views65 Folien

Más contenido relacionado

Was ist angesagt?

Artificial Intelligence in Medicine von
Artificial Intelligence in MedicineArtificial Intelligence in Medicine
Artificial Intelligence in MedicineNancy Gertrudiz
761 views19 Folien
Techniques of Brain Cancer Detection from MRI using Machine Learning von
Techniques of Brain Cancer Detection from MRI using Machine LearningTechniques of Brain Cancer Detection from MRI using Machine Learning
Techniques of Brain Cancer Detection from MRI using Machine LearningIRJET Journal
663 views6 Folien
Autoencoder von
AutoencoderAutoencoder
AutoencoderWataru Hirota
1.3K views32 Folien
Plant disease detection using machine learning algorithm-1.pptx von
Plant disease detection using machine learning algorithm-1.pptxPlant disease detection using machine learning algorithm-1.pptx
Plant disease detection using machine learning algorithm-1.pptxRummanHajira
1.4K views28 Folien
AI: AI & Problem Solving von
AI: AI & Problem SolvingAI: AI & Problem Solving
AI: AI & Problem SolvingDataminingTools Inc
36.4K views17 Folien
Supervised learning network von
Supervised learning networkSupervised learning network
Supervised learning networkDr. C.V. Suresh Babu
79 views61 Folien

Was ist angesagt?(20)

Artificial Intelligence in Medicine von Nancy Gertrudiz
Artificial Intelligence in MedicineArtificial Intelligence in Medicine
Artificial Intelligence in Medicine
Nancy Gertrudiz761 views
Techniques of Brain Cancer Detection from MRI using Machine Learning von IRJET Journal
Techniques of Brain Cancer Detection from MRI using Machine LearningTechniques of Brain Cancer Detection from MRI using Machine Learning
Techniques of Brain Cancer Detection from MRI using Machine Learning
IRJET Journal663 views
Plant disease detection using machine learning algorithm-1.pptx von RummanHajira
Plant disease detection using machine learning algorithm-1.pptxPlant disease detection using machine learning algorithm-1.pptx
Plant disease detection using machine learning algorithm-1.pptx
RummanHajira1.4K views
"Introduction to Machine Learning and its Applications" at sapthgiri engineer... von Sachin Nagargoje
"Introduction to Machine Learning and its Applications" at sapthgiri engineer..."Introduction to Machine Learning and its Applications" at sapthgiri engineer...
"Introduction to Machine Learning and its Applications" at sapthgiri engineer...
Sachin Nagargoje1.4K views
Google Media Pipe and Computer Vision von Matias Iacono
Google Media Pipe and Computer VisionGoogle Media Pipe and Computer Vision
Google Media Pipe and Computer Vision
Matias Iacono548 views
TinyML: Machine Learning for Microcontrollers von Robert John
TinyML: Machine Learning for MicrocontrollersTinyML: Machine Learning for Microcontrollers
TinyML: Machine Learning for Microcontrollers
Robert John472 views
The rise of AI in medical imaging von Ran Klein
The rise of AI in medical imagingThe rise of AI in medical imaging
The rise of AI in medical imaging
Ran Klein929 views
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS von ESCOM
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMSNEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS
ESCOM3.8K views
Stock Market Prediction using Machine Learning von Aravind Balaji
Stock Market Prediction using Machine LearningStock Market Prediction using Machine Learning
Stock Market Prediction using Machine Learning
Aravind Balaji28.6K views
Neural Network Based Brain Tumor Detection using MR Images von Aisha Kalsoom
Neural Network Based Brain Tumor Detection using MR ImagesNeural Network Based Brain Tumor Detection using MR Images
Neural Network Based Brain Tumor Detection using MR Images
Aisha Kalsoom7.4K views
AI in Healthcare von Peter Morgan
AI in HealthcareAI in Healthcare
AI in Healthcare
Peter Morgan14.8K views
Classification of Apple diseases through machine learning von Muqaddas Bin Tahir
Classification of Apple diseases through machine learningClassification of Apple diseases through machine learning
Classification of Apple diseases through machine learning
Muqaddas Bin Tahir1.7K views

Destacado

영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표 von
영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표
영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표Youngjun Kim
3.8K views51 Folien
ICT in Healthcare von
ICT in HealthcareICT in Healthcare
ICT in HealthcareNawanan Theera-Ampornpunt
9.1K views113 Folien
Health 4.0 von
Health 4.0Health 4.0
Health 4.0Napier University
8.5K views38 Folien
정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블) von
정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블)정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블)
정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블)Hyung Jin Choi
696 views128 Folien
(2017/06)Practical points of deep learning for medical imaging von
(2017/06)Practical points of deep learning for medical imaging(2017/06)Practical points of deep learning for medical imaging
(2017/06)Practical points of deep learning for medical imagingKyuhwan Jung
7.3K views84 Folien
2017 상반기 미디어 이슈 von
2017 상반기 미디어 이슈2017 상반기 미디어 이슈
2017 상반기 미디어 이슈Nasmedia
3.4K views26 Folien

Destacado(6)

영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표 von Youngjun Kim
영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표
영상기반 딥러닝 의료 분야 응용 (KIST 김영준) - 2017 대한의료영상학회 발표
Youngjun Kim3.8K views
정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블) von Hyung Jin Choi
정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블)정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블)
정밀의료와 다차원 의료데이터(유전자, Ehr, 국가자료, 영상, 센서-웨어러블)
Hyung Jin Choi696 views
(2017/06)Practical points of deep learning for medical imaging von Kyuhwan Jung
(2017/06)Practical points of deep learning for medical imaging(2017/06)Practical points of deep learning for medical imaging
(2017/06)Practical points of deep learning for medical imaging
Kyuhwan Jung7.3K views
2017 상반기 미디어 이슈 von Nasmedia
2017 상반기 미디어 이슈2017 상반기 미디어 이슈
2017 상반기 미디어 이슈
Nasmedia3.4K views

Similar a Recent advances of AI for medical imaging : Engineering perspectives

Deep learning lecture - part 1 (basics, CNN) von
Deep learning lecture - part 1 (basics, CNN)Deep learning lecture - part 1 (basics, CNN)
Deep learning lecture - part 1 (basics, CNN)SungminYou
648 views39 Folien
AI Class Topic 6: Easy Way to Learn Deep Learning AI Technologies von
AI Class Topic 6: Easy Way to Learn Deep Learning AI TechnologiesAI Class Topic 6: Easy Way to Learn Deep Learning AI Technologies
AI Class Topic 6: Easy Way to Learn Deep Learning AI TechnologiesValue Amplify Consulting
107 views61 Folien
IRJET-Breast Cancer Detection using Convolution Neural Network von
IRJET-Breast Cancer Detection using Convolution Neural NetworkIRJET-Breast Cancer Detection using Convolution Neural Network
IRJET-Breast Cancer Detection using Convolution Neural NetworkIRJET Journal
83 views3 Folien
Convolutional neural networks 이론과 응용 von
Convolutional neural networks 이론과 응용Convolutional neural networks 이론과 응용
Convolutional neural networks 이론과 응용홍배 김
2.2K views63 Folien
Image classification with Deep Neural Networks von
Image classification with Deep Neural NetworksImage classification with Deep Neural Networks
Image classification with Deep Neural NetworksYogendra Tamang
7.6K views29 Folien
Reservoir computing fast deep learning for sequences von
Reservoir computing   fast deep learning for sequencesReservoir computing   fast deep learning for sequences
Reservoir computing fast deep learning for sequencesClaudio Gallicchio
393 views56 Folien

Similar a Recent advances of AI for medical imaging : Engineering perspectives(20)

Deep learning lecture - part 1 (basics, CNN) von SungminYou
Deep learning lecture - part 1 (basics, CNN)Deep learning lecture - part 1 (basics, CNN)
Deep learning lecture - part 1 (basics, CNN)
SungminYou648 views
IRJET-Breast Cancer Detection using Convolution Neural Network von IRJET Journal
IRJET-Breast Cancer Detection using Convolution Neural NetworkIRJET-Breast Cancer Detection using Convolution Neural Network
IRJET-Breast Cancer Detection using Convolution Neural Network
IRJET Journal83 views
Convolutional neural networks 이론과 응용 von 홍배 김
Convolutional neural networks 이론과 응용Convolutional neural networks 이론과 응용
Convolutional neural networks 이론과 응용
홍배 김2.2K views
Image classification with Deep Neural Networks von Yogendra Tamang
Image classification with Deep Neural NetworksImage classification with Deep Neural Networks
Image classification with Deep Neural Networks
Yogendra Tamang7.6K views
Reservoir computing fast deep learning for sequences von Claudio Gallicchio
Reservoir computing   fast deep learning for sequencesReservoir computing   fast deep learning for sequences
Reservoir computing fast deep learning for sequences
Claudio Gallicchio393 views
Machine learning in science and industry — day 4 von arogozhnikov
Machine learning in science and industry — day 4Machine learning in science and industry — day 4
Machine learning in science and industry — day 4
arogozhnikov8.7K views
Generalizing Scientific Machine Learning and Differentiable Simulation Beyond... von Chris Rackauckas
Generalizing Scientific Machine Learning and Differentiable Simulation Beyond...Generalizing Scientific Machine Learning and Differentiable Simulation Beyond...
Generalizing Scientific Machine Learning and Differentiable Simulation Beyond...
Chris Rackauckas17 views
Deep Learning - 인공지능 기계학습의 새로운 트랜드 :김인중 von datasciencekorea
Deep Learning - 인공지능 기계학습의 새로운 트랜드 :김인중Deep Learning - 인공지능 기계학습의 새로운 트랜드 :김인중
Deep Learning - 인공지능 기계학습의 새로운 트랜드 :김인중
datasciencekorea8.6K views
Improving Hardware Efficiency for DNN Applications von Chester Chen
Improving Hardware Efficiency for DNN ApplicationsImproving Hardware Efficiency for DNN Applications
Improving Hardware Efficiency for DNN Applications
Chester Chen1.5K views
MULTI-DOMAIN UNPAIRED ULTRASOUND IMAGE ARTIFACT REMOVAL USING A SINGLE CONVOL... von JaeyoungHuh2
MULTI-DOMAIN UNPAIRED ULTRASOUND IMAGE ARTIFACT REMOVAL USING A SINGLE CONVOL...MULTI-DOMAIN UNPAIRED ULTRASOUND IMAGE ARTIFACT REMOVAL USING A SINGLE CONVOL...
MULTI-DOMAIN UNPAIRED ULTRASOUND IMAGE ARTIFACT REMOVAL USING A SINGLE CONVOL...
JaeyoungHuh217 views
FACE RECOGNITION USING ELM-LRF von Aras Masood
FACE RECOGNITION USING ELM-LRFFACE RECOGNITION USING ELM-LRF
FACE RECOGNITION USING ELM-LRF
Aras Masood77 views
imageclassification-160206090009.pdf von KammetaJoshna
imageclassification-160206090009.pdfimageclassification-160206090009.pdf
imageclassification-160206090009.pdf
KammetaJoshna7 views
Details of Lazy Deep Learning for Images Recognition in ZZ Photo app von PAY2 YOU
Details of Lazy Deep Learning for Images Recognition in ZZ Photo appDetails of Lazy Deep Learning for Images Recognition in ZZ Photo app
Details of Lazy Deep Learning for Images Recognition in ZZ Photo app
PAY2 YOU9.3K views
(Research Note) Delving deeper into convolutional neural networks for camera ... von Jacky Liu
(Research Note) Delving deeper into convolutional neural networks for camera ...(Research Note) Delving deeper into convolutional neural networks for camera ...
(Research Note) Delving deeper into convolutional neural networks for camera ...
Jacky Liu281 views
Deep Learning in Recommender Systems - RecSys Summer School 2017 von Balázs Hidasi
Deep Learning in Recommender Systems - RecSys Summer School 2017Deep Learning in Recommender Systems - RecSys Summer School 2017
Deep Learning in Recommender Systems - RecSys Summer School 2017
Balázs Hidasi11.3K views

Más de Namkug Kim

의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptx von
의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptx의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptx
의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptxNamkug Kim
303 views48 Folien
젊은 연구자를 위한 국가과제 연구비 소개 2020 von
젊은 연구자를 위한 국가과제 연구비 소개 2020젊은 연구자를 위한 국가과제 연구비 소개 2020
젊은 연구자를 위한 국가과제 연구비 소개 2020Namkug Kim
520 views72 Folien
Ccids 2019 cutting edges of ai technology in medicine von
Ccids 2019 cutting edges of ai technology in medicineCcids 2019 cutting edges of ai technology in medicine
Ccids 2019 cutting edges of ai technology in medicineNamkug Kim
926 views137 Folien
Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3 von
Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3
Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3Namkug Kim
5.4K views164 Folien
젊은 연구자를 위한 국가과제 연구비 소개_2019 von
젊은 연구자를 위한 국가과제 연구비 소개_2019젊은 연구자를 위한 국가과제 연구비 소개_2019
젊은 연구자를 위한 국가과제 연구비 소개_2019Namkug Kim
1.2K views67 Folien
남과 다른 사진찍기 von
남과 다른 사진찍기남과 다른 사진찍기
남과 다른 사진찍기Namkug Kim
246 views120 Folien

Más de Namkug Kim(20)

의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptx von Namkug Kim
의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptx의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptx
의료용 디지털트윈의 정의, 주요 동향(정책, 기술)과 바이오헬스 분야 응용.pptx
Namkug Kim303 views
젊은 연구자를 위한 국가과제 연구비 소개 2020 von Namkug Kim
젊은 연구자를 위한 국가과제 연구비 소개 2020젊은 연구자를 위한 국가과제 연구비 소개 2020
젊은 연구자를 위한 국가과제 연구비 소개 2020
Namkug Kim520 views
Ccids 2019 cutting edges of ai technology in medicine von Namkug Kim
Ccids 2019 cutting edges of ai technology in medicineCcids 2019 cutting edges of ai technology in medicine
Ccids 2019 cutting edges of ai technology in medicine
Namkug Kim926 views
Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3 von Namkug Kim
Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3
Raai 2019 clinical unmet needs and its solutions of deep learning in medicine3
Namkug Kim5.4K views
젊은 연구자를 위한 국가과제 연구비 소개_2019 von Namkug Kim
젊은 연구자를 위한 국가과제 연구비 소개_2019젊은 연구자를 위한 국가과제 연구비 소개_2019
젊은 연구자를 위한 국가과제 연구비 소개_2019
Namkug Kim1.2K views
남과 다른 사진찍기 von Namkug Kim
남과 다른 사진찍기남과 다른 사진찍기
남과 다른 사진찍기
Namkug Kim246 views
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives von Namkug Kim
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' PerspectivesIFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives
Namkug Kim1K views
3D 프린팅 수가정책과제 발표자료 (심평원) von Namkug Kim
3D 프린팅 수가정책과제 발표자료 (심평원)3D 프린팅 수가정책과제 발표자료 (심평원)
3D 프린팅 수가정책과제 발표자료 (심평원)
Namkug Kim1.7K views
Interpretability and informatics of deep learning in medical images3 von Namkug Kim
Interpretability and informatics of deep learning in medical images3Interpretability and informatics of deep learning in medical images3
Interpretability and informatics of deep learning in medical images3
Namkug Kim560 views
인공지능 논문작성과 심사에관한요령 von Namkug Kim
인공지능 논문작성과 심사에관한요령인공지능 논문작성과 심사에관한요령
인공지능 논문작성과 심사에관한요령
Namkug Kim2.5K views
연구자를 위한 국가과제 및 연구실 운영 소개 von Namkug Kim
연구자를 위한 국가과제 및 연구실 운영 소개연구자를 위한 국가과제 및 연구실 운영 소개
연구자를 위한 국가과제 및 연구실 운영 소개
Namkug Kim3.1K views
4차산업혁명시대의 의과학교육 von Namkug Kim
4차산업혁명시대의 의과학교육4차산업혁명시대의 의과학교육
4차산업혁명시대의 의과학교육
Namkug Kim2K views
4차산업시대와 의료환경 von Namkug Kim
4차산업시대와 의료환경 4차산업시대와 의료환경
4차산업시대와 의료환경
Namkug Kim630 views
Medical Ai perspectives von Namkug Kim
Medical Ai perspectivesMedical Ai perspectives
Medical Ai perspectives
Namkug Kim3.5K views
뇌인지과학과 뇌질환 원인규명 및 치료법 연구 ; von Namkug Kim
뇌인지과학과 뇌질환 원인규명 및 치료법 연구 ; 뇌인지과학과 뇌질환 원인규명 및 치료법 연구 ;
뇌인지과학과 뇌질환 원인규명 및 치료법 연구 ;
Namkug Kim893 views
의료 분야에서의 3DP communicator 역할 (코어라인소프트 장세명이사) von Namkug Kim
의료 분야에서의 3DP communicator 역할 (코어라인소프트 장세명이사)의료 분야에서의 3DP communicator 역할 (코어라인소프트 장세명이사)
의료 분야에서의 3DP communicator 역할 (코어라인소프트 장세명이사)
Namkug Kim1.6K views
최근 3D 프린팅 관련 식약처 규제완화 정책 및 법령 정비 사항 von Namkug Kim
최근 3D 프린팅 관련 식약처 규제완화 정책 및 법령 정비 사항최근 3D 프린팅 관련 식약처 규제완화 정책 및 법령 정비 사항
최근 3D 프린팅 관련 식약처 규제완화 정책 및 법령 정비 사항
Namkug Kim2.1K views
Basics of computational assessment for COPD: IWPFI 2017 von Namkug Kim
Basics of computational assessment for COPD: IWPFI 2017Basics of computational assessment for COPD: IWPFI 2017
Basics of computational assessment for COPD: IWPFI 2017
Namkug Kim1.4K views
공간정보(Gis) 기반의 multi national mortality data 및 다변량 데이터 탐색적 분석 기법 von Namkug Kim
공간정보(Gis) 기반의 multi national mortality data 및 다변량 데이터 탐색적 분석 기법공간정보(Gis) 기반의 multi national mortality data 및 다변량 데이터 탐색적 분석 기법
공간정보(Gis) 기반의 multi national mortality data 및 다변량 데이터 탐색적 분석 기법
Namkug Kim441 views
의료영역에서의3D 프린팅적용을위한의료영상모델링 von Namkug Kim
의료영역에서의3D 프린팅적용을위한의료영상모델링의료영역에서의3D 프린팅적용을위한의료영상모델링
의료영역에서의3D 프린팅적용을위한의료영상모델링
Namkug Kim3.5K views

Último

Stone Masonry and Brick Masonry.pdf von
Stone Masonry and Brick Masonry.pdfStone Masonry and Brick Masonry.pdf
Stone Masonry and Brick Masonry.pdfMohammed Abdullah Laskar
20 views6 Folien
802.11 Computer Networks von
802.11 Computer Networks802.11 Computer Networks
802.11 Computer NetworksTusharChoudhary72015
9 views33 Folien
Activated sludge process .pdf von
Activated sludge process .pdfActivated sludge process .pdf
Activated sludge process .pdf8832RafiyaAltaf
8 views32 Folien
DevOps to DevSecOps: Enhancing Software Security Throughout The Development L... von
DevOps to DevSecOps: Enhancing Software Security Throughout The Development L...DevOps to DevSecOps: Enhancing Software Security Throughout The Development L...
DevOps to DevSecOps: Enhancing Software Security Throughout The Development L...Anowar Hossain
12 views34 Folien
An approach of ontology and knowledge base for railway maintenance von
An approach of ontology and knowledge base for railway maintenanceAn approach of ontology and knowledge base for railway maintenance
An approach of ontology and knowledge base for railway maintenanceIJECEIAES
12 views14 Folien
Pull down shoulder press final report docx (1).pdf von
Pull down shoulder press final report docx (1).pdfPull down shoulder press final report docx (1).pdf
Pull down shoulder press final report docx (1).pdfComsat Universal Islamabad Wah Campus
10 views25 Folien

Último(20)

DevOps to DevSecOps: Enhancing Software Security Throughout The Development L... von Anowar Hossain
DevOps to DevSecOps: Enhancing Software Security Throughout The Development L...DevOps to DevSecOps: Enhancing Software Security Throughout The Development L...
DevOps to DevSecOps: Enhancing Software Security Throughout The Development L...
Anowar Hossain12 views
An approach of ontology and knowledge base for railway maintenance von IJECEIAES
An approach of ontology and knowledge base for railway maintenanceAn approach of ontology and knowledge base for railway maintenance
An approach of ontology and knowledge base for railway maintenance
IJECEIAES12 views
Update 42 models(Diode/General ) in SPICE PARK(DEC2023) von Tsuyoshi Horigome
Update 42 models(Diode/General ) in SPICE PARK(DEC2023)Update 42 models(Diode/General ) in SPICE PARK(DEC2023)
Update 42 models(Diode/General ) in SPICE PARK(DEC2023)
NEW SUPPLIERS SUPPLIES (copie).pdf von georgesradjou
NEW SUPPLIERS SUPPLIES (copie).pdfNEW SUPPLIERS SUPPLIES (copie).pdf
NEW SUPPLIERS SUPPLIES (copie).pdf
georgesradjou14 views
_MAKRIADI-FOTEINI_diploma thesis.pptx von fotinimakriadi
_MAKRIADI-FOTEINI_diploma thesis.pptx_MAKRIADI-FOTEINI_diploma thesis.pptx
_MAKRIADI-FOTEINI_diploma thesis.pptx
fotinimakriadi7 views
A multi-microcontroller-based hardware for deploying Tiny machine learning mo... von IJECEIAES
A multi-microcontroller-based hardware for deploying Tiny machine learning mo...A multi-microcontroller-based hardware for deploying Tiny machine learning mo...
A multi-microcontroller-based hardware for deploying Tiny machine learning mo...
IJECEIAES13 views
Generative AI Models & Their Applications von SN
Generative AI Models & Their ApplicationsGenerative AI Models & Their Applications
Generative AI Models & Their Applications
SN6 views
Thermal aware task assignment for multicore processors using genetic algorithm von IJECEIAES
Thermal aware task assignment for multicore processors using genetic algorithm Thermal aware task assignment for multicore processors using genetic algorithm
Thermal aware task assignment for multicore processors using genetic algorithm
IJECEIAES31 views
zincalume water storage tank design.pdf von 3D LABS
zincalume water storage tank design.pdfzincalume water storage tank design.pdf
zincalume water storage tank design.pdf
3D LABS5 views

Recent advances of AI for medical imaging : Engineering perspectives

  • 1. Deep learning and its application for radiologists Namkug Kim, PhD namkugkim@gmail.com Medical Imaging N Robotics Lab. http://mirl.ulsan.ac.kr Convergence Medicine/Radiology Biomedical Engineering Center Asan Medical Center/Univ. of Ulsan College of Medicine
  • 2. Researches with Hyundai Heavy Industries Co. Ltd. LG Electronics Coreline Soft Inc. Osstem Implant CGBio VUNO Kakaobrain 이해상충 Stockholder Coreline Soft, Inc. AnyMedi Co-Founder Somansa Inc. Cybermed Inc. Clinical Imaging Solution, Inc AnyMedi, Inc. Selected Grants as PI 한국연구재단, NRF, South Korea 7T용 4D 자기공명유속영상을 이용한 심뇌혈관 질환의 in-vivo 유동 정량화 SW개발, 2016 4D flow MRI을 이용한 심혈관 질환의 in-vivo 유동 연구, 2015-7 자기공명분광영상 및 MRI의 통합 분석 소프트웨어 개발 산업부, KEIT, South Korea 의료영상 인공지능 과제, 2016-20 3DP 척추 맞춤형 임플란트, 2016-20 3D 프린터 기반 무치악 및 두개악안면결손 환자용 수복 보철물 제작, 재건 시스템 개발, 2015-9 근골격계 복구 수술 로봇 개발, 2012-7 영상중재시술 로봇시스템 개발, 2012-7 보건복지부, KHIDI, South Korea 영상 뇌졸중 예후 예측 및 치료방침 결정 시스템 개발, 2012-8 관동맥 관류 CT 의 자동 진단 프로그램을 활용한 허혈성 질환의 진단과 치료, 2013-6 산학협력 Hyundai Heavy Industry, Osstem Implant, S&G Biotech, Coreline soft, Midas IT, AnyMedi, Hitachi Medical, Japan,
  • 3. 5 MBC 다큐 스페셜 ‘미래인간 AI’, 2016.12.05 영상유도 중재 수술로봇 ; 현대중공업 폐영상 분석 SW ; 코어라인 소프트 3D 프린터 응용 ; 서울아산병원 Movie Clips
  • 4. Major Breakthroughs in Feedforward NN K. Fukushima Yann Lecun G. Hinton, S. Ruslan Neocognitron Neocognitron (1979) • By Kunihiko Fukushima • First proposed CNN Convolutional Neural Networks (1989) • Yann Lecun et.al • Back propagation for CNN • Theoretically learn any function LeNet-5 architecture Alex krizhevsky , Hinton LeNet-5 (1998) • Convolutional networks Improved by Yann Lecun et.al • Classify handwritten digits D. Rumelhart, G. Hinton, R. Wiliams 1960 1970 1980 1990 2000 2010 2012 Perceptron XOR Problem Golden Age 1957 1969 1986 F. Rosenblatt M. Minsky, S. Papert • Adjustable weights • Weights are not learned • XOR problem is not linearly separble • Solution to nonlinearity separable problems • Big computation, local optima and overfiting CNN Breakthrough (2012) • By Alex Krizhevsky et al. • Winner of ILSVRC2012 by large marginDark Age (AI winter) Back propagation (1981) • Train multiple layers Multi-layer Perceptron (1986)
  • 5. Why Deep Learning ? From MS speech group Speech Recognition 6.66% 5.98% 5.10% 4.94% 4.80% 0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% 7.00% Google 2014 Baidu 2015 Human Level MS 2015 Google 2015 Object (Image) Recognition (ImageNet) Face Verification Gene Network Structure Inference 97.35% 97.45% 97.53% 98.52% 99.15% 99.47% 99.63% 96.00% 96.50% 97.00% 97.50% 98.00% 98.50% 99.00% 99.50% 100.00% ←Using DL
  • 6. Major Components in Deep Learning Breakthroughs Algorithms Parallel computing Big Data Unsupervised pre-training Supervised training for deeper models NVIDIA® CUDA® 코어 5760 메모리 클럭 7.0 Gbp 표준 메모리 설정 12288 MB
  • 7. Where do we use Deep Learning ? Auto drive (pedestrian/traffic sign recognition) • Netflix movie recommendation • Language translation • Breast cancer detection • Skype translator • Gesture/pose detection* *Neverova, Natalia, et al. "Hand Pose Estimation through Weakly-Supervised Learning of a Rich Intermediate Representation." arXiv preprint arXiv:1511.06728 (2015)
  • 8. 영상 인식 Image tagging, retrievalObject recognition Scene segmentation
  • 9. 비디오 인식 Video understanding (Google, 2014) Scene parsing (NYU/Facebook , 2014) NVIDIA DRIVE PX, 2015
  • 10. 음성 및 번역 Speech Recognition Machine Translation Speech Recognition + Machine Translation
  • 11. 영상 해석/자막 생성 Image Caption Generation Video Caption Generation
  • 12. 질의 문답 Image Question Answering Speech Recognition + Image Questing AnsweringVideo Question Answering Text Question Answering
  • 14. 생성 모델 (풍경) Eyescream project in Facebook AI Research (using Laplacian pyramid Generative adversarial network; LAPGAN) http://soumith.ch/eyescream/
  • 15. 생성 모델 (얼굴, 거실) Deep Convolutional Generative Adversarial Networks (DCGAN) Rotations are linear in latent space Bedroom generation Arithmetic on faces
  • 17. Comparison btw Brain and NN 36
  • 18. Comparison btw Brain and NN 37 1. 10 billion neurons 2. 60 trillion synapses 3. Distributed processing 4. Nonlinear processing 5. Parallel processing 6. Efficiency (20~25W, 하루섭취량의 20~25%) 1. Faster than neuron (10-9 sec) cf. neuron: 10-3 sec 3. Central processing 4. Arithmetic operation (linearity) 5. Relatively Sequential processing 6. Efficiency (Titan X : 250W) cf. 1[kcal] = 1.16[Wh], 1W=1J=1Nm/s, 1cal=4.2J=1.163mWh Brain Computer
  • 19. Bio Plausible Neural Network Mimic human visual recognition system Each unit connected to a small subset of other units Based on what it sees, it decides what it wants to say Units must learn to cooperate to accomplish the task 38
  • 20. Hierarchical Representations pixels edges object parts (combination of edges) object models H. Lee, R. Grosse, R. Ranganath, A. Y. Ng, “Convolutional Deep Belief Networks for Scalable Unsupervised Learning of Hierachical Representations," in ICML 2009.
  • 21. From Shallow to Deep Learning
  • 22. From Shallow to Deep Learning Shallow learning SVM Linear & Kernel Regression Hidden Markov Models (HMM) Gaussian Mixture Models (GMM) Single hidden layer MLP Artificial Neural Net (ANN) ... Limited modeling capability of concepts Cannot make use of unlabeled data
  • 24. Neural Networks • Machine Learning • Knowledge from high dimensional data • Classification • Input: features of data • supervised vs unsupervised • labeled data • Neurons
  • 25. Neural Networks (NN); 반복적인 에러교정 Forward propagation Sum inputs, produce activation, feed-forward Input 𝑋 Hidden 𝑊1 𝑊2 Neuron y 𝑥1 𝑥2 𝑥3 𝑥 𝑛−1 𝑥 𝑛 … Inputs Output 𝑧 = 𝑏 + 𝑖 𝑥𝑖 𝑤𝑖 𝑦 = 𝐻(𝑧) Output 𝑌 =f(x) Scaling function Activation function Activation function Information Propagation Error BackPropagation Output Comparison Weights
  • 29. Need for Multiple Units and Multiple Layers Multiple boundaries are needed (e.g. XOR problem) -> Multiple Units More complex regions are needed (e.g. Polygons) -> Multiple Layers 48
  • 30. From Shallow to Deep Learning 49
  • 31. Best Practice Normalization Prevent very high weights, Oscillation Overfitting/Generalisation Validation Set, Early Stopping Mini-Batch Learning update weights with multiple input vectors combined
  • 32. Challenges in Training Feedforward NN Training multiple layers Vanishing gradient problem Gradients are diluted as layers go deep Only use labeled data most data is unlabeled Stuck in local minima Over-fitting Too many hyper parameters
  • 33. Problems with Backpropagation Limitations Get stuck in local optima start weights from random positions Error attenuation, long fruitless training Slow convergence to optimum large training set needed Only use labeled data most data is unlabeled Backpropagation (BP) barely changes lower-layer parameters (vanishing gradient) Therefore, deep networks cannot be fully (effectively) trained with backpropagation 52
  • 34. Breakthrough with Backpropagation Breakthrough Recent – Long patient training with GPUs and special hardware Deep belief networks (unsupervised pre-training) Convolutional neural networks (reducing redundant parameters) Rectified linear unit (constant gradient propagation) 53
  • 35. Rectified Linear Units More efficient gradient propagation, derivative is 0 or constant, just fold into learning rate More efficient computation: Only comparison, addition and mul tiplication. Leaky ReLU f(x) = x if > 0 else ax where 0 ≤ a <= 1, so that derivat e is not 0 and can do some learning for that case. Lots of other variations Sparse activation: For example, in a randomly initialized network s, only about 50% of hidden units are activated (having a non-z ero output) CS 678 – Deep Learning 54
  • 36. Convolutional Neural Networks (CNN) A type of feed-forward neural network Inspired by biological process Weight sharing (convolution) + Subsampling (pooling) Reducing the number of parameters (Reduce over- fitting) Translation invariance Input 28 × 28 Feature maps 4@24 × 24 Feature maps 4@8 × 8 Feature maps 8@4 × 4 Feature maps 8@2 × 2 Feature maps 8 ⋅ 2 ⋅ 2 × 1 Output 10 × 1 Convolution layer Max-pooling layer Convolution layer Max-pooling layer Reshape Linear layer [LeCun, 1998]
  • 38. Convolutional Neural Networks (CNN) Convolution과 Pooling (Subsampling)을 반복하여 상위 Feature 를 구성 Convolution은 Local영역에서의 특정 Feature를 얻는 과정 Pooling은 Dimension을 줄이면서도, Translation-invariant한 Feature를 얻는 과정
  • 39. Convolutional Neural Networks (CNN) Neural network with sparse connections Learning algorithm: Backpropagation on convolution layers and fully-connected layers
  • 41. Visualization of FilterBank VGG16 architecture ImageNet most filters identical, rotated by some non-random factor (typically 90 degrees). – rotation-invariant. The rotation observation holds true in block4_conv1. Textures similar to that found in the objects in block5_conv2 61
  • 43. 26.2% 16.4% 13.5% 12.9% 11.8% 7.3% 6.7% 4.9% 4.8% 3.6% Breakthroughs in CNN 2012 2013 2014 2015 Alexnet (2012) • 1st place in 2012 • 5 conv layers + 3 fully connected layers • Dropout & ReLU ImageNet Large Scale Visual Recognition Challenge Results SIFT + FVs (2012) • 2nd place in 2012 • SIFT + fisher vectors • No CNNs ReLU (Rectified Linear Units) Alexnet architecture Data augmentation (flip, random crop) Dropout
  • 44. 26.2% 16.4% 13.5% 12.9% 11.8% 7.3% 6.7% 4.9% 4.8% 3.6% Recent Breakthroughs in CNN 2012 2013 2014 2015 ZF Net (2013) • 3rd place in 2013 • By Matthiew Zeiler & Rob Fergus • Variant of Alexnet Alexnet (2012) • 1st place in 2012 • 5 conv layers + 3 fully connected layers • Dropout & ReLU Clarifai (2013) • 1st place in 2013 • Deep Learning startup • Founded by Matthiew Zeiler • Variant of Alexnet VGG Networks (2014) • 2nd place in 2014 • By Oxford computer vision group • 19 layers deep GoogLeNet (2014) • 1st place in 2014 • 24 layers of convolution • Memory efficient network ImageNet Large Scale Visual Recognition Challenge Results SIFT + FVs (2012) • 2nd place in 2012 • SIFT + fisher vectors • No CNNs Overfeat (2013) • 2nd place in 2013 • By NYU • Variant of Alexnet GoogLeNet architecture
  • 45. 26.2% 16.4% 13.5% 12.9% 11.8% 7.3% 6.7% 4.9% 4.8% 3.6% Recent Breakthroughs in CNN 2012 2013 2014 2015 ZF Net (2013) • 3rd place in 2013 • By Matthiew Zeiler & Rob Fergus • Variant of Alexnet Alexnet (2012) • 1st place in 2012 • 5 conv layers + 3 fully connected layers • Dropout & ReLU Clarifai (2013) • 1st place in 2013 • Deep Learning startup • Founded by Matthiew Zeiler • Variant of Alexnet VGG Networks (2014) • 2nd place in 2014 • By Oxford computer vision group • 19 layers deep GoogLeNet (2014) • 1st place in 2014 • 24 layers of convolution • Memory efficient network Batch normalization (2015) • By Google • Simple but powerful normalization algorithm Parametric ReLU (2015) • By Facebook ImageNet Large Scale Visual Recognition Challenge Results SIFT + FVs (2012) • 2nd place in 2012 • SIFT + fisher vectors • No CNNs Overfeat (2013) • 2nd place in 2013 • By NYU • Variant of Alexnet Parametric ReLU Residual Learning (add skip connections) Deep Residual Network (2016) • Winner of ILSVRC 2015 • By MSRA • More than 100 layers deep with skip connections
  • 52. Chest X-ray 75 Data cleansing SW with AI Gold Standard (Generate 1000 data per class) Manual Drawing SW Nodule Interstitial Opacity Consolidation Pleural Effusion
  • 53. Anonymization 77 Data 수집 - 정상 1,821,455 - 비정상 287,626 DICOM header Meta inf. Chest (AP,Lateral) Study description filtering Anonymizer 1. Batch conversion 2. 익명화 1. DICOM patient info 0008 tag 2. PatientID, SEX, AGE, NAME, Birth date, etc 3. Research ID generation 3. Efficient Error Handling MIRL@AMC ananoymizer Anonymizer@MIRL*
  • 54. Cleansing Data Description - # 9,589 from 500,000 - Resize into 100x100 - Standardization CNN Training Data (3,000/3,000 ) Validation Data (1,000/1,000) Test Data (1000/589) Normal Class Abnormal Class Model Precision Sensitivity Specificity CNN 99.3% 99.8% 98.4% CNNBN 100% 99.9% 100% H. C. Shin,“Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning, “IEEE Transactions on Medical Imaging, 2016, pp. 1285-1298.
  • 55. Chest X-ray CAM Weakly Supervised Learning + Class Activation Map - Resnet50 ILSVRC Pre-trained Model
  • 57. Interesting Case 85 Normal Cardiomegaly Ground Truth Prediction 수술자국
  • 58. Chest PA YOLO • Fine-Tuned You Only Look Once(YOLO) model • YOLO model-26 layers including 24 convolution layers and 2 fully connected layers • Only final layer training • Only 6 classes of abnormal lesion, our last layer requires C = 6
  • 59. Predicted Con. True Con. 1 0 1 164 56 0 50 941 1. nodule 2. consolidation 3. interstitial opacity 4. cardiomegaly 5. pleural effusion 6. pneumothorax Predicted Con. True Con. 1 0 1 123 41 0 32 982 Predicted Con. True Con. 1 0 1 82 14 0 25 1023 Predicted Con. True Con. 1 0 1 414 9 0 80 691 Predicted Con. True Con. 1 0 1 267 32 0 43 838 Predicted Con. True Con. 1 0 1 55 19 0 22 1050 Results
  • 60. • Nodule result ROI 1. Two nodules are detected in in Chest X-ray image 2. Nodule are only detected in Chest X-ray image • Consolidation result ROI 1. Consolidation and pleural effusion are simultaneously detected in Chest X-ray image 2. Consolidation is only detected in Chest X-ray image ① ② ① ② Examples
  • 61. • Interstitial Opacity and Cardiomegaly result ROI 1. Two interstitial opacity are detected in Chest X-ray image. 2. Cardiomegaly is detected in Chest X-ray image . 3. Cardiomegaly and two pleural effusion are detected in Chest X-ray image. ②① ③ Examples
  • 62. • Pleural Effusion result ROI 1. Cardiomegaly and Pleural Effusion are detected in Chest X-ray image • Pneumothorax result ROI 1. Pneumothorax are detected in Chest X-ray image 2. Pneumothorax and pleural effusion are simultaneously detected in Chest X-ray image ① ②① Examples
  • 63. AI in Radiology CAD for COPD Compare various classifiers for OLD classification Lee Y., et al, CMPB, 2009. 93(2): p. 206-15. Adding shape features for accuracy enhancement of emphysema quantification., J Digit Imaging, 22:136-148, 2009, SPIE 2007 Honorable Mention Poster Award / JDI Best paper 2009 Compare texture-based quantification vs density-based quantification for emphysema quantification, Investigative Radiology 43:395-402 , 2008 CAD for DILD Study feasibility for DILD, Korean J Radiol 10:455-463, 2009 Context sensitive SVM for whole lung quantification, IFMIA 2011, IWPFI 2011, JDI 2011 CrossVendor study for ILD (GE vs Siemens), Kim N , et al, RSNA 2010, Med Phys 2013 Texture based segmentation for iodine quantification in DECT of ILD, RSNA 2013, ER 2015 3D extension of texture analysis in DILD, RSNA 2014 CADD for DILD with regional parenchyma classification, WIP in Med Phys Deep learning for DILD, WIP in JDI DILD lung segmentation using deep learning, WIP in Med Phys Noise reduction of HRCT using auto-encoder, WIP in Med Phys Ensemble method for ILD classifier, JDI accepted CAD for AVUS (Ultrasounds), Thyroid Nodule (Med Phys), Lymph node meta (Acta Radiol), Pulmonary Embolism (Eur Radiol), …
  • 64. Semantic Segmentation on DILD HRCT 92
  • 65. Image Pattern Classification on DILD HRCT (a) Consolidation (b) Emphysema (c) Normal (d) Ground-glass opacity (e) Honeycombing (f) Reticular opacity DEEP Learning Training dataset Testing set Accuracy (%) Ensemble Classifiers GE GE 96.06 ± 1.61 Siemens Siemens 96.11 ± 1.19 GE+Siemens GE+Siemens 85.12 ± 1.91 +5~10% improvement (breakthrough in 5 years!) O = GE; X = SiemensSVM CNN
  • 66. 3D DILD 2 Class : Normal Raw Data ROI Extraction 3D 3 Channel 2D 2D
  • 67. Results 2D (82.7%) • 3 X 2D (81.7%) • 3D (84.7%) • 비교를 위해 augmentation 적용 안됨. 따라서 3D의 경우 variation이 큼. (많은 parameter)
  • 68. CNN Airway Segmentation 80 COPD Patients’ Inspiration CT 69 CT volumes are included in training 11 CT volumes are NOT included in training GS : Manual segmentation 107 Axial 3 slices, Sagittal 3 slices, Coronal 3 slices 32 x 32 x 3 x 3 Weights are shared 2-class classification CT Volume Probabili ty Volume CNN For each voxels inside lungs Segment ed Airway Hard thresholding (0.51) and Select the connected component
  • 70. 내시경 영상 판별 (대장) NICEI실제 클래스 예측 클래스 클래스별 확률 Patch Image 결과 Whole Image 결과
  • 71. 결과 검토 (-) (+) (0) *0의 위치는 다를 수 있음. 정답 NICEIII NICEI CAM NICEII CAM NICEIII CAM 예측 NICEIII 이상 영역 marking
  • 73. 113 The Previous Research for Quantification Definition of Similar Lung Images Extraction of Distribution Features Extraction of Distribution Features * https://en.wikipedia.org/wiki/Large_margin_nearest_neighbor * Y.J.Chang, et al,. “A support vector machine classifier reduces interscanner variation in the HRCT classification of regional disease pattern in diffuse lung disease: Comparison to a Baysian classifier”, Medical Physics 40 (5), 051912 (2013)
  • 74. 114 Evaluation: Statistics ( * p-value < 0.01, ** p-value < 0.001 ) Evaluation: Recall Accuracy Fitting Test Data Acknowledgement This work was supported by the Industrial Strategic technology development program (10072064, Development of Novel Artificial Intelligence Technologies To Assist Imaging Diagnosis of Pulmonary, Hepatic, and Cardiac Diseases and Their Integration into Commercial Clinical PACS Platforms) funded by the Ministry of Trade Industry and Energy (MI, Korea)
  • 76. 116 AP 통신 : 인공지능(로봇)이 기사 작성, 기술 공개 초당 2,000개 기사 작성 가능 기존의 300개 기업 실적 -> 3000개 기업 실적 커버
  • 77. 117 Right now, about 80% of Americans who need a lawyer can't afford one "With ROSS, lawyers can scale their abilities and start to service this very large untapped market of Americans in need,"
  • 80. 정밀의료(Precision Medicine) Initiative “And that’s why we’re here today. Because something called precision medicine … gives us one of the greatest opportunities for new medical breakthroughs that we have ever seen.” President Barack Obama January 30, 2015 정밀의료 치료기기 핵심 : 빅데이터 + 인공지능 빅데이터 ▶ 인공지능 ▶ 환자 맞춤형 진단/치료/공공 120/27
  • 81. 의료 빅데이터 + 인공지능 의료 빅데이터를 이용하는 정밀의료 실현 사물인터넷 유전자검사 의료영상 환자 모니터링 하루 37만건의 의무기록 (연 간 약 1TB) 연간 약 2백만 영상데이터 (30TB) 3.8×109 염기쌍을 가진 DNA 정보 대형 종합 병원
  • 82. 의료비 절감의 필요성 의료효율성 국내현황 Efficacy (효용성) Cost (비용) Equation = Efficacy / Cost • 정밀치료를 통한 효용성 극대화 • 효율적 의료시스템 구축을 통한 비 용 감소 • 고령화 사회 등으로 GDP대비 의료비 비율 빠르게 증가하고 있으며 상승폭 또한 커지고 있 음 • 국민 소득 및 소비수준을 고려하여 의료비는 변화에 반비례하여 감소되어야 할 필요성 존재
  • 83. 인공지능 관련 국내외 시장 현황 사회 및 산업 각 분야시장의 핵으 로 부상중인 인공지능 2020년까지 연평균 53.65% 성장 률을 기록할 것으로 예상1) 국내의 인공지능 시장 규모 예상 2020년 2조2천억원 2025년 11조원 2030년 27.5조원2) 1) 2016 market and market 2) KT경제경영연구소 123
  • 84. Global Trends Drive Momentum in health care industry Data Explosion 150+ exabytes Amount of healthcare data today1 Over 230K Active clinical trials2 80% Healthcare data that comes from unstructured data sources3 Dynamic Delivery Environment 50% Expected alternative payments form the Centers of Medicare and Medicaid by 2018 4 75%+ Percentage of patients expected to use digital health services in the future5 90K Expected shortage of physicians by 2020 6 Value vs Volume 4.7 trillion Estimated global economic impact of chronic disease by 2030 7 3 trillion Estimated US healthcare spending 8 100’s Approx. amount of decisions a person living with Type 1 Diabetes makes a day9 Efficient and effective R&D 1 in 10 Clinical trials in cancer that are shut down due to lack of participation10 2.6B Average costs to develop a new pharma drug11 <10% Amount of drug currently in development that make it market12 124 1: NCBI: big data analytics in healthcare: promise and potential 2: ClinicalTrials.gov, 3: NIH 4 CMS 5: McKinsey Healthcare's Digital Future July 2014, 6: AAMC Report The complexities from 2014 ro 2025 7: WEF global economic burden non-communication diseases 8: Health affaires. Team analysis 9: OpenAps.org 10: Bio-clinical development success rates 20, Health Economic volume 47, May 2016 1. Life expectancy data, WHO, 2012 2. 2015 Global life sciences outlook: adapting in an era of transformation, Deloitte DTTL, 2014 3. Informa Pic Market Line Extracted Oct 2014
  • 85. Opportunity 126 8 trillion : Industry Size 2 trillion : waste in industry Better experience Imaging : Unnecessary tests Lower cost Oncology: Variability of Care Better outcomes Life sciences: Failed clinical trials Government: Fraud, Waste and Abuse Value Based Care: Cost of chronic disease 360 billion : total IT and healthcare market opportunity *IBM Watson
  • 86. 국제적 동향 인공지능 의료적용 현황 (해외) 1) marketandmarkets 2016.02 세계 시장 규모는 5.05조원 (2020년 예측치), 향후 5년 성장률 53.65%으로 新성장 사업분야1) 의료 문서, 영상 빅 데이터를 활용한 진단 및 처방이 가능한 기술을 개발 헬스케어 분야부터 정밀의료 산 업까지 폭넓은 기술 로드맵 보유 폐암진단을 위한 딥러닝 기반 분석 시스템 131
  • 87. 133
  • 88. AI in healthcare 134 2011년부터 해외 인공지능 기반 의료 스타트 업 기업들은 8억7천만 달러(1조 원) 규모의 투자를 유치
  • 89. 인공지능 의료적용 현황 (국내) 국내 동향 • 삼성과 LG를 비롯한 IT 기업의 높은 인공지능 분야 기술력 • 세계적 수준의 임상의료 및 임상시험 기술 • 최근 뷰노코리아, 루닛 등 인공지능기반 의료분야 스타트업들이 조기성과를 보이는 등 의료분야 인공지능 산업 기반 확대 미만성폐질환에서 CT로부터 질병을 판별. 질병 판독. 질병의 진행상황, 치료법 등 전문의 판단 을 도움을 주는 시스템 개발 소아 골연령 판정 보조 시스템 개발 흉부단순촬영 등 영상 데이터를 딥러닝 기술로 학습, 결핵, 유방암등 검출하는 기술을 개발 시스템생물학에 인공지능을 결합시 켜 기존 약물 개발 과정을 개선 135
  • 90. 인공지능 의료적용 분야 인공지능 분야 시각지능 언어지능 판단지능 자동분류 요약/창작 공간지능 임상시험 케이스선정 신약개발프로세스 진료보조 비서서비스 음성인식 의무기록 데이터기반 정밀의료 유전체분석 약혼합사용 및 합병증 예측 진단검사추천 판독보조 정상유무판정 유사증례검색 판독문 생성 병리분야 판독 보조 물류, 수술실, 병실 운영 로봇수술 의료 인공지능 인공지능 의료적용 분야 136
  • 91. 인공지능 의료적용 분야 (임상 중심 분류) 정상 유무 판정 초기진단 빅 데이터와 차세대 인공지능 기법을 통해 정밀판독 이 없이도 초기 감별진단 유사증례검색 DB의 수많은 증례로부터 유사 한 증례를 검색, 시각화하여 진 단에 도움 예비 판독문 생성 인공지능 기반 의료영상 분석기 법과 자연어 처리기술을 융합, 영상전문의의 판독을 보조할 수 있는 수준의 판독문 자동생성 병리분야 판독 보조 병리영상 빅데이터에 인공지능 기법의 적용, 진단, 발병기전 분 석, 예후 예측 등에 활용 임상시험 인공지능 기술을 기반으로 단일 병원 및 병원 클러스터간 물류, 수 술실, 병실의 효율적 운영 물류, 수술실, 병실 운영 로봇수술 인공지능 기술을 통해 의료 로봇 의 수술을 계획, 위험도 예측, 침 습부위 최소화 진료보조 데이터기반 정밀의료 판독보조 신약 약물 개발과정에 인공지능 기법을 적용, 질병 치 료에 더욱 효과적인 약물의 조합과 용도 변경 탐 색, 약물 후보군 및 임상환자 군의 최적화 케이스 선정 인공지능 기반 검색기법을 통해 적합한 질병 및 환자 를 탐색, 임상 시험의 준비 기간을 단축, 객관성 향상 비서서비스 IoT기술, 음성 인식기술 및 인공지능 기술을 융합, 효율적인 예약, 진단 및 진료 프로세스, 업무 정보 업데이트 및 맞춤형 큐레이션 음성인식 의무기록 진단 및 판독 내용의 기록을 자동화, 의료분야 전문용어를 판독 및 구조 화할 수 있는 수준의 음성 인식 및 문서 생성 기술 진단 검사 추천 정밀 진단 및 치료를 위해 인공지능 및 빅데이터를 활용, 정확도를 높이 고 위험도를 낮추기 위한 추가 진단 검사 프로세스를 추천 유전체 맞춤 의료를 위해 유전체, 멀티모달 의료 영 상 및 임상병리 빅데이터를 바탕으로 연관 성을 분석, 모델링하여 예후 예측, 진단 및 치료 약 혼합 사용 및 합병증 예측 치료 및 약물 사용 시 사례기반 위험성 또는 합병 증의 위험을 알려 주어 의사의 최종 결정을 보조 137
  • 92. 인공지능 의료적용의 어려움과 대책 139 의료환경의 표준화의 어려움으로 인공지능 훈련을 위한 라벨링 어려움 개인정보보호에 따른 데이터의 비개방성 의료기기 인허가, 신의료기술인정 및 수가 장벽 인공지능 전문가와 의료전문가의 단절 의료빅데이터의 표준화와 연계 연구 필요 인공지능기술개발을 위한 공개 데이터 개발과 이를 이용한 그랜드 챌린지 방식의 기술 진흥 필요 연구기획 초기부터 관련 종합 대책 정책적 지원 방안 마련 인공지능 교육, 연구개발 산업화를 망라하는 생태계 구축 필요 어려움 대책
  • 93. Strategy Proof of principal Google Retinopathy, JAMA 2016 – 130k images, 54 ophthalmologists – Reference standard – Severity, image quality, L/R, FoV Ophthalmology, Pathology, Radiology, Dermatology Cloud/Vision/Speech API AI make AI Amazon, MS Naver J Proj. Clova, 오감 AI Line chat bot since 2014 141
  • 94. Technical Issues on Medical Imaging Data collection More (clean) data Accurate annotation Legal issues Models selection Deeper network Off the shelf models Result Interpretations Neural network visualization Human-friendly interpretation 142
  • 95. Machine Operable, Human Readable Visual attention Category – feature mapping Sparsity and diversity 150
  • 96. Machine Operable, Human Readable Evidence hotpot for lesion visualization “SpineNet: Automatically Pinpointing Classification Evidence in Spinal MRIs” 151
  • 97. Machine Operable, Human Readable Visualization of salient region in bone x-ray 152
  • 98. MD-friendly Interpretation Breast cancer risk prediction through BI-RADS categorization of mammography Analysis and visualization for breast density prediction Mapping and visualization of patient by predicted breast cancer risk score 153
  • 99. MD-friendly Interpretation 154 Learning to read chest x-ray Automated annotation with attention and visual- sentence mapping
  • 100. MD-friendly Interpretation 155 Contents-based case retrieval Suggest similar cases with the clinically matching context
  • 102. 167
  • 103. Collaborators • MIRL • Clinical Collaborators@Asan Medical Center, SNU BH – Radiology : Chest, Cardiac, Abdomen, Neuro/Brain – Neurology :Dongwha Kang, Chongsik Lee, Jaehong Lee, Sangbeom Jun, Misun Kwon, Beomjun Kim – Cardiology ; Jaekwan Song, Jongmin Song, Younghak Kim – Internal Medicine : Jeongsik Byeon – Pathology : Hyunhee Go – Surgery : Bumsuk Go, JongHun Jeong, Songchuk Kim • Academy • KAIST EE, CS, Math • Related Companies • LG, VUNO, Coreline Soft, MIDAS IT, KakaoBrain