Learning the implicit strain reconstruction in ultrasound elastography using privileged information

Quasi-static ultrasound elastography is an importance imaging technology to assess the conditions of various diseases through reconstructing the tissue strain from radio frequency data. State-of-the-art strain reconstruction techniques suffer from the inexperienced user unfriendliness, high model bias, and low effectiveness-to-efficiency ratio. The three challenges result from the explicitness characteristic (i.e. explicit formulation of the reconstruction model) in these techniques. For these challenges, we are the first to develop an implicit strain reconstruction framework by a deep neural network architecture. However, the classic neural network methods are unsuitable to the strain reconstruction task because they are difficult to impose any direct influence on the intermediate state of the learning process. This may lead the map learned by the neural network to be biased with the desired map. In order to correct the intermediate state of the learning process, our framework proposes the learning-using-privileged-information (LUPI) paradigm with causality in the network. It provides the causal privileged information besides the training examples to help the network learning, while makes these privileged information unavailable at the test stage. This improvement can narrow the search region of the map learned by the network, and thus prompts the network to evolve towards the actual ultrasound elastography process. Moreover, in order to ensure the causality in LUPI, our framework proposes a physically-based data generation strategy to produce the triplets of privileged information, training examples and labels. This data generation process can approximately describes the actual ultrasound elastography process by the numerical simulation based on the tissue biomechanics and ultrasound physics. It thus can build the causal relationship between the privileged information and training examples/labels. It can also address the medical data insufficiency problem. The performance of our framework has been validated on 100 simulation data, 42 phantom data and 4 real clinical data by comparing with the ground truth performed by an ultrasound simulation system and four state-of-the-art methods. The experimental results show that our framework is well agreed (average bias is 0.065 for strain reconstruction) with the ground truth, as well as superior to these state-of-the-art methods. These results can demonstrate the effectiveness of our framework in the strain reconstruction.

[1]  Fatma Taher,et al.  A deep learning-based approach for automatic segmentation and quantification of the left ventricle from cardiac cine MR images , 2020, Comput. Medical Imaging Graph..

[2]  Tsuyoshi Shiina,et al.  WFUMB guidelines and recommendations for clinical use of ultrasound elastography: Part 3: liver. , 2015, Ultrasound in medicine & biology.

[3]  E. M. A. Anas,et al.  Direct mean strain estimation for elastography using nearest-neighbor weighted least-squares approach in the frequency domain. , 2012, Ultrasound in medicine & biology.

[4]  Silvio Savarese,et al.  Deep Learning Under Privileged Information Using Heteroscedastic Dropout , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[5]  Hassan Rivaz,et al.  Global Time-Delay Estimation in Ultrasound Elastography , 2017, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.

[6]  Heye Zhang,et al.  A stochastic filtering approach to recover strain images from quasi-static ultrasound elastography , 2014, Biomedical engineering online.

[7]  Nir Friedman,et al.  Probabilistic Graphical Models - Principles and Techniques , 2009 .

[8]  Jianwen Luo,et al.  Performance comparison of rigid and affine models for motion estimation using ultrasound radio-frequency signals , 2015, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.

[9]  Andrew Dibattista,et al.  An efficient block matching and spectral shift estimation algorithm with applications to ultrasound elastography , 2014, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.

[10]  F. Schaefer,et al.  EFSUMB Guidelines and Recommendations on the Clinical Use of Ultrasound Elastography. Part 1: Basic Principles and Technology , 2013, Ultraschall in der Medizin.

[11]  Huafeng Liu,et al.  Direct delineation of myocardial infarction without contrast agents using a joint motion feature learning architecture , 2018, Medical Image Anal..

[12]  Guang Yang,et al.  Holistic and Deep Feature Pyramids for Saliency Detection , 2018, BMVC.

[13]  Anna Choromanska,et al.  VisualBackProp for learning using privileged information with CNNs , 2018, ArXiv.

[14]  Samy Bengio,et al.  Understanding deep learning requires rethinking generalization , 2016, ICLR.

[15]  Jan Feyereisl,et al.  Object Localization based on Structural SVM using Privileged Information , 2014, NIPS.

[16]  Gregory D. Hager,et al.  Tracked Regularized Ultrasound Elastography for Targeting Breast Radiotherapy , 2009, MICCAI.

[17]  Shan Gao,et al.  Gated recurrent unit-based heart sound analysis for heart failure screening , 2020, BioMedical Engineering OnLine.

[18]  Michael I. Jordan,et al.  Graphical Models, Exponential Families, and Variational Inference , 2008, Found. Trends Mach. Learn..

[19]  Dong Liang,et al.  Motion Tracking of the Carotid Artery Wall From Ultrasound Image Sequences: a Nonlinear State-Space Approach , 2018, IEEE Transactions on Medical Imaging.

[20]  Yu Zhang,et al.  Learning to Transfer , 2017, ArXiv.

[21]  J. Jensen,et al.  Calculation of pressure fields from arbitrarily shaped, apodized, and excited ultrasound transducers , 1992, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.

[22]  Ryota Tomioka,et al.  Norm-Based Capacity Control in Neural Networks , 2015, COLT.

[23]  Huafeng Liu,et al.  A strain-based ultrasound elastography using phase shift with prior estimates and meshfree shape function , 2011, 2011 IEEE International Symposium on Biomedical Imaging: From Nano to Macro.

[24]  Andreas K. Maier,et al.  Deep Learning Computed Tomography , 2016, MICCAI.

[25]  Guillermo Sapiro,et al.  Lessons from the Rademacher complexity for deep learning , 2016, ICLR 2016.

[26]  Urszula Zaleska-Dorobisz,et al.  Ultrasound elastography - review of techniques and its clinical applications. , 2014, Advances in clinical and experimental medicine : official organ Wroclaw Medical University.

[27]  Daniel Rueckert,et al.  A Deep Cascade of Convolutional Neural Networks for MR Image Reconstruction , 2017, IPMI.

[28]  Hang Li Learning to Rank , 2017, Encyclopedia of Machine Learning and Data Mining.

[29]  Daniel Rueckert,et al.  A Deep Cascade of Convolutional Neural Networks for Dynamic MR Image Reconstruction , 2017, IEEE Transactions on Medical Imaging.

[30]  J. P. Lewis Fast Normalized Cross-Correlation , 2010 .

[31]  Adam Roberts,et al.  Latent Constraints: Learning to Generate Conditionally from Unconditional Generative Models , 2017, ICLR.

[32]  Tsuyoshi Shiina,et al.  WFUMB guidelines and recommendations for clinical use of ultrasound elastography: Part 1: basic principles and terminology. , 2015, Ultrasound in medicine & biology.

[33]  Ambuj Tewari,et al.  On the Complexity of Linear Prediction: Risk Bounds, Margin Bounds, and Regularization , 2008, NIPS.

[34]  Ying Wang,et al.  Ultrasound characteristics of cervical lesions in patients with radioiodine refractory differentiated thyroid cancer , 2019, Medicine.

[35]  Vladimir Vapnik,et al.  Statistical learning theory , 1998 .

[36]  M. Omair Ahmad,et al.  Supervised Classification of the Accuracy of the Time Delay Estimation in Ultrasound Elastography , 2018, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.

[37]  Yanning Zhu,et al.  A Modified Block Matching Method for Real-Time Freehand Strain Imaging , 2002, Ultrasonic imaging.

[38]  Heye Zhang,et al.  Robust estimation of carotid artery wall motion using the elasticity‐based state‐space approach , 2017, Medical Image Anal..

[39]  Tsuyoshi Shiina,et al.  WFUMB guidelines and recommendations for clinical use of ultrasound elastography: Part 2: breast. , 2015, Ultrasound in medicine & biology.

[40]  Che-Rung Lee,et al.  Escaping from Collapsing Modes in a Constrained Space , 2018, ECCV.

[41]  Michèle Sebag,et al.  Learning Functional Causal Models with Generative Neural Networks , 2018 .

[42]  M. Omidyeganeh,et al.  Estimation of Strain Elastography from Ultrasound Radio-Frequency Data by Utilizing Analytic Gradient of the Similarity Metric , 2017, IEEE Transactions on Medical Imaging.

[43]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[44]  Olivier Basset,et al.  2-D Locally Regularized Tissue Strain Estimation From Radio-Frequency Ultrasound Images: Theoretical Developments and Results on Experimental Data , 2008, IEEE Transactions on Medical Imaging.

[45]  Bernt Schiele,et al.  Learning using privileged information: SV M+ and weighted SVM , 2013, Neural Networks.

[46]  François Chollet,et al.  Xception: Deep Learning with Depthwise Separable Convolutions , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[47]  Defeng Wang,et al.  Robust recovery of myocardial kinematics using dual ℋ∞$\mathcal {H}_{\infty }$ criteria , 2017, Multimedia Tools and Applications.

[48]  Kim-Han Thung,et al.  A brief review on multi-task learning , 2018, Multimedia Tools and Applications.

[49]  Shihui Ying,et al.  Improving Single-Modal Neuroimaging Based Diagnosis of Brain Disorders via Boosted Privileged Information Learning Framework , 2016, MLMI@MICCAI.

[50]  R. Righetti,et al.  Trade-offs between the axial resolution and the signal-to-noise ratio in elastography. , 2003 .

[51]  Feng Lin,et al.  Low-Dose CT With a Residual Encoder-Decoder Convolutional Neural Network , 2017, IEEE Transactions on Medical Imaging.

[52]  Yuan-Ting Zhang,et al.  Motion Estimation of Common Carotid Artery Wall Using a H ∞ Filter Based Block Matching Method , 2015, MICCAI.

[53]  Gregory D. Hager,et al.  Ultrasound elastography using multiple images , 2014, Medical Image Anal..

[54]  Jahan Tavakkoli,et al.  Performance study of a new time-delay estimation algorithm in ultrasonic echo signals and ultrasound elastography. , 2016, Ultrasonics.

[55]  Septimiu E. Salcudean,et al.  Motion Estimation in Ultrasound Images Using Time Domain Cross Correlation With Prior Estimates , 2006, IEEE Transactions on Biomedical Engineering.

[56]  J M Bland,et al.  Statistical methods for assessing agreement between two methods of clinical measurement , 1986 .

[57]  Vladimir Vapnik,et al.  A new learning paradigm: Learning using privileged information , 2009, Neural Networks.

[58]  Roch L. Maurice,et al.  Optical-Flow-Based B-Mode Elastography: Application in the Hypertensive Rat Carotid , 2010, IEEE Transactions on Medical Imaging.

[59]  Yu Zhang,et al.  A Survey on Multi-Task Learning , 2017, IEEE Transactions on Knowledge and Data Engineering.

[60]  Jingfeng Jiang,et al.  A coupled subsample displacement estimation method for ultrasound-based strain elastography , 2015, Physics in medicine and biology.

[61]  Brian W. Anthony,et al.  Multi-frame elastography using a handheld force-controlled ultrasound probe , 2015, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.

[62]  Yaoqin Xie,et al.  Carotid Artery Wall Motion Estimated from Ultrasound Imaging Sequences Using a Nonlinear State Space Approach , 2016, MICCAI.

[63]  Qi Zhang,et al.  B-Mode Ultrasound Based Diagnosis of Liver Cancer With CEUS Images as Privileged Information , 2018, 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC).

[64]  O H Gilja,et al.  EFSUMB Guidelines and Recommendations on the Clinical Use of Ultrasound Elastography.Part 2: Clinical Applications , 2013, Ultraschall in der Medizin.

[65]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[66]  Andrei Zinovyev,et al.  Independent Component Analysis for Unraveling the Complexity of Cancer Omics Datasets , 2019, International journal of molecular sciences.

[67]  Wei Li,et al.  The value of ultrasound elastography in differential diagnosis of superficial lymph nodes , 2009 .

[68]  Tae-Kyun Kim,et al.  Learning and Refining of Privileged Information-Based RNNs for Action Recognition from Depth Sequences , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[69]  Trevor Darrell,et al.  Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation , 2013, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[70]  Xiabi Liu,et al.  A multi-level similarity measure for the retrieval of the common CT imaging signs of lung diseases , 2020, Medical & Biological Engineering & Computing.

[71]  Jianwen Luo,et al.  A fast normalized cross-correlation calculation method for motion estimation , 2010, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.

[72]  Jianwen Luo,et al.  Robust Segmentation of Intima–Media Borders With Different Morphologies and Dynamics During the Cardiac Cycle , 2018, IEEE Journal of Biomedical and Health Informatics.

[73]  Hassan Rivaz,et al.  GLUENet: Ultrasound Elastography Using Convolutional Neural Network , 2018, POCUS/BIVPCS/CuRIOUS/CPM@MICCAI.

[74]  V. Vapnik,et al.  On the theory of learning with Privileged Information , 2010, NIPS 2010.

[75]  Frédérique Frouin,et al.  Ultrasound elastography based on multiscale estimations of regularized displacement fields , 2004, IEEE Transactions on Medical Imaging.

[76]  Bo Peng,et al.  Cascaded Multi-Column RVFL+ Classifier for Single-Modal Neuroimaging-Based Diagnosis of Parkinson's Disease , 2019, IEEE Transactions on Biomedical Engineering.

[77]  Leon A. Gatys,et al.  Texture Synthesis Using Convolutional Neural Networks , 2015, NIPS.

[78]  David Cosgrove,et al.  WFUMB Guidelines and Recommendations on the Clinical Use of Ultrasound Elastography: Part 5. Prostate. , 2017, Ultrasound in medicine & biology.

[79]  Yoshua Bengio,et al.  Learning deep representations by mutual information estimation and maximization , 2018, ICLR.

[80]  Rauf Izmailov,et al.  Learning using privileged information: similarity control and knowledge transfer , 2015, J. Mach. Learn. Res..

[81]  W. Walker,et al.  A fundamental limit on delay estimation using partially correlated speckle signals , 1995, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.

[82]  Sylvia Richardson,et al.  Markov Chain Monte Carlo in Practice , 1997 .

[83]  Ohad Shamir,et al.  Size-Independent Sample Complexity of Neural Networks , 2017, COLT.

[84]  Tomy Varghese,et al.  Wavelet denoising of displacement estimates in elastography. , 2004, Ultrasound in medicine & biology.

[85]  Jianwen Luo,et al.  Direct Reconstruction of Ultrasound Elastography Using an End-to-End Deep Neural Network , 2018, MICCAI.

[86]  Christoph H. Lampert,et al.  Learning to Rank Using Privileged Information , 2013, 2013 IEEE International Conference on Computer Vision.

[87]  Zhuowen Tu,et al.  Deeply-Supervised Nets , 2014, AISTATS.

[88]  Peter L. Bartlett,et al.  Rademacher and Gaussian Complexities: Risk Bounds and Structural Results , 2003, J. Mach. Learn. Res..

[89]  Gregory D. Hager,et al.  Real-Time Regularized Ultrasound Elastography , 2011, IEEE Transactions on Medical Imaging.

[90]  Graham M. Treece,et al.  Real-time quasi-static ultrasound elastography , 2011, Interface Focus.

[91]  Bernhard Schölkopf,et al.  On Estimation of Functional Causal Models , 2015, ACM Trans. Intell. Syst. Technol..

[92]  José Eduardo Corrente,et al.  Statistical methods for assessing agreement , 2011 .

[93]  Lin Chen,et al.  Incorporating Privileged Genetic Information for Fundus Image Based Glaucoma Detection , 2014, MICCAI.

[94]  D. Altman,et al.  STATISTICAL METHODS FOR ASSESSING AGREEMENT BETWEEN TWO METHODS OF CLINICAL MEASUREMENT , 1986, The Lancet.

[95]  K. Boone,et al.  Effect of skin impedance on image quality and variability in electrical impedance tomography: a model study , 1996, Medical and Biological Engineering and Computing.

[96]  Christoph H. Lampert,et al.  Learning to Transfer Privileged Information , 2014, ArXiv.

[97]  M. Insana,et al.  An information-based machine learning approach to elasticity imaging , 2017, Biomechanics and modeling in mechanobiology.

[98]  P. Fearnhead Markov chain Monte Carlo, Sufficient Statistics, and Particle Filters , 2002 .

[99]  J. Ophir,et al.  Elastography: A Quantitative Method for Imaging the Elasticity of Biological Tissues , 1991, Ultrasonic imaging.

[100]  Michael S. Lew,et al.  Deep learning for visual understanding: A review , 2016, Neurocomputing.

[101]  Lili Yuan,et al.  Analytical phase-tracking-based strain estimation for ultrasound elasticity , 2015, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control.