Title: Rapid Diagnosis of Acute Heart Disease by Cloud-based High Performance Computing for Computer Vision
1Rapid Diagnosis of Acute Heart Disease by
Cloud-based High Performance Computing for
Computer Vision
- Oleksii Morozov
- Physics in Medicine Research Group
- University Hospital of Basel
- Switzerland
April 8, 2010
2The Heart
- Life-sustaining pump 2500000 L/year of vital
blood - Coronary artery disease (CAD) is most frequent
cause of heart malfunction and death - World largest killer (WHO)
- 29 of global death
- 17100000 lives/year
3Cardiology yesterday
- Tools with relatively low information content
4Cardiology today
- More tools, more information
- Subjective decision mostly relying on experience
of a doctor
5Cardiology tomorrow
- More advanced technologies
- Multidimensional information
- High quality, high resolution data
- Multimodal information
- Quantitative, objective, integrative computer
based analysis - Worldwide-networked standards and databases
- Problems
- Need for high performance computing in a
distributed environment but only for a fraction
of the time - Global storage network for storing large datasets
6Cardiac Ultrasound
- One of the modern tools for evaluation of the
heart function - HF sound waves No Radiation/Ionization
- Safe, Non-invasive, Fast, Portable, Cheap
- - Rather low signal to noise ratio
7Cardiac UltrasoundDiagnostic value
8Cardiac UltrasoundDiagnostic value
9Cardiac UltrasoundDiagnostic value
103D Cardiac Ultrasound
- Explore heart in 3D
- Freehand ultrasound (Manual sweeping)
113D Cardiac Ultrasound
- Explore heart in 3D
- Freehand ultrasound (Manual sweeping)
- Mechanical sweeping ultrasound (Motor driven)
123D Cardiac Ultrasound
- Explore heart in 3D
- Freehand ultrasound (Manual sweeping)
- Mechanical sweeping ultrasound (Motor driven)
- Live 3D ultrasound (2D arrays with electrical
sweeping)
13Cardiac Ultrasound
- Ultrasound machine a transducer a
supercomputer
50000 500000 USD
Idle 90 of the time
14Computational problems in Cardiac Ultrasound
Non-uniformly sampled measurements
Complete gridded or continuous data
representation
153Dtime signal reconstruction
- Inherent non-uniformity of scanning
- Spatial non-uniformity
- Serialism in scanning
163Dtime signal reconstruction
- Inherent non-uniformity of scanning
- Spatial non-uniformity
- Serialism in scanning
- Non-uniformity in synchronization (ECG)
173Dtime signal reconstruction
- Inherent non-uniformity of scanning
- Spatial non-uniformity
- Serialism in scanning
- Non-uniformity in synchronization (ECG)
- Body motion artifacts (breathing)
I(x,y,z,t) ?
4D non-uniform data
183Dtime signal reconstructionA spline solution
- B-spline non-uniform interpolation by
Arigovindan, Unser (EPFL, Switzerland 2005) - Robust global interpolation handles oversampling
and undersampling (gaps) in the data - Sparse and well-conditioned alternative to the
optimal RBF solution - Enjoys multiresolution properties (way to fast
solving) - Parallelizability of solving process
- Successfully applied to 2D problems
193Dtime signal reconstructionA spline solution
- Obstacles in 3D/4D
- Complexity is exponentially dependent on the data
size 128 x 128 x 128 x 18 gt 78752009856
non-zeros (312 Gbyte in single precision) - Tensor based approach by Morozov, Hunziker, Unser
2009 - Tensor decomposition of the problem
- Relaxed storage requirements
- Feasibility on standard workstations
- 9 millions of measurements with size 128 x
128 x 128 x 18 -gt 30 minutes on my dual core
laptop
203Dtime signal reconstructionA spline solution
- Tensor based approach applied to ultrasound data
from continuously rotating transducer
21Computational problems in Cardiac Ultrasound
- Tissue/blood motion estimation
- Doppler Ultrasound imaging (State of the art)
- Semi-quantitative measurements
- Full motion reconstruction
- Generalization of B-spline reconstruction to
vector valued data (Arigovindan, Unser 2005) - Employing additional constraints from physics of
fluids (incompressibility, Navier-Stokes
equations)
22Computational problems in Cardiac Ultrasound
- B-spline based tissue motion reconstruction
- Continuous
- Fully quantifiable
- Can be combined with Doppler for better
robustness
23Computational problems in Cardiac Ultrasound
- Blood flow reconstruction
- Resolves ambiguity of Doppler measurements
- Continuous
- Fully quantifiable
24Pathway to distributed supercomputing
- Multicore (IBM Power7) claimed 260 GFLOP/chip
- Cluster (UniBasel) 34500 GFLOP/400 cores
- GPGPU (ATI 4870X2) 2000 GFLOP/card
- GPGPU array
- FPGA accelerator cards dozens of GFLOP/chip, up
to 512 chips per system, low power - In exploration within ICES Microsoft project
- Cloud - Microsoft Azure
25Cloud Ultrasound Processing Service
- Reasons
- Processing of large multidimensional multimodal
medical data requires vast computational power - Building/maintaining own HPC infrastructure is
overly expensive - Relatively rare use of HPC power (few times per
day) - Availability at multiple points of care (medical
practices and hospital emergency rooms) - Unified storage/access of the multimodal medical
data
26Cloud Ultrasound Processing Service
Record data
User
Cloud
4D acquisition with real-time on board
visualization
Visualization/Analysis parameters
Interactive web-based visualization of the result
Rendered images and quantitative information
27Cloud Ultrasound Processing Service
- Record data
- Raw data
- 180 beams x 500 samples x 100 frames x 10 sec -gt
85 Mb - Additional information (geometry) -gt few Kb
- Lossless compressed DICOM
- Low latency response to the user by sending first
a subpart of the data for coarser resolution
reconstruction
28Cloud Ultrasound Processing Service Signal
reconstruction
- Problem is very large for solving using direct
solvers -gt use iterative solver
Ci1 Ci OP(Ci) OP linear operator
2 iterations
50 iterations
80 iterations
29Cloud Ultrasound Processing Service Signal
reconstruction
- Iteration can be distributed relative to the grid
dx
dx, dy grid spacing
C1,1i1 C1,1i OP1,1(Ci) C1,2i1
C1,2i OP1,2(Ci) C2,1i1 C2,1i
OP2,1(Ci) C2,2i1 C2,2i OP2,2(Ci)
dy
Completely independent output
Ck,m solution subpart dedicated to a compute
unit
OPk,m() operator applied by a k,ms compute
unit
30Cloud Ultrasound Processing Service Signal
reconstruction
OPk,m() uses data outside the bounds of Ck,m
- Extents of dependent input data 3 samples for
cubic spline - At each iteration this data is transferred among
adjacent units - Performance limiting factor
Ck,m
31Cloud Ultrasound Processing Service Signal
reconstruction
- Data dependency
- Data size 512 x 512 x 512 x 64
- Single precision 32 GB
- Infiniband QDR 12X ( 12GB/s )
Number of units Size of dependent data per unit, MB Total data transfers for single iteration, MB Maximal number of iterations/s (excluding CPU time)
64 72 4608 166
128 48 6144 250
256 30 7680 400
512 18 9216 666
1024 12 12288 1000
32Cloud Ultrasound Processing Service Signal
reconstruction
- Computational load
- Data size 512 x 512 x 512 x 64
- Intel Quad Core 2.67 GHz (30 GFLOP/s in single
precision) - PC3-10600 DDR3-SDRAM (30 GB/s)
- 115000000 data samples
- Total requirements 5000 GFLOP, 3000 GB
of memory transfers
Number of units Maximal number of iterations/s (including inter-unit communication)
64 0.24
128 0.48
256 0.96
512 1.92
1024 3.84
33Cloud Ultrasound Processing Service Signal
reconstruction
- Multiresolution
- Coarse to scale propagation getting general
from coarser scales and improving details on
finer scales - Inherent spline inter-scale relation
34Cloud Ultrasound Processing Service Signal
reconstruction
- Multiresolution in solving algorithm
- Coarse to scale propagation getting general
from coarser scales and improving details on
finer scales - Inherent spline inter-scale relation
- Multigrid solving algorithm
35Cloud Ultrasound Processing Service Signal
reconstruction
- Multiresolution in solving algorithms
- Coarse to scale propagation getting general
from coarser scales and improving details on
finer scales - Inherent spline inter-scale relation
- Multigrid solving algorithm
- Few iterations needed at each scale to get
reasonably good solution - With each coarser scale the cost of iteration
decreases exponentially - In total requires much less computational load
than pure iteration
36Cloud Ultrasound Processing Service Signal
reconstruction
- Total requirements per compute unit
- Data size 512 x 512 x 512 x 64
- 6 scales including finest scale
- 16 iterations at each scale
- Motion reconstruction algorithms require
9 times more computational load
Number of units Memory, GB CPU time, s
64 2.7 62
128 1.36 31.2
256 0.68 15.6
512 0.34 7.8
1024 0.17 3.9
37Cloud Ultrasound Processing Service
Data size 512x512x512x64
Storage, GB 32
Number of instances 1024
Compute hours 0.0011
hourinstance 1.13
- Switzerland (1/1000 of world population)
- 700 cardiologists/7000000 population
- 1000 echocardiograms per year per cardiologist
- Multiple views(3) per patient
- Multiple analyses (3) per view
7000000 use cases/year 3043 use
cases/hour Full loaded 3 x 1024
instances (Uniform load in Switzerland)
38Collaborative Research enabled by the Cloud
- Globally available service for processing,
storing and accessing medical data - Standardized DICOM interface for unifying data
access - Involve all interested parties around the world
- World-wide large scale trials are possible
- Getting more statistics for rare cases
- Building reference datasets for known cases
39Conclusion
- An approach to rapid diagnosis of heart disease
using cloud based distributed computing - Replace ultrasound machines supercomputer by a
cloud service for remote processing and storage - Miniaturization of the medical equipment and
decrease of its costs - Availability of advanced analysis technologies
for objective analysis - Availability at multiple points of care
- Unified storage and access of medical data
- Enables collaborative research