toward the future of ai-driven medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... ·...
TRANSCRIPT
![Page 1: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/1.jpg)
Toward the Future of
AI-Driven Medicine
葉肇元 醫師
雲象科技執行長
![Page 2: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/2.jpg)
雲象科技
Bring state-of-the-art technology to healthcare.
Our Core
Our Mission
Our Goal
We’re a Medical Image AI company.
Empower medical imaging with A.I.
![Page 3: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/3.jpg)
A survey of deep learning in medical image analysis
Mammographic Mass Classification
Diabetic RetinopathyDetection
Breast CancerMetastasis Detection
Brain Lesion Segmentation
Airway Segmentation of Chest CT Image
Lung Nodule Detection
Bone Suppression in X-Ray Image
Skin DiseaseClassification
Prostate
Segmentation
![Page 4: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/4.jpg)
Organs of interest for Medical Image AI
• Brain
• Brain tumor segmentation
• Disease classification
• Survival Prediction
• Eyes (Retina)
• DM retinopathy, cardiac risk factor
• Lungs
• Lung nodule detection
• Breast
• Breast cancer screening
• Heart
• Cardiac image analysis
• Intestine
• Polyp classification
• Prostate
• Prostate segmentation
• Bones
• Age determination
• Skins
• Disease classification
• Blood Vessels
• Blood vessel segmentation
• Blood
• Blood cell counting and classification
![Page 5: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/5.jpg)
• Authored by Google, Verily Life Sciences, and Stanford School of Medicine
• Inception-V3 Model trained on data from 236,234 patients from EyePACS , 48,101(UK Biobank), validated on data from 12,026 patients from UK Biobank, and 999 patients from EyePACS.
• Used Retinal Fundus Image to predict
• Age, gender, smoking status, BMI, systolic blood pressure, diastolic blood pressure
Poplin R, et al. Nature Biomedical Engineeringvolume 2, pages158–164(2018)
![Page 6: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/6.jpg)
MAE : Mean Absolute ErrorFor continuous risk factors (like age), the baseline value is the Mean Absolute Error of Predicting the mean value for all patients.
![Page 7: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/7.jpg)
![Page 8: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/8.jpg)
The cost of making medical image AI not often talked about :
Time
![Page 9: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/9.jpg)
Expected Timeline for a Medical Image AI Project
Required Skill Category:• Interdisciplinary Knowledge• Hospital Information System
Time(Month)
Identify Topic
Collect & Process
Data
Train & Validate ModelCollect More Data
Train & Validate Model
Deploy
2 4 6 8
• AI Software and Hardware• Healthcare Workflow
![Page 10: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/10.jpg)
In reality..
Time(Month)
Identify Topic Collect, Process and Label Data Train & Validate Model
2 4 6 8
![Page 11: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/11.jpg)
Houston, we’ve got a problem.
• So it takes ten months to make one AI model happen (if you’re lucky).
• But there are thousands of clinical tasks that could potentially benefit from the help of A.I. !
• (How on earth can we replace Drs. with A.I. ?)
![Page 12: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/12.jpg)
How Do We Get There ?
Time(Month)
2 4 6 81
Identify TopicCollect Data
Train and Validate Model
Deploy
![Page 13: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/13.jpg)
What’s holding us back? Infrastructure.
• Hospital Information System
• AI Software and Hardware
• Interdisciplinary Knowledge
• Healthcare Workflow
![Page 14: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/14.jpg)
Interdisciplinary Knowledge
![Page 15: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/15.jpg)
Essential Ingredient of a Successful Medical Image AI Project
• Interdisciplinary knowledge
• Intricacies of medical diagnostic procedures
• Capabilities of different neural network models
• How medical data can be digested by neural networks and turned into insight
![Page 16: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/16.jpg)
Our first attempt at Digital Pathology AI• Lymphoma screening using whole slide image
![Page 17: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/17.jpg)
Digging into data : examining raw input
Dark Zone
Light Zone
Follicular Lymphoma
Mantle Zone
Tinged-Body Macrophage
![Page 18: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/18.jpg)
Web interface for deep learning inferencing
![Page 19: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/19.jpg)
Training statistics
![Page 20: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/20.jpg)
Lymphoma Screening Model Used on Whole Slide Image
![Page 21: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/21.jpg)
Improved Tools for Whole Slide Image Labeling
![Page 22: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/22.jpg)
Dataset Statistics
• Labeled Training Slides : 56 Cancer, 56 Benign
• Total number of extracted patches
• Validation: ~40,000 patches
• Testing : ~40,000 patches
Benign Cancer Background
4,460,452 147,533 87,974
![Page 23: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/23.jpg)
Neural Network Architecture
• Modified ResNet-50:
• Dense layer after Global Average Pooling for tissue / background binary classification
• Separate path with additional dense layers for cell type (cancer / benign) classification
![Page 24: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/24.jpg)
Neural Network Training
• Heavy data augmentation
• Flipping (Up-down, left-right), Add, Multiply, Add to Hue and Saturation, Contrast Normalization, Gaussian Blur, Gaussian Noise
• Class balancing : random sampling of equal number from each class
• Optimizer : Adam Optimizer
• Early Stopping
![Page 25: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/25.jpg)
Training Result
Foreground / background classification
Benign / Cancer classification
Loss
Accuracy
![Page 26: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/26.jpg)
Statistics Of Validation Result
![Page 27: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/27.jpg)
Testing Result
Recall = SensitivityPrecision = Positive Predictive Rate
![Page 28: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/28.jpg)
Prediction on Separate Test SlidePrediction by Neural NetworkGround Truth
Yellow : Cancer, Blue : Benign Red : Predicted cancer region
Accuracy : 90.4 %, Precision : 93.4% , Recall : 93.0 %
![Page 29: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/29.jpg)
Class Activation Map
![Page 30: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/30.jpg)
AI Software and Hardware
![Page 31: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/31.jpg)
• 1 Digital slide is larger than the entire CIFAR-10 dataset
• Digital slide : 80000*60000
• CIFAR-10 : 32*32*60000
![Page 32: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/32.jpg)
Medical Images AI Needs a Lot of Memory
• Medical images have very high spatial resolution:
• Radiography image : 5000*4000 uint16
• CT image : 512*512*300 uint16
• Digital Slide : 60000*60000*3 uint8
• Average ImageNet image : 469*387*3 uint8
![Page 33: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/33.jpg)
GPU memory alone is not sufficientfor Medical Image AI
• For VGG-16, during training
• A GTX-1080Ti can take an image up to 1200*1200
• A Tesla P40 can take an image up to 1700*1700
• A Tesla V100 can take an image up to 2100*2100
• CUDA unified memory
![Page 34: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/34.jpg)
CUDA Unified Memory in Tensorflow
![Page 35: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/35.jpg)
Specialized Hardware for AI Compute
A BREAKTHROUGH IN TRAINING AND INFERENCEEach of Tesla V100's 640 Tensor Cores operates on a 4x4 matrix, and their associated data paths are custom-designed to dramatically increase floating-point compute throughput with high-energy efficiency.
This key capability enables Volta to deliver 3X performance speedups in training and inference over the previous generation.
![Page 36: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/36.jpg)
The Power of Tensor Cores
0
2
4
6
8
10
12
14
16
GTX 1080 TI TITAN XP TITAN V
Ba
tch
es
pe
r se
con
d
Float 16 Batchsize 512
Development environment:
GTX 1080 Ti : Tensorflow 1.4, CUDA 8, cuDNN 5, nvidia-381 driver
Titan Xp : Tensorflow 1.4, CUDA 9, cuDNN 7, nvidia-387 driver
Titan V : Tensorflow 1.4, CUDA 9, cuDNN 7, nvidia-387 driver
Neural Network : Convolution * 6 + fully connected * 2 , trained on cifar-10* 2
0
2
4
6
8
10
12
14
16
GTX 1080 TI TITAN XP TITAN V
Ba
tch
es
pe
r se
con
dFloat32 Batchsize 512
![Page 37: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/37.jpg)
GPU is often thirsty : The Importance of Pipelining
9.7
15.2
8.1
4.2
2.33 2.28
0
2
4
6
8
10
12
14
16
1 CPU 2 CPU 4 CPU 8CPU 16 CPU
Training time per epoch
Without Queue
With Queue
Without Queue
With Queue
![Page 38: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/38.jpg)
Healthcare Information System
![Page 39: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/39.jpg)
Problems with Existing Hospital Information System
• Databases are not tightly connected
• Limited search functions
• The majority of data exists in unstructured format (.txt, .pdf, etc)
![Page 40: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/40.jpg)
Unified Web Interface for Medical Image AI• Web-based system that integrates:
• Clinical data
• Digital slides
• DICOM images / videos
• Deep learning annotation, training and inferencing
![Page 41: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/41.jpg)
Annotation Interface with Structured Reporting
![Page 42: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/42.jpg)
Annotation and Image Markup (AIM)
• An NCI initiated project that provides a solution to the following imaging challenges:
• No agreed upon syntax for annotation and markup
• No agreed upon semantics to describe annotations
• No standard format (for example, DICOM, XML, HL7) for annotations and markup
• The link between the semantics and image annotation will help make more useful and more interpretable medical image AI.
https://wiki.nci.nih.gov/display/AIM/Annotation+and+Image+Markup+-+AIM
![Page 43: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/43.jpg)
AIM Example
![Page 44: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/44.jpg)
Medical Record De-Identification
• Due to privacy concerns, AI research requires that personal identification information be removed from medical record.
• It’s hard to achieve satisfactory result using regular expression or other rule-based methods.
• Using tools like NeuroNER (name entity recognition), we’ve successfully achieved an F1 score of >97% on public dataset.
![Page 45: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/45.jpg)
Next Generation IT Infrastructure for AI-Powered Hospital
Clinical Terminal• Structured Report for Both Clinical and AI use
Hybrid Storage• Fast : Cache for AI training• Slow : Data Archive
AI Training Server• High Compute Capacity• Job Queues for Non-Stop Learning
Main Server• High Availability• Advanced Database System• Job Flow Control
AI Inferencing Server•Virtualization for On-Demand AI Inferencing • Optimized for Inferencing Speed
Clinical Data Clinical Data
AI Model
AI Model
AI-Powered Diagnostic Aid
![Page 46: Toward the Future of AI-Driven Medicineon-demand.gputechconf.com/gtc-taiwan/2018/pdf/3-3... · •Authored by Google, Verily Life Sciences, and Stanford School of Medicine •Inception-V3](https://reader035.vdocuments.net/reader035/viewer/2022070713/5ed0e1b3694f1a065759bd48/html5/thumbnails/46.jpg)
Acknowledgement
• 長庚醫院病理科莊文郁副教授
• 長庚醫院巨量資料及統計中心張尚宏主任
• 臺大醫院心臟內科王宗道教授
• 臺大醫院影像醫學科李文正醫師
• 雲象科技張哲惟
• 雲象科技游為翔
• 雲象科技楊証琨
• 雲象科技蔡岳霖