FR-Net: Joint Reconstruction and Segmentation in Compressed Sensing Cardiac MRI

We provide a novel solution to the inverse problem in medical imaging that takes as input the undersampled k-space data from Magnetic Resonance Imaging (MRI) scans and outputs both the reconstructed images and the segmented myocardium. Previously, the undersampled k-space data is first transformed into a reconstructed MRI image. From this image, the myocardium is contours are subsequently extracted using a segmentation method. However, this sequential approach is not optimal and requires manual intervention. In order to automate and improve the results of these approaches, we propose a new method to solve the reconstruction and segmentation problems simultaneously. Our method is based on a novel deep learning approach we term “Joint-FR-Net”, which consists of a reconstruction module derived from the fast iterative shrinkage-thresholding algorithm (FISTA) and a segmentation module. We test our approach on an undersampled short-axis (SAX) cardiac dataset and show the effectiveness of the Joint FR-Net in both image reconstruction and myocardium joint segmentation.

[1]  Jong Chul Ye,et al.  Improved k–t BLAST and k–t SENSE using FOCUSS , 2007, Physics in medicine and biology.

[2]  D. Donoho,et al.  Sparse MRI: The application of compressed sensing for rapid MR imaging , 2007, Magnetic resonance in medicine.

[3]  Shiqian Ma,et al.  An efficient algorithm for compressed MR imaging using total variation and wavelets , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[4]  M. Lustig,et al.  Compressed Sensing MRI , 2008, IEEE Signal Processing Magazine.

[5]  Marc Teboulle,et al.  A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems , 2009, SIAM J. Imaging Sci..

[6]  Yoram Bresler,et al.  MR Image Reconstruction From Highly Undersampled k-Space Data by Dictionary Learning , 2011, IEEE Transactions on Medical Imaging.

[7]  Daniel Rueckert,et al.  Application-Driven MRI: Joint Reconstruction and Segmentation from Undersampled MRI Data , 2014, MICCAI.

[8]  Jian Sun,et al.  Deep ADMM-Net for Compressive Sensing MRI , 2016, NIPS.

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

[10]  Jong Chul Ye,et al.  Deep residual learning for compressed sensing MRI , 2017, 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017).

[11]  Dimitris N. Metaxas,et al.  Multi-component deformable models coupled with 2D-3D U-Net for automated probabilistic segmentation of cardiac walls and blood , 2018, 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018).

[12]  Daniel C. Castro,et al.  Cardiac MR Segmentation from Undersampled k-space Using Deep Latent Representation Learning , 2018, MICCAI.

[13]  Jin Keun Seo,et al.  Deep learning for undersampled MRI reconstruction , 2017, Physics in medicine and biology.

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

[15]  Dimitris N. Metaxas,et al.  MRI Reconstruction Via Cascaded Channel-Wise Attention Network , 2018, 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019).