Presentation file for "Medical Image Synthesis with Improved Cycle-GAN: CT from CECT" presented at the Workshop on Deep Learning for Biomedical Image Reconstruction of IEEE Internetional Symposium on Biomedical Imaging, ISBI 2020.
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion ModelBoahKim2
Presentation file for "DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model" presented at European Conference on Computer Vision, ECCV 2022.
Diffusion Deformable Model for 4D Temporal Medical Image GenerationBoahKim2
Presentation file for "Diffusion Deformable Model for 4D Temporal Medical Image Generation" presented at the International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022.
Photo-realistic Single Image Super-resolution using a Generative Adversarial ...Hansol Kang
* Ledig, Christian, et al. "Photo-realistic single image super-resolution using a generative adversarial network." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.
발표자: 박태성 (UC Berkeley 박사과정)
발표일: 2017.6.
Taesung Park is a Ph.D. student at UC Berkeley in AI and computer vision, advised by Prof. Alexei Efros.
His research interest lies between computer vision and computational photography, such as generating realistic images or enhancing photo qualities. He received B.S. in mathematics and M.S. in computer science from Stanford University.
개요:
Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs.
However, for many tasks, paired training data will not be available.
We present an approach for learning to translate an image from a source domain X to a target domain Y in the absence of paired examples.
Our goal is to learn a mapping G: X → Y such that the distribution of images from G(X) is indistinguishable from the distribution Y using an adversarial loss.
Because this mapping is highly under-constrained, we couple it with an inverse mapping F: Y → X and introduce a cycle consistency loss to push F(G(X)) ≈ X (and vice versa).
Qualitative results are presented on several tasks where paired training data does not exist, including collection style transfer, object transfiguration, season transfer, photo enhancement, etc.
Quantitative comparisons against several prior methods demonstrate the superiority of our approach.
DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion ModelBoahKim2
Presentation file for "DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model" presented at European Conference on Computer Vision, ECCV 2022.
Diffusion Deformable Model for 4D Temporal Medical Image GenerationBoahKim2
Presentation file for "Diffusion Deformable Model for 4D Temporal Medical Image Generation" presented at the International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022.
Photo-realistic Single Image Super-resolution using a Generative Adversarial ...Hansol Kang
* Ledig, Christian, et al. "Photo-realistic single image super-resolution using a generative adversarial network." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.
발표자: 박태성 (UC Berkeley 박사과정)
발표일: 2017.6.
Taesung Park is a Ph.D. student at UC Berkeley in AI and computer vision, advised by Prof. Alexei Efros.
His research interest lies between computer vision and computational photography, such as generating realistic images or enhancing photo qualities. He received B.S. in mathematics and M.S. in computer science from Stanford University.
개요:
Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs.
However, for many tasks, paired training data will not be available.
We present an approach for learning to translate an image from a source domain X to a target domain Y in the absence of paired examples.
Our goal is to learn a mapping G: X → Y such that the distribution of images from G(X) is indistinguishable from the distribution Y using an adversarial loss.
Because this mapping is highly under-constrained, we couple it with an inverse mapping F: Y → X and introduce a cycle consistency loss to push F(G(X)) ≈ X (and vice versa).
Qualitative results are presented on several tasks where paired training data does not exist, including collection style transfer, object transfiguration, season transfer, photo enhancement, etc.
Quantitative comparisons against several prior methods demonstrate the superiority of our approach.
Lec11: Active Contour and Level Set for Medical Image SegmentationUlaş Bağcı
ActiveContour(Snake) • LevelSet
• Applications
Enhancement, Noise Reduction, and Signal Processing • MedicalImageRegistration • MedicalImageSegmentation • MedicalImageVisualization • Machine Learning in Medical Imaging • Shape Modeling/Analysis of Medical Images Deep Learning in Radiology Fuzzy Connectivity (FC) – Affinity functions • Absolute FC • Relative FC (and Iterative Relative FC) • Successful example applications of FC in medical imaging • Segmentation of Airway and Airway Walls using RFC based method Energyfunctional – Data and Smoothness terms • GraphCut – Min cut – Max Flow • ApplicationsinRadiologyImages
classify images from the CIFAR-10 dataset. The dataset consists of airplanes, dogs, cats, and other objects.we'll preprocess the images, then train a convolutional neural network on all the samples. The images need to be normalized and the labels need to be one-hot encoded.
Codetecon #KRK 3 - Object detection with Deep LearningMatthew Opala
There’s been enormous progress in object detection algorithms. Starting from multi-stage ones like R-CNN to end-to-end ones like SSD or YOLO, accuracy of the methods improved significantly. Current applications include pedestrian detection for cars and face detection on facebook.
But that’s just the beginning. I am going to show the algorithms for solving the problem, show what’s currently possible, and what will be possible in the near future.
Towards Total Recall in Industrial Anomaly Detectionharmonylab
公開URL:https://openaccess.thecvf.com/content/CVPR2022/papers/Roth_Towards_Total_Recall_in_Industrial_Anomaly_Detection_CVPR_2022_paper.pdf
出典:Karsten Roth, Latha Pemula, Joaquin Zepeda, Bernhard Schölkopf, Thomas Brox, Peter Gehler: Towards Total Recall in Industrial Anomaly Detection, Conference on Computer Vision and Pattern Recognition (CVPR), pp. 14318-14328 (2022)
概要:本論文では位置情報を考慮した特徴量の集合和であるメモリバンクとCoresetによる画像パッチ特徴量の削減を行うPatchCoreアルゴリズムを提案する.結果として、異常検出のベンチマークであるMVTecにおいてAUROC99%以上の精度を出力し,2022年時点でのSoTAを記録した.また,PatchCoreによる特徴量削減により,学習のサンプル数を20%に減らした場合でも以前のSoTAに匹敵する精度となった.
This talk will cover various medical applications of deep learning including tumor segmentation in histology slides, MRI, CT, and X-Ray data. Also, more complicated tasks such as cell counting where the challenge is to count how many objects are in an image. It will also cover generative adversarial networks and how they can be used for medical applications. This presentation is accessible to non-doctors and non-computer scientists.
Lec14: Evaluation Framework for Medical Image SegmentationUlaş Bağcı
How to evaluate accuracy of image segmentation?
– Gold standard ~ surrogate of truths
– Qualitative • Visual
• Inter-andintra-observeragreementrates – Quantitative
• Volumetricmeasurements(regression) • Regionoverlaps
• Shapebasedmeasurements
• Theoreticalcomparisons
• STAPLE,Uncertaintyguidance,andevaluationw/otruths
Clustering – K-means – FCM (fuzzyc-means) – SMC (simple membership based clustering) – AP(affinity propagation) – FLAB(fuzzy locally adaptive Bayesian) – Spectral Clustering Methods ShapeModeling – M-reps – Active Shape Models (ASM) – Oriented Active Shape Models (OASM) – Application in anatomy recognition and segmentation – Comparison of ASM and OASM ActiveContour(Snake) • LevelSet • Applications Enhancement, Noise Reduction, and Signal Processing • MedicalImageRegistration • MedicalImageSegmentation • MedicalImageVisualization • Machine Learning in Medical Imaging • Shape Modeling/Analysis of Medical Images Deep Learning in Radiology Fuzzy Connectivity (FC) – Affinity functions • Absolute FC • Relative FC (and Iterative Relative FC) • Successful example applications of FC in medical imaging • Segmentation of Airway and Airway Walls using RFC based method Energy functional – Data and Smoothness terms • GraphCut – Min cut – Max Flow • ApplicationsinRadiologyImages
Lec11: Active Contour and Level Set for Medical Image SegmentationUlaş Bağcı
ActiveContour(Snake) • LevelSet
• Applications
Enhancement, Noise Reduction, and Signal Processing • MedicalImageRegistration • MedicalImageSegmentation • MedicalImageVisualization • Machine Learning in Medical Imaging • Shape Modeling/Analysis of Medical Images Deep Learning in Radiology Fuzzy Connectivity (FC) – Affinity functions • Absolute FC • Relative FC (and Iterative Relative FC) • Successful example applications of FC in medical imaging • Segmentation of Airway and Airway Walls using RFC based method Energyfunctional – Data and Smoothness terms • GraphCut – Min cut – Max Flow • ApplicationsinRadiologyImages
classify images from the CIFAR-10 dataset. The dataset consists of airplanes, dogs, cats, and other objects.we'll preprocess the images, then train a convolutional neural network on all the samples. The images need to be normalized and the labels need to be one-hot encoded.
Codetecon #KRK 3 - Object detection with Deep LearningMatthew Opala
There’s been enormous progress in object detection algorithms. Starting from multi-stage ones like R-CNN to end-to-end ones like SSD or YOLO, accuracy of the methods improved significantly. Current applications include pedestrian detection for cars and face detection on facebook.
But that’s just the beginning. I am going to show the algorithms for solving the problem, show what’s currently possible, and what will be possible in the near future.
Towards Total Recall in Industrial Anomaly Detectionharmonylab
公開URL:https://openaccess.thecvf.com/content/CVPR2022/papers/Roth_Towards_Total_Recall_in_Industrial_Anomaly_Detection_CVPR_2022_paper.pdf
出典:Karsten Roth, Latha Pemula, Joaquin Zepeda, Bernhard Schölkopf, Thomas Brox, Peter Gehler: Towards Total Recall in Industrial Anomaly Detection, Conference on Computer Vision and Pattern Recognition (CVPR), pp. 14318-14328 (2022)
概要:本論文では位置情報を考慮した特徴量の集合和であるメモリバンクとCoresetによる画像パッチ特徴量の削減を行うPatchCoreアルゴリズムを提案する.結果として、異常検出のベンチマークであるMVTecにおいてAUROC99%以上の精度を出力し,2022年時点でのSoTAを記録した.また,PatchCoreによる特徴量削減により,学習のサンプル数を20%に減らした場合でも以前のSoTAに匹敵する精度となった.
This talk will cover various medical applications of deep learning including tumor segmentation in histology slides, MRI, CT, and X-Ray data. Also, more complicated tasks such as cell counting where the challenge is to count how many objects are in an image. It will also cover generative adversarial networks and how they can be used for medical applications. This presentation is accessible to non-doctors and non-computer scientists.
Lec14: Evaluation Framework for Medical Image SegmentationUlaş Bağcı
How to evaluate accuracy of image segmentation?
– Gold standard ~ surrogate of truths
– Qualitative • Visual
• Inter-andintra-observeragreementrates – Quantitative
• Volumetricmeasurements(regression) • Regionoverlaps
• Shapebasedmeasurements
• Theoreticalcomparisons
• STAPLE,Uncertaintyguidance,andevaluationw/otruths
Clustering – K-means – FCM (fuzzyc-means) – SMC (simple membership based clustering) – AP(affinity propagation) – FLAB(fuzzy locally adaptive Bayesian) – Spectral Clustering Methods ShapeModeling – M-reps – Active Shape Models (ASM) – Oriented Active Shape Models (OASM) – Application in anatomy recognition and segmentation – Comparison of ASM and OASM ActiveContour(Snake) • LevelSet • Applications Enhancement, Noise Reduction, and Signal Processing • MedicalImageRegistration • MedicalImageSegmentation • MedicalImageVisualization • Machine Learning in Medical Imaging • Shape Modeling/Analysis of Medical Images Deep Learning in Radiology Fuzzy Connectivity (FC) – Affinity functions • Absolute FC • Relative FC (and Iterative Relative FC) • Successful example applications of FC in medical imaging • Segmentation of Airway and Airway Walls using RFC based method Energy functional – Data and Smoothness terms • GraphCut – Min cut – Max Flow • ApplicationsinRadiologyImages
Enlarge Medical Image using Line-Column Interpolation (LCI) Method IJECEIAES
Quality of medical image has an important role in constructing right medical diagnosis. This paper recommends a method to improve the quality of medical images by increasing the size of the image pixels. By increasing the size of pixels, the size of the objects contained therein is also greater, making it easier to observe. In this study medical images of Brain CT-Scan, Chest X-Ray and Panoramic X-Ray were processed using Line-Column Interpolation (LCI) Method. The results of the treatment are then compared to Nearest Neighbor Interpolation (NNI), Bilinear Interpolation (BLI) and Bicubic Interpolation (BCI) processing results. The experiment shows that Line-Column Interpolation Method produces a larger image with details of the objects in it are not blurred and has equal visual effects. Thus, this method is expected to be a reference material in enlarging the size of the medical image for ease in clinical analysis.
MICCS: A Novel Framework for Medical Image Compression Using Compressive Sens...IJECEIAES
The vision of some particular applications such as robot-guided remote surgery where the image of a patient body will need to be captured by the smart visual sensor and to be sent on a real-time basis through a network of high bandwidth but yet limited. The particular problem considered for the study is to develop a mechanism of a hybrid approach of compression where the Region-ofInterest (ROI) should be compressed with lossless compression techniques and Non-ROI should be compressed with Compressive Sensing (CS) techniques. So the challenge is gaining equal image quality for both ROI and Non-ROI while overcoming optimized dimension reduction by sparsity into Non-ROI. It is essential to retain acceptable visual quality to Non-ROI compressed region to obtain a better reconstructed image. This step could bridge the trade-off between image quality and traffic load. The study outcomes were compared with traditional hybrid compression methods to find that proposed method achieves better compression performance as compared to conventional hybrid compression techniques on the performances parameters e.g. PSNR, MSE, and Compression Ratio.
Unsupervised Deformable Image Registration Using Cycle-Consistent CNNBoahKim2
Presentation file for "Unsupervised Deformable Image Registration Using Cycle-Consistent CNN" presented at the International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2019.
Welcome to International Journal of Engineering Research and Development (IJERD)IJERD Editor
call for paper 2012, hard copy of journal, research paper publishing, where to publish research paper,
journal publishing, how to publish research paper, Call For research paper, international journal, publishing a paper, IJERD, journal of science and technology, how to get a research paper published, publishing a paper, publishing of journal, publishing of research paper, reserach and review articles, IJERD Journal, How to publish your research paper, publish research paper, open access engineering journal, Engineering journal, Mathemetics journal, Physics journal, Chemistry journal, Computer Engineering, Computer Science journal, how to submit your paper, peer reviw journal, indexed journal, reserach and review articles, engineering journal, www.ijerd.com, research journals
Two Dimensional Image Reconstruction Algorithmsmastersrihari
Convolution Back-Projection (CBP) Algorithm was used for the reconstruction of the image. The performance was compared by implementing the algorithm by using RAM- LAK filter, Shepp- Logan filter and also No filter being used.
An efficient image compression algorithm using dct biorthogonal wavelet trans...eSAT Journals
Abstract
Recently the digital imaging applications is increasing significantly and it leads the requirement of effective image compression techniques. Image compression removes the redundant information from an image. By using it we can able to store only the necessary information which helps to reduce the transmission bandwidth, transmission time and storage size of image. This paper proposed a new image compression technique using DCT-Biorthogonal Wavelet Transform with arithmetic coding for improvement the visual quality of an image. It is a simple technique for getting better compression results. In this new algorithm firstly Biorthogonal wavelet transform is applied and then 2D DCT-Biorthogonal wavelet transform is applied on each block of the low frequency sub band. Finally, split all values from each transformed block and arithmetic coding is applied for image compression.
Keywords: Arithmetic coding, Biorthogonal wavelet Transform, DCT, Image Compression.
BIG DATA-DRIVEN FAST REDUCING THE VISUAL BLOCK ARTIFACTS OF DCT COMPRESSED IM...IJDKP
The Urban Surveillance Systems generate huge amount of video and image data and impose high pressure
onto the recording disks. It is obvious that the research of video is a key point of big data research areas.
Since videos are composed of images, the degree and efficiency of image compression are of great
importance. Although the DCT based JPEG standard are widely used, it encounters insurmountable
problems. For instance, image encoding deficiencies such as block artifacts have to be removed frequently.
In this paper, we propose a new, simple but effective method to fast reduce the visual block artifacts of DCT
compressed images for urban surveillance systems. The simulation results demonstrate that our proposed
method achieves better quality than widely used filters while consuming much less computer CPU
resources.
Diabetic retinopathy is one of the leading complication of diabetes and also one of the leading preventable blindness. Early diagnosis and treatment may prevent such condition or in other words, annoyance of the disease may be overcome. The fundus images produced by automated fundus camera are often noisy making it difficult for doctors to precisely detect the abnormalities in fundus images. In the present paper, we propose to use vessel extraction of Retinal image enhancement and implemented in Raspberry Pi board using opencv library for faster execution and cost effective processing unit which helps during mass screening of diabetic retinopathy. The effectiveness of the proposed techniques is evaluated using different metrics and Micro-aneurysms. Finally, a considerable improvement in the enhancement of the Diabetic Retinopathy images is achieved.
Water scarcity is the lack of fresh water resources to meet the standard water demand. There are two type of water scarcity. One is physical. The other is economic water scarcity.
Cosmetic shop management system project report.pdfKamal Acharya
Buying new cosmetic products is difficult. It can even be scary for those who have sensitive skin and are prone to skin trouble. The information needed to alleviate this problem is on the back of each product, but it's thought to interpret those ingredient lists unless you have a background in chemistry.
Instead of buying and hoping for the best, we can use data science to help us predict which products may be good fits for us. It includes various function programs to do the above mentioned tasks.
Data file handling has been effectively used in the program.
The automated cosmetic shop management system should deal with the automation of general workflow and administration process of the shop. The main processes of the system focus on customer's request where the system is able to search the most appropriate products and deliver it to the customers. It should help the employees to quickly identify the list of cosmetic product that have reached the minimum quantity and also keep a track of expired date for each cosmetic product. It should help the employees to find the rack number in which the product is placed.It is also Faster and more efficient way.
Student information management system project report ii.pdfKamal Acharya
Our project explains about the student management. This project mainly explains the various actions related to student details. This project shows some ease in adding, editing and deleting the student details. It also provides a less time consuming process for viewing, adding, editing and deleting the marks of the students.
Final project report on grocery store management system..pdfKamal Acharya
In today’s fast-changing business environment, it’s extremely important to be able to respond to client needs in the most effective and timely manner. If your customers wish to see your business online and have instant access to your products or services.
Online Grocery Store is an e-commerce website, which retails various grocery products. This project allows viewing various products available enables registered users to purchase desired products instantly using Paytm, UPI payment processor (Instant Pay) and also can place order by using Cash on Delivery (Pay Later) option. This project provides an easy access to Administrators and Managers to view orders placed using Pay Later and Instant Pay options.
In order to develop an e-commerce website, a number of Technologies must be studied and understood. These include multi-tiered architecture, server and client-side scripting techniques, implementation technologies, programming language (such as PHP, HTML, CSS, JavaScript) and MySQL relational databases. This is a project with the objective to develop a basic website where a consumer is provided with a shopping cart website and also to know about the technologies used to develop such a website.
This document will discuss each of the underlying technologies to create and implement an e- commerce website.
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
1. Medical Image Synthesis with Improved Cycle-GAN
CT from CECT
Workshop: Deep Learning for Biomedical Image Reconstruction
Boah Kim, Eung Yeop Kim, Jong Chul Ye
2. Contents
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 2
Boah Kim
1. Introduction
2. Method
3. Experiment
4. Conclusion
3. Introduction
Medical image synthesis
An approach to modeling a mapping from given images to unknown images
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
• Potential risks of radiation exposure for multiple acquisition of medical images (ex. CT, PET)
• Not always accessible modalities for every patient
• Alignment issue on the analysis of multi-images
Multi-modality image synthesis Inter-modality image synthesis
MRI CT 3T MRI 7T MRI
Necessity
ISBI2020 Workshop Presentation 3
Boah Kim
Nie, Dong, et al. "Medical image synthesis with deep convolutional adversarial networks." IEEE Transactions on Biomedical Engineering
4. Introduction
Deep learning approaches for image synthesis
Based on generative adversarial networks (GAN) model
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
Conditional GAN model
Auto-context model with GAN
• Image generation by solving the min-max problem with generator and discriminator
• Useful for the dataset that does not have ground-truth images
ISBI2020 Workshop Presentation 4
Boah Kim
DCGAN model
Zhang, Qianqian, et al. ICHI, 2018 Dar, Salman UH, et al. IEEE TMI, 2019
Nie, Dong, et al. IEEE TBE, 2018
5. Introduction
CT image synthesis from CECT
CECT and unenhanced CT
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 5
Boah Kim
CECT Unenhanced CT
• CECT : highlight specific tissues or parts of body by contrast agent
• Useful for diagnosis by extracting certain organ (ex. bone)
Challenge
Design “CT image synthesis model using unsupervised deep learning”
• Decrease the risk of radiation exposure
• Do not require alignment for image analysis
• Not aligned CT & CECT → Do not have ground-truth images
• Only have to change the enhanced blood vessels
• Do not generate or remove certain organs/tissues in medical images
6. Contents
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 6
Boah Kim
1. Introduction
2. Method
3. Experiment
4. Conclusion
7. Method
Motivation: cycleGAN
One-to-one mapping using cyclic constraint
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 7
Boah Kim
Zhu, Jun-Yan, et al. "Unpaired image-to-image translation using
cycle-consistent adversarial networks,” CVPR, 2017
• Two Generators
Generate the unknown image from source image
• Two Discriminators
Distinguish between real and fake images
• Loss function
= 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐷𝐷𝐴𝐴 + 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐵𝐵
+ 𝜆𝜆1𝐿𝐿𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵 + 𝜆𝜆2𝐿𝐿𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵)
8. Method
Motivation: cycleGAN
Why using cycleGAN?
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 8
Boah Kim
B.Sim et al. “Optimal transport, cycleGAN, and penalized LS for unsupervised learning in inverse problems”
• Stochastic generalization of penalized least square (PLS)
Penalized least square loss with deep learning prior
Optimal transport (OT)
x: stochastic variable, y : given variable
x: stochastic variable, y: stochastic variable
9. Method
Motivation: cycleGAN
Why using cycleGAN?
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 9
Boah Kim
• Stochastic generalization of penalized least square (PLS)
B.Sim et al. “Optimal transport, cycleGAN, and penalized LS for unsupervised learning in inverse problems”
Penalized least square loss with deep learning prior
Optimal transport (OT)
x: stochastic variable, y : given variable
x: stochastic variable, y: stochastic variable
Cyclic loss GAN loss
- cycleGAN can be explained by optimal transport.
- Cyclic loss is derived from the PLS.
10. Method
Overall framework
Improved cycleGAN with residual learning
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 10
Boah Kim
11. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 11
Boah Kim
• Residual learning for generators
CNN
Prevent loss of
medical information
Method
Overall framework
Improved cycleGAN with residual learning
• Three Discriminators
𝐷𝐷𝐴𝐴 : Distinguish real and fake CT
𝐷𝐷𝐵𝐵 : Distinguish real and fake CECT
𝑫𝑫𝑪𝑪 : Distinguish real and fake pair of CT & CECT
• Two Generators
𝐺𝐺𝐴𝐴𝐴𝐴 : CECT → CT (generate synthetic CT)
𝐺𝐺𝐵𝐵𝐵𝐵 : CT → CECT (generate synthetic CECT)
12. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 12
Boah Kim
Method
Overall framework
Improved cycleGAN with residual learning
• Residual learning for generators
CNN
Prevent loss of
medical information
• Three Discriminators
𝐷𝐷𝐴𝐴 : Distinguish real and fake CT
𝐷𝐷𝐵𝐵 : Distinguish real and fake CECT
𝑫𝑫𝑪𝑪 : Distinguish real and fake pair of CT & CECT
• Two Generators
𝐺𝐺𝐴𝐴𝐴𝐴 : CECT → CT (generate synthetic CT)
𝐺𝐺𝐵𝐵𝐵𝐵 : CT → CECT (generate synthetic CECT)
13. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 13
Boah Kim
Method
Overall framework
Improved cycleGAN with residual learning
• Residual learning for generators
CNN
Prevent loss of
medical information
• Three Discriminators
𝐷𝐷𝐴𝐴 : Distinguish real and fake CT
𝐷𝐷𝐵𝐵 : Distinguish real and fake CECT
𝑫𝑫𝑪𝑪 : Distinguish real and fake pair of CT & CECT
• Two Generators
𝐺𝐺𝐴𝐴𝐴𝐴 : CECT → CT (generate synthetic CT)
𝐺𝐺𝐵𝐵𝐵𝐵 : CT → CECT (generate synthetic CECT)
14. Training networks by solving the optimization problem
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 14
Boah Kim
𝐿𝐿 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵, 𝐷𝐷𝐶𝐶
= 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐷𝐷𝐴𝐴 + 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐵𝐵 + 𝐿𝐿𝐺𝐺𝐺𝐺𝑁𝑁′ 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐶𝐶
+ 𝜆𝜆1𝐿𝐿𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵 + 𝜆𝜆2𝐿𝐿𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵)
min
𝐺𝐺𝐴𝐴𝐴𝐴,𝐺𝐺𝐵𝐵𝐵𝐵
max
𝐷𝐷𝐴𝐴,𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶
𝐿𝐿(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶)
Method
Loss function
15. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 15
Boah Kim
𝑳𝑳𝑮𝑮𝑮𝑮𝑮𝑮(𝑮𝑮𝑨𝑨𝑨𝑨, 𝑫𝑫𝑨𝑨)
min
𝐺𝐺𝐴𝐴𝐴𝐴
𝔼𝔼𝑥𝑥𝐴𝐴~𝑃𝑃𝐴𝐴
[ 𝐷𝐷𝐴𝐴 𝐺𝐺𝐴𝐴𝐴𝐴 𝑥𝑥𝐴𝐴 − 1
2
]
min
𝐷𝐷𝐴𝐴
1
2
𝔼𝔼𝑥𝑥𝐵𝐵~𝑃𝑃𝐵𝐵
[ 𝐷𝐷𝐴𝐴 𝑥𝑥𝐵𝐵 − 1 2
] +
1
2
𝔼𝔼𝑥𝑥𝐴𝐴~𝑃𝑃𝐴𝐴
[𝐷𝐷𝐴𝐴 𝐺𝐺𝐴𝐴𝐴𝐴 𝑥𝑥𝐴𝐴
2
]
• Adversarial loss,
- Produce realistic images
- Apply to input 𝑥𝑥
Training networks by solving the optimization problem
𝐿𝐿 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵, 𝐷𝐷𝐶𝐶
= 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐷𝐷𝐴𝐴 + 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐵𝐵 + 𝐿𝐿𝐺𝐺𝐺𝐺𝑁𝑁′ 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐶𝐶
+ 𝜆𝜆1𝐿𝐿𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵 + 𝜆𝜆2𝐿𝐿𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵)
min
𝐺𝐺𝐴𝐴𝐴𝐴,𝐺𝐺𝐵𝐵𝐵𝐵
max
𝐷𝐷𝐴𝐴,𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶
𝐿𝐿(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶)
Method
Loss function
16. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 16
Boah Kim
𝑳𝑳𝑮𝑮𝑮𝑮𝑵𝑵′(𝑮𝑮𝑨𝑨𝑨𝑨, 𝑮𝑮𝑩𝑩𝑩𝑩, 𝑫𝑫𝑪𝑪)
- Produce realistic images
- Apply to a paired input 𝑥𝑥𝐴𝐴 and 𝑥𝑥𝐵𝐵
min
𝐺𝐺𝐴𝐴𝐴𝐴,𝐺𝐺𝐵𝐵𝐵𝐵
𝔼𝔼𝑥𝑥𝐴𝐴~𝑃𝑃𝐴𝐴,𝑥𝑥𝐵𝐵~𝑃𝑃𝐵𝐵
[ 𝐷𝐷𝐶𝐶 𝐺𝐺𝐴𝐴𝐴𝐴 𝑥𝑥𝐴𝐴 , 𝐺𝐺𝐵𝐵𝐵𝐵(𝑥𝑥𝐵𝐵 − 1 2
]
min
𝐷𝐷𝐶𝐶
1
2
𝔼𝔼𝑥𝑥𝐴𝐴~𝑃𝑃𝐴𝐴,𝑥𝑥𝐵𝐵~𝑃𝑃𝐵𝐵
𝐷𝐷𝐶𝐶 𝑥𝑥𝐵𝐵, 𝑥𝑥𝐴𝐴 − 1 2
+
1
2
𝔼𝔼𝑥𝑥𝐴𝐴~𝑃𝑃𝐴𝐴,𝑥𝑥𝐵𝐵~𝑃𝑃𝐵𝐵
[𝐷𝐷𝐶𝐶 𝐺𝐺𝐴𝐴𝐴𝐴 𝑥𝑥𝐴𝐴 , 𝐺𝐺𝐵𝐵𝐵𝐵(𝑥𝑥𝐵𝐵) 2
]
Training networks by solving the optimization problem
𝐿𝐿 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵, 𝐷𝐷𝐶𝐶
= 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐷𝐷𝐴𝐴 + 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐵𝐵 + 𝐿𝐿𝐺𝐺𝐺𝐺𝑁𝑁′ 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐶𝐶
+ 𝜆𝜆1𝐿𝐿𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵 + 𝜆𝜆2𝐿𝐿𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵)
min
𝐺𝐺𝐴𝐴𝐴𝐴,𝐺𝐺𝐵𝐵𝐵𝐵
max
𝐷𝐷𝐴𝐴,𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶
𝐿𝐿(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶)
Method
Loss function
• Adversarial loss,
17. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 17
Boah Kim
• Cyclic loss, 𝑳𝑳𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄(𝑮𝑮𝑨𝑨𝑨𝑨, 𝑮𝑮𝑩𝑩𝑩𝑩)
Guarantee one-to-one mapping of CT & CECT
𝔼𝔼𝑥𝑥𝐴𝐴~𝑃𝑃𝐴𝐴
𝐺𝐺𝐵𝐵𝐵𝐵 𝐺𝐺𝐴𝐴𝐴𝐴 𝑥𝑥𝐴𝐴 − 𝑥𝑥𝐴𝐴 1
+𝔼𝔼𝑥𝑥𝐵𝐵~𝑃𝑃𝐵𝐵
𝐺𝐺𝐴𝐴𝐴𝐴 𝐺𝐺𝐵𝐵𝐵𝐵 𝑥𝑥𝐵𝐵 − 𝑥𝑥𝐵𝐵 1
Training networks by solving the optimization problem
𝐿𝐿 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵, 𝐷𝐷𝐶𝐶
= 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐷𝐷𝐴𝐴 + 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐵𝐵 + 𝐿𝐿𝐺𝐺𝐺𝐺𝑁𝑁′ 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐶𝐶
+ 𝜆𝜆1𝐿𝐿𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵 + 𝜆𝜆2𝐿𝐿𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵)
min
𝐺𝐺𝐴𝐴𝐴𝐴,𝐺𝐺𝐵𝐵𝐵𝐵
max
𝐷𝐷𝐴𝐴,𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶
𝐿𝐿(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶)
Method
Loss function
18. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 18
Boah Kim
• Identity loss, 𝑳𝑳𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊(𝑮𝑮𝑨𝑨𝑨𝑨, 𝑮𝑮𝑩𝑩𝑩𝑩)
Preserve medical information between CT & CECT
𝔼𝔼𝑥𝑥𝐴𝐴~𝑃𝑃𝐴𝐴
𝐺𝐺𝐵𝐵𝐵𝐵 𝑥𝑥𝐴𝐴 − 𝑥𝑥𝐴𝐴 1
+𝔼𝔼𝑥𝑥𝐵𝐵~𝑃𝑃𝐵𝐵
𝐺𝐺𝐴𝐴𝐴𝐴 𝑥𝑥𝐵𝐵 − 𝑥𝑥𝐵𝐵 1
𝑮𝑮𝑩𝑩𝑩𝑩
𝑮𝑮𝑨𝑨𝑨𝑨
CT
CECT
CT
CECT
Training networks by solving the optimization problem
𝐿𝐿 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵, 𝐷𝐷𝐶𝐶
= 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐷𝐷𝐴𝐴 + 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐵𝐵 + 𝐿𝐿𝐺𝐺𝐺𝐺𝑁𝑁′ 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐶𝐶
+ 𝜆𝜆1𝐿𝐿𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵 + 𝜆𝜆2𝐿𝐿𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵)
min
𝐺𝐺𝐴𝐴𝐴𝐴,𝐺𝐺𝐵𝐵𝐵𝐵
max
𝐷𝐷𝐴𝐴,𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶
𝐿𝐿(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶)
Method
Loss function
19. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 19
Boah Kim
• Minimize our designed loss function
(adversarial loss + cyclic loss + identity loss)
Training networks by solving the optimization problem
𝐿𝐿 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵, 𝐷𝐷𝐶𝐶
= 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐷𝐷𝐴𝐴 + 𝐿𝐿𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐵𝐵 + 𝐿𝐿𝐺𝐺𝐺𝐺𝑁𝑁′ 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐶𝐶
+ 𝜆𝜆1𝐿𝐿𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵 + 𝜆𝜆2𝐿𝐿𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵)
min
𝐺𝐺𝐴𝐴𝐴𝐴,𝐺𝐺𝐵𝐵𝐵𝐵
max
𝐷𝐷𝐴𝐴,𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶
𝐿𝐿(𝐺𝐺𝐴𝐴𝐴𝐴, 𝐺𝐺𝐵𝐵𝐵𝐵, 𝐷𝐷𝐴𝐴, 𝐷𝐷𝐵𝐵,𝐷𝐷𝐶𝐶)
Method
Loss function
21. Contents
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 21
Boah Kim
1. Introduction
2. Method
3. Experiment
4. Conclusion
22. Experiment
Head CT & CECT scans
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 22
Boah Kim
• Provided by Gacheon University College of Medicine
• 10 pair of CT & CECT scans (not aligned)
• 8 scans for training / 2 scans for test
CECT
CT
Dataset
23. Experiment
Head CT & CECT scans
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 23
Boah Kim
• Provided by Gacheon University College of Medicine
• 10 pair of CT & CECT scans (not aligned)
• 8 scans for training / 2 scans for test
CECT
CT
Dataset
Implementation details
• Batch size = 4
• Learning rate = 0.0002 (decayed to zero until last epoch)
• Training for 200 epoch using a single GPU NVIDIA Geforce GTX 1080 Ti
• Data augmentation : 256x256 random crop, random horizontal/vertical flip, rotation
24. Synthetic CT image results from CECT
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 24
Boah Kim
Input: CECT
[-300,600] HU
Target: CT cycleGAN Ours
Input - Target Input - cycleGAN Input - Ours
[-300,300] HU
Experiment
Results
25. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 25
Boah Kim
Synthetic CT image results from CECT
Input: CECT
[-300,600] HU
Target: CT cycleGAN Ours
Input - Target Input - cycleGAN Input - Ours
[-300,300] HU
Experiment
Results
26. Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 26
Boah Kim
Synthetic CT image results from CECT
Input: CECT
[-300,600] HU
Target: CT cycleGAN Ours
Input - Target Input - cycleGAN Input - Ours
[-300,300] HU
Experiment
Results
27. Contents
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 27
Boah Kim
1. Introduction
2. Method
3. Experiment
4. Conclusion
28. Conclusion
Present a medical image synthesis method with improved cycleGAN
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 28
Boah Kim
Validate our model by synthesizing CT image from CECT image using real dataset
Can be an important platform for medical image synthesis
29. Acknowledgement
• This work was supported by Institute of Information & Communications Technology Planning & Evaluation
(IITP) grant funded by the Korea government(MSIT) [2016-0-00562(R0124-16-0002), Emotional
Intelligence Technology to Infer Human Emotion and Carry on Dialogue Accordingly]
• This work was also 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).
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 29
Boah Kim
30. Workshop: Deep Learning for Biomedical Image Reconstruction
Medical Image Synthesis with Improved Cycle-GAN: CT from CECT
ISBI2020 Workshop Presentation 30
Boah Kim
Thank you.