SlideShare a Scribd company logo
1 of 12
DiffuseMorph: Unsupervised Deformable Image
Registration Using Diffusion Model
Boah Kim, Inhwa Han, and Jong Chul Ye
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Research objectives
Deformable image registration
• To estimate non-rigid voxel correspondences between moving & fixed images
• Important for medical image analysis (ex. disease diagnosis)
Multiphase abdominal CT
PET MRI
Types Advantage Limitation
Classical iterative method High topology preservation Substantial time, high computational cost
Supervised learning No parameter tuning Depending on the ground-truth fields
Unsupervised learning No parameter tuning + No labels Potential for folding problem on the fields
[1] Du, Juan, et al., International Journal of Computer Mathematics, 2006.
2
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Research objectives
Deformable image registration
• To estimate non-rigid voxel correspondences between moving & fixed images
• Important for medical image analysis (ex. disease diagnosis)
Multiphase abdominal CT
PET MRI
• To perform image registration along the continuous trajectory
• To apply the denoising diffusion probabilistic model to image registration
Research goal
[1] Du, Juan, et al., International Journal of Computer Mathematics, 2006.
3
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Background: Deformable image registration
• Deformation field: A vector field of displacement vectors for all coordinates in images
• Spatial transformation: Grid sampling to warp moving image into fixed image (ex. linear interpolation)
𝜙∗
= argmin
𝜙
𝑳𝒔𝒊𝒎 𝑚 𝜙 , 𝑓 + 𝑳𝒓𝒆𝒈(𝜙)
Similarity Regularization
Optimization problem in classical algorithms
• Similarity: To evaluate the shape differences between
deformed images and fixed images
• Regularization: To penalize deformation fields
ST: Spatial transformation
𝑓
Fixed image
𝑚
Moving image
ST
𝜙
Deformation field
𝑚(𝜙)
Deformed image
4
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Background: Denoising diffusion probabilistic model (DDPM)
[1] Jonathan Ho et al., NeurIPS 2020.
• To convert a data distribution to Gaussian noise
Forward diffusion process (𝒙𝟎 → 𝒙𝑻)
Gaussian transition for each target data 𝒙0:
⇒ 𝑝 𝒙𝑡 𝒙0 = 𝒩 𝒙𝑡; 𝛼𝑡𝒙0, 1 − 𝛼𝑡 𝑰
where 𝛼𝑡 = ς𝑠=1
𝑡
(1 − 𝛽𝑠)
𝑝 𝒙𝑡 𝒙𝑡−1 = 𝒩(𝒙𝑡; 1 − 𝛽𝑡𝒙𝑡−1, 𝛽𝑡𝑰)
• To sample image from Gaussian noise
Reverse generative process (𝒙𝑻 → 𝒙𝟎)
Parameterized Gaussian process:
𝑝𝜃 𝒙𝑡−1 𝒙𝑡 = 𝒩 𝒙𝑡−1; 𝜇𝜃 𝒙𝑡, 𝑡 , 𝜎𝑡
2
𝑰
𝒙𝑁~𝒩(0, 𝑰)
𝒙𝑡−1 =
1
1−𝛽𝑡
𝒙𝑡 −
𝛽𝑡
1−𝛼𝑡
𝝐𝜃 𝒙𝑡, 𝑡 + 𝜎𝑡𝒛
Generation for 𝑡 = 𝑁, 𝑁 − 1, … , 1:
Parameterized model
Image perturbation: 𝒙𝑡 = 𝛼𝑡𝒙0 + 1 − 𝛼𝑡𝝐
5
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Proposed method: DiffuseMorph
Training phase
min
𝐺𝜃,𝑀𝜓
𝑳𝒅𝒊𝒇𝒇𝒖𝒔𝒊𝒐𝒏 𝑐, 𝑥𝑡, 𝑡 + 𝜆𝑳𝒓𝒆𝒈𝒊𝒔𝒕(𝑚, 𝑓)
𝑳𝒅𝒊𝒇𝒇𝒖𝒔𝒊𝒐𝒏 𝑐, 𝑥𝑡, 𝑡 = 𝔼𝜖,𝑥𝑡,𝑡 𝐺𝜃 𝑐, 𝑥𝑡, 𝑡 − 𝜖 2
2
𝑳𝒓𝒆𝒈𝒊𝒔𝒕 𝑚, 𝑓 = − 𝑚 𝜙 ⊗ 𝑓 + 𝜆𝜙∑ 𝛻𝜙 2
• Diffusion network: To estimate a conditional score function
• Deformation network: To yield the registration field & provide the deformed image
Training in an end-to-end learning manner
Loss function
6
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Proposed method: DiffuseMorph
Inference phase
 Continuous image registration  Synthetic image generation
7
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Experiment 1
Intra-subject 2D face image registration
[1] Balakrishnan, G. et al., CVPR 2018. [2] Dalca, A.V. MICCAI 2018.
8
Methods NMSE x10−1
SSIM 𝐽𝜙 ≤ 0 (%)
Initial 0.363 (0.268) 0.823 (0.066) -
VM [1] 0.047 (0.057) 0.936 (0.024) 0.050 (0.106)
VM-diff [2] 0.034 (0.015) 0.957 (0.013) 0.014 (0.065)
Ours 0.032 (0.017) 0.964 (0.011) 0.017 (0.056)
• RaFD dataset (3 different gazed 8 expressions / Input size = 128x128)
• Training : Validation: Test = 53 : 7 : 7 (subjects)
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Experiment 2
Intra-subject 3D cardiac MR image registration
[1] Balakrishnan, G. et al., CVPR 2018. [2] Dalca, A.V. MICCAI 2018.
9
Methods Dice 𝐽𝜙 ≤ 0 (%)
Initial 0.642 (0.188) -
VM [1] 0.787 (0.113) 0.169 (0.109)
VM-diff [2] 0.794 (0.104) 0.291 (0.188)
Ours 0.802 (0.109) 0.161 (0.082)
• ACDC dataset (4D temporal cardiac MRI data/ input size = 128×128×32)
• Training : Test = 90 : 10 (scans)
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Experiment 3
Atlas-based 3D Brain MR Image Registration
10
Methods Dice 𝐽𝜙 ≤ 0 (%) Time (min)
Initial 0.616 (0.171) - -
SyN [1] 0.752 (0.140) 0.400 (0.100) 122, CPU
VM [2] 0.749 (0.145) 0.553 (0.075) 0.01, GPU
VM-diff [3] 0.731 (0.139) 0.631 (0.073) 0.01, GPU
SYMNet [4] 0.733 (0.148) 0.547 (0.049) 0.43, GPU
MSDIRNet [5] 0.751 (0.142) 0.804 (0.089) 2.06, GPU
CM [6] 0.750 (0.144) 0.510 (0.087) 0.01, GPU
Ours 0.756 (0.139) 0.505 (0.058) 0.01, GPU
• OASIS-3 dataset (T1-weights MRI data / input size = 160×192×224)
• Training : Validation : Test = 1027 : 93 : 129 (scans)
[1] Avants, B.B.et al., Medical image analysis, 2008 [2] Balakrishnan, G. et al., CVPR 2018. [3] Dalca, A.V. MICCAI 2018.
[4] Mok, T.C. et al., CVPR 2020 [5] Lei, Y. et al., Physics in Medicine & Biology, 2020. [6] Kim, B. et al., Medical image analysis 2021.
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Conclusion
• DiffuseMorph learns the conditional score function of deformation
→ To generate synthetic deformed images
→ To provide high-quality image registration from the continuous deformation
• Experimental results on 2D and 3D data suggest the superiority of our method in image registration.
• DiffuseMorph can be used to generate temporal data from the moving to the fixed images.
Propose a novel diffusion-based unsupervised image registration model, called DiffuseMorph.
11
Thank you.
12

More Related Content

What's hot

Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT BoahKim2
 
Brain tumor mri image segmentation and detection
Brain tumor mri image segmentation and detectionBrain tumor mri image segmentation and detection
Brain tumor mri image segmentation and detectioneSAT Publishing House
 
IMAGE SEGMENTATION TECHNIQUES
IMAGE SEGMENTATION TECHNIQUESIMAGE SEGMENTATION TECHNIQUES
IMAGE SEGMENTATION TECHNIQUESVicky Kumar
 
Lec6: Pre-Processing for Nuclear Medicine Images
Lec6: Pre-Processing for Nuclear Medicine ImagesLec6: Pre-Processing for Nuclear Medicine Images
Lec6: Pre-Processing for Nuclear Medicine ImagesUlaş Bağcı
 
Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)
Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)
Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)Universitat Politècnica de Catalunya
 
Semantic Segmentation - Fully Convolutional Networks for Semantic Segmentation
Semantic Segmentation - Fully Convolutional Networks for Semantic SegmentationSemantic Segmentation - Fully Convolutional Networks for Semantic Segmentation
Semantic Segmentation - Fully Convolutional Networks for Semantic Segmentation岳華 杜
 
Deep Learning in Computer Vision
Deep Learning in Computer VisionDeep Learning in Computer Vision
Deep Learning in Computer VisionSungjoon Choi
 
Introduction to Generative Adversarial Networks (GAN) with Apache MXNet
Introduction to Generative Adversarial Networks (GAN) with Apache MXNetIntroduction to Generative Adversarial Networks (GAN) with Apache MXNet
Introduction to Generative Adversarial Networks (GAN) with Apache MXNetAmazon Web Services
 
Lec15: Medical Image Registration (Introduction)
Lec15: Medical Image Registration (Introduction)Lec15: Medical Image Registration (Introduction)
Lec15: Medical Image Registration (Introduction)Ulaş Bağcı
 
Skin lesion detection from dermoscopic images using Convolutional Neural Netw...
Skin lesion detection from dermoscopic images using Convolutional Neural Netw...Skin lesion detection from dermoscopic images using Convolutional Neural Netw...
Skin lesion detection from dermoscopic images using Convolutional Neural Netw...Adrià Romero López
 
Medical Segmentation Decathalon
Medical Segmentation DecathalonMedical Segmentation Decathalon
Medical Segmentation Decathalonimgcommcall
 
Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...
Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...
Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...Ulaş Bağcı
 
CNN and its applications by ketaki
CNN and its applications by ketakiCNN and its applications by ketaki
CNN and its applications by ketakiKetaki Patwari
 
Lec2: Digital Images and Medical Imaging Modalities
Lec2: Digital Images and Medical Imaging ModalitiesLec2: Digital Images and Medical Imaging Modalities
Lec2: Digital Images and Medical Imaging ModalitiesUlaş Bağcı
 

What's hot (20)

Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
 
Brain tumor mri image segmentation and detection
Brain tumor mri image segmentation and detectionBrain tumor mri image segmentation and detection
Brain tumor mri image segmentation and detection
 
IMAGE SEGMENTATION TECHNIQUES
IMAGE SEGMENTATION TECHNIQUESIMAGE SEGMENTATION TECHNIQUES
IMAGE SEGMENTATION TECHNIQUES
 
Computer Vision
Computer VisionComputer Vision
Computer Vision
 
Lec6: Pre-Processing for Nuclear Medicine Images
Lec6: Pre-Processing for Nuclear Medicine ImagesLec6: Pre-Processing for Nuclear Medicine Images
Lec6: Pre-Processing for Nuclear Medicine Images
 
Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)
Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)
Image Segmentation (D3L1 2017 UPC Deep Learning for Computer Vision)
 
Semantic Segmentation - Fully Convolutional Networks for Semantic Segmentation
Semantic Segmentation - Fully Convolutional Networks for Semantic SegmentationSemantic Segmentation - Fully Convolutional Networks for Semantic Segmentation
Semantic Segmentation - Fully Convolutional Networks for Semantic Segmentation
 
Deep Learning in Computer Vision
Deep Learning in Computer VisionDeep Learning in Computer Vision
Deep Learning in Computer Vision
 
Deep Learning for Computer Vision: Object Detection (UPC 2016)
Deep Learning for Computer Vision: Object Detection (UPC 2016)Deep Learning for Computer Vision: Object Detection (UPC 2016)
Deep Learning for Computer Vision: Object Detection (UPC 2016)
 
Introduction to Generative Adversarial Networks (GAN) with Apache MXNet
Introduction to Generative Adversarial Networks (GAN) with Apache MXNetIntroduction to Generative Adversarial Networks (GAN) with Apache MXNet
Introduction to Generative Adversarial Networks (GAN) with Apache MXNet
 
Object detection
Object detectionObject detection
Object detection
 
Cnn
CnnCnn
Cnn
 
Lec15: Medical Image Registration (Introduction)
Lec15: Medical Image Registration (Introduction)Lec15: Medical Image Registration (Introduction)
Lec15: Medical Image Registration (Introduction)
 
Skin lesion detection from dermoscopic images using Convolutional Neural Netw...
Skin lesion detection from dermoscopic images using Convolutional Neural Netw...Skin lesion detection from dermoscopic images using Convolutional Neural Netw...
Skin lesion detection from dermoscopic images using Convolutional Neural Netw...
 
Object tracking
Object trackingObject tracking
Object tracking
 
Medical Segmentation Decathalon
Medical Segmentation DecathalonMedical Segmentation Decathalon
Medical Segmentation Decathalon
 
Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...
Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...
Lec7: Medical Image Segmentation (I) (Radiology Applications of Segmentation,...
 
CNN and its applications by ketaki
CNN and its applications by ketakiCNN and its applications by ketaki
CNN and its applications by ketaki
 
Mask R-CNN
Mask R-CNNMask R-CNN
Mask R-CNN
 
Lec2: Digital Images and Medical Imaging Modalities
Lec2: Digital Images and Medical Imaging ModalitiesLec2: Digital Images and Medical Imaging Modalities
Lec2: Digital Images and Medical Imaging Modalities
 

Similar to DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model

CycleMorph: Cycle consistent unsupervised deformable image registration
CycleMorph: Cycle consistent unsupervised deformable image registrationCycleMorph: Cycle consistent unsupervised deformable image registration
CycleMorph: Cycle consistent unsupervised deformable image registrationBoahKim2
 
SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)Reheman Baikejiang
 
PES ncetec conference
PES ncetec conferencePES ncetec conference
PES ncetec conferenceAvinash P M
 
CARI-2020, Application of LSTM architectures for next frame forecasting in Se...
CARI-2020, Application of LSTM architectures for next frame forecasting in Se...CARI-2020, Application of LSTM architectures for next frame forecasting in Se...
CARI-2020, Application of LSTM architectures for next frame forecasting in Se...Mokhtar SELLAMI
 
Research Paper v2.0
Research Paper v2.0Research Paper v2.0
Research Paper v2.0Kapil Tiwari
 
Automatic System for Detection and Classification of Brain Tumors
Automatic System for Detection and Classification of Brain TumorsAutomatic System for Detection and Classification of Brain Tumors
Automatic System for Detection and Classification of Brain TumorsFatma Sayed Ibrahim
 
MRI_processing
MRI_processingMRI_processing
MRI_processingLitu Rout
 
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...yudhveersingh18
 
Towards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryTowards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryOlalekan Ogunmolu
 
Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...
Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...
Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...AIST
 
Visual Impression Localization of Autonomous Robots_#CASE2015
Visual Impression Localization of Autonomous Robots_#CASE2015Visual Impression Localization of Autonomous Robots_#CASE2015
Visual Impression Localization of Autonomous Robots_#CASE2015Soma Boubou
 
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...sugiuralab
 
COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING
COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING
COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING Pratyusha Mahavadi
 
20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptxssuser7807522
 
Lung Tumour Detection using Image Processing
Lung Tumour Detection using Image Processing Lung Tumour Detection using Image Processing
Lung Tumour Detection using Image Processing Aviral Chaurasia
 
Adaptive unsharp masking
Adaptive unsharp maskingAdaptive unsharp masking
Adaptive unsharp maskingRavi Teja
 
ADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORS
ADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORSADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORS
ADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORSSoma Boubou
 
Brain Tumor - Final Presentation.pdf
Brain Tumor - Final Presentation.pdfBrain Tumor - Final Presentation.pdf
Brain Tumor - Final Presentation.pdfTechNest1
 
Speaker Diarization
Speaker DiarizationSpeaker Diarization
Speaker DiarizationHONGJOO LEE
 

Similar to DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model (20)

CycleMorph: Cycle consistent unsupervised deformable image registration
CycleMorph: Cycle consistent unsupervised deformable image registrationCycleMorph: Cycle consistent unsupervised deformable image registration
CycleMorph: Cycle consistent unsupervised deformable image registration
 
SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)
 
PES ncetec conference
PES ncetec conferencePES ncetec conference
PES ncetec conference
 
CARI-2020, Application of LSTM architectures for next frame forecasting in Se...
CARI-2020, Application of LSTM architectures for next frame forecasting in Se...CARI-2020, Application of LSTM architectures for next frame forecasting in Se...
CARI-2020, Application of LSTM architectures for next frame forecasting in Se...
 
Research Paper v2.0
Research Paper v2.0Research Paper v2.0
Research Paper v2.0
 
Automatic System for Detection and Classification of Brain Tumors
Automatic System for Detection and Classification of Brain TumorsAutomatic System for Detection and Classification of Brain Tumors
Automatic System for Detection and Classification of Brain Tumors
 
MRI_processing
MRI_processingMRI_processing
MRI_processing
 
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
 
Towards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryTowards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgery
 
Inverse problems in medical imaging
Inverse problems in medical imagingInverse problems in medical imaging
Inverse problems in medical imaging
 
Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...
Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...
Artem Baklanov - Votes Aggregation Techniques in Geo-Wiki Crowdsourcing Game:...
 
Visual Impression Localization of Autonomous Robots_#CASE2015
Visual Impression Localization of Autonomous Robots_#CASE2015Visual Impression Localization of Autonomous Robots_#CASE2015
Visual Impression Localization of Autonomous Robots_#CASE2015
 
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
 
COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING
COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING
COMPUTING THE GROWTH RATE OF STEM CELLS USING DIGITAL IMAGE PROCESSING
 
20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx
 
Lung Tumour Detection using Image Processing
Lung Tumour Detection using Image Processing Lung Tumour Detection using Image Processing
Lung Tumour Detection using Image Processing
 
Adaptive unsharp masking
Adaptive unsharp maskingAdaptive unsharp masking
Adaptive unsharp masking
 
ADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORS
ADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORSADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORS
ADAPTIVE FILTER FOR DENOISING 3D DATA CAPTURED BY DEPTH SENSORS
 
Brain Tumor - Final Presentation.pdf
Brain Tumor - Final Presentation.pdfBrain Tumor - Final Presentation.pdf
Brain Tumor - Final Presentation.pdf
 
Speaker Diarization
Speaker DiarizationSpeaker Diarization
Speaker Diarization
 

Recently uploaded

Oxy acetylene welding presentation note.
Oxy acetylene welding presentation note.Oxy acetylene welding presentation note.
Oxy acetylene welding presentation note.eptoze12
 
Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024hassan khalil
 
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionSachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionDr.Costas Sachpazis
 
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ
 
Electronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdfElectronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdfme23b1001
 
Introduction-To-Agricultural-Surveillance-Rover.pptx
Introduction-To-Agricultural-Surveillance-Rover.pptxIntroduction-To-Agricultural-Surveillance-Rover.pptx
Introduction-To-Agricultural-Surveillance-Rover.pptxk795866
 
CCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdf
CCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdfCCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdf
CCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdfAsst.prof M.Gokilavani
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...Soham Mondal
 
IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024Mark Billinghurst
 
HARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IVHARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IVRajaP95
 
chaitra-1.pptx fake news detection using machine learning
chaitra-1.pptx  fake news detection using machine learningchaitra-1.pptx  fake news detection using machine learning
chaitra-1.pptx fake news detection using machine learningmisbanausheenparvam
 
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETEINFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETEroselinkalist12
 
Artificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptxArtificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptxbritheesh05
 
Application of Residue Theorem to evaluate real integrations.pptx
Application of Residue Theorem to evaluate real integrations.pptxApplication of Residue Theorem to evaluate real integrations.pptx
Application of Residue Theorem to evaluate real integrations.pptx959SahilShah
 
Current Transformer Drawing and GTP for MSETCL
Current Transformer Drawing and GTP for MSETCLCurrent Transformer Drawing and GTP for MSETCL
Current Transformer Drawing and GTP for MSETCLDeelipZope
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSKurinjimalarL3
 
Introduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptxIntroduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptxvipinkmenon1
 

Recently uploaded (20)

Oxy acetylene welding presentation note.
Oxy acetylene welding presentation note.Oxy acetylene welding presentation note.
Oxy acetylene welding presentation note.
 
Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024
 
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionSachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
 
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
 
Electronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdfElectronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdf
 
Introduction-To-Agricultural-Surveillance-Rover.pptx
Introduction-To-Agricultural-Surveillance-Rover.pptxIntroduction-To-Agricultural-Surveillance-Rover.pptx
Introduction-To-Agricultural-Surveillance-Rover.pptx
 
CCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdf
CCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdfCCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdf
CCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdf
 
young call girls in Green Park🔝 9953056974 🔝 escort Service
young call girls in Green Park🔝 9953056974 🔝 escort Serviceyoung call girls in Green Park🔝 9953056974 🔝 escort Service
young call girls in Green Park🔝 9953056974 🔝 escort Service
 
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
 
IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024
 
HARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IVHARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IV
 
chaitra-1.pptx fake news detection using machine learning
chaitra-1.pptx  fake news detection using machine learningchaitra-1.pptx  fake news detection using machine learning
chaitra-1.pptx fake news detection using machine learning
 
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
 
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETEINFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
 
Artificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptxArtificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptx
 
Application of Residue Theorem to evaluate real integrations.pptx
Application of Residue Theorem to evaluate real integrations.pptxApplication of Residue Theorem to evaluate real integrations.pptx
Application of Residue Theorem to evaluate real integrations.pptx
 
Current Transformer Drawing and GTP for MSETCL
Current Transformer Drawing and GTP for MSETCLCurrent Transformer Drawing and GTP for MSETCL
Current Transformer Drawing and GTP for MSETCL
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
 
Introduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptxIntroduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptx
 

DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model

  • 1. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Boah Kim, Inhwa Han, and Jong Chul Ye
  • 2. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Research objectives Deformable image registration • To estimate non-rigid voxel correspondences between moving & fixed images • Important for medical image analysis (ex. disease diagnosis) Multiphase abdominal CT PET MRI Types Advantage Limitation Classical iterative method High topology preservation Substantial time, high computational cost Supervised learning No parameter tuning Depending on the ground-truth fields Unsupervised learning No parameter tuning + No labels Potential for folding problem on the fields [1] Du, Juan, et al., International Journal of Computer Mathematics, 2006. 2
  • 3. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Research objectives Deformable image registration • To estimate non-rigid voxel correspondences between moving & fixed images • Important for medical image analysis (ex. disease diagnosis) Multiphase abdominal CT PET MRI • To perform image registration along the continuous trajectory • To apply the denoising diffusion probabilistic model to image registration Research goal [1] Du, Juan, et al., International Journal of Computer Mathematics, 2006. 3
  • 4. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Background: Deformable image registration • Deformation field: A vector field of displacement vectors for all coordinates in images • Spatial transformation: Grid sampling to warp moving image into fixed image (ex. linear interpolation) 𝜙∗ = argmin 𝜙 𝑳𝒔𝒊𝒎 𝑚 𝜙 , 𝑓 + 𝑳𝒓𝒆𝒈(𝜙) Similarity Regularization Optimization problem in classical algorithms • Similarity: To evaluate the shape differences between deformed images and fixed images • Regularization: To penalize deformation fields ST: Spatial transformation 𝑓 Fixed image 𝑚 Moving image ST 𝜙 Deformation field 𝑚(𝜙) Deformed image 4
  • 5. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Background: Denoising diffusion probabilistic model (DDPM) [1] Jonathan Ho et al., NeurIPS 2020. • To convert a data distribution to Gaussian noise Forward diffusion process (𝒙𝟎 → 𝒙𝑻) Gaussian transition for each target data 𝒙0: ⇒ 𝑝 𝒙𝑡 𝒙0 = 𝒩 𝒙𝑡; 𝛼𝑡𝒙0, 1 − 𝛼𝑡 𝑰 where 𝛼𝑡 = ς𝑠=1 𝑡 (1 − 𝛽𝑠) 𝑝 𝒙𝑡 𝒙𝑡−1 = 𝒩(𝒙𝑡; 1 − 𝛽𝑡𝒙𝑡−1, 𝛽𝑡𝑰) • To sample image from Gaussian noise Reverse generative process (𝒙𝑻 → 𝒙𝟎) Parameterized Gaussian process: 𝑝𝜃 𝒙𝑡−1 𝒙𝑡 = 𝒩 𝒙𝑡−1; 𝜇𝜃 𝒙𝑡, 𝑡 , 𝜎𝑡 2 𝑰 𝒙𝑁~𝒩(0, 𝑰) 𝒙𝑡−1 = 1 1−𝛽𝑡 𝒙𝑡 − 𝛽𝑡 1−𝛼𝑡 𝝐𝜃 𝒙𝑡, 𝑡 + 𝜎𝑡𝒛 Generation for 𝑡 = 𝑁, 𝑁 − 1, … , 1: Parameterized model Image perturbation: 𝒙𝑡 = 𝛼𝑡𝒙0 + 1 − 𝛼𝑡𝝐 5
  • 6. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Proposed method: DiffuseMorph Training phase min 𝐺𝜃,𝑀𝜓 𝑳𝒅𝒊𝒇𝒇𝒖𝒔𝒊𝒐𝒏 𝑐, 𝑥𝑡, 𝑡 + 𝜆𝑳𝒓𝒆𝒈𝒊𝒔𝒕(𝑚, 𝑓) 𝑳𝒅𝒊𝒇𝒇𝒖𝒔𝒊𝒐𝒏 𝑐, 𝑥𝑡, 𝑡 = 𝔼𝜖,𝑥𝑡,𝑡 𝐺𝜃 𝑐, 𝑥𝑡, 𝑡 − 𝜖 2 2 𝑳𝒓𝒆𝒈𝒊𝒔𝒕 𝑚, 𝑓 = − 𝑚 𝜙 ⊗ 𝑓 + 𝜆𝜙∑ 𝛻𝜙 2 • Diffusion network: To estimate a conditional score function • Deformation network: To yield the registration field & provide the deformed image Training in an end-to-end learning manner Loss function 6
  • 7. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Proposed method: DiffuseMorph Inference phase  Continuous image registration  Synthetic image generation 7
  • 8. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Experiment 1 Intra-subject 2D face image registration [1] Balakrishnan, G. et al., CVPR 2018. [2] Dalca, A.V. MICCAI 2018. 8 Methods NMSE x10−1 SSIM 𝐽𝜙 ≤ 0 (%) Initial 0.363 (0.268) 0.823 (0.066) - VM [1] 0.047 (0.057) 0.936 (0.024) 0.050 (0.106) VM-diff [2] 0.034 (0.015) 0.957 (0.013) 0.014 (0.065) Ours 0.032 (0.017) 0.964 (0.011) 0.017 (0.056) • RaFD dataset (3 different gazed 8 expressions / Input size = 128x128) • Training : Validation: Test = 53 : 7 : 7 (subjects)
  • 9. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Experiment 2 Intra-subject 3D cardiac MR image registration [1] Balakrishnan, G. et al., CVPR 2018. [2] Dalca, A.V. MICCAI 2018. 9 Methods Dice 𝐽𝜙 ≤ 0 (%) Initial 0.642 (0.188) - VM [1] 0.787 (0.113) 0.169 (0.109) VM-diff [2] 0.794 (0.104) 0.291 (0.188) Ours 0.802 (0.109) 0.161 (0.082) • ACDC dataset (4D temporal cardiac MRI data/ input size = 128×128×32) • Training : Test = 90 : 10 (scans)
  • 10. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Experiment 3 Atlas-based 3D Brain MR Image Registration 10 Methods Dice 𝐽𝜙 ≤ 0 (%) Time (min) Initial 0.616 (0.171) - - SyN [1] 0.752 (0.140) 0.400 (0.100) 122, CPU VM [2] 0.749 (0.145) 0.553 (0.075) 0.01, GPU VM-diff [3] 0.731 (0.139) 0.631 (0.073) 0.01, GPU SYMNet [4] 0.733 (0.148) 0.547 (0.049) 0.43, GPU MSDIRNet [5] 0.751 (0.142) 0.804 (0.089) 2.06, GPU CM [6] 0.750 (0.144) 0.510 (0.087) 0.01, GPU Ours 0.756 (0.139) 0.505 (0.058) 0.01, GPU • OASIS-3 dataset (T1-weights MRI data / input size = 160×192×224) • Training : Validation : Test = 1027 : 93 : 129 (scans) [1] Avants, B.B.et al., Medical image analysis, 2008 [2] Balakrishnan, G. et al., CVPR 2018. [3] Dalca, A.V. MICCAI 2018. [4] Mok, T.C. et al., CVPR 2020 [5] Lei, Y. et al., Physics in Medicine & Biology, 2020. [6] Kim, B. et al., Medical image analysis 2021.
  • 11. DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model Conclusion • DiffuseMorph learns the conditional score function of deformation → To generate synthetic deformed images → To provide high-quality image registration from the continuous deformation • Experimental results on 2D and 3D data suggest the superiority of our method in image registration. • DiffuseMorph can be used to generate temporal data from the moving to the fixed images. Propose a novel diffusion-based unsupervised image registration model, called DiffuseMorph. 11