Deep learning–guided weighted averaging for signal dropout compensation in DWI of the liver

Magnetic Resonance in Medicine - Tập 88 Số 6 - Trang 2679-2693 - 2022
Fasil Gadjimuradov1,2, Thomas Benkert1, Dominik Nickel1, Tobit Führes3, Marc Saake3, Andreas Maier2
1MR Application Predevelopment, Siemens Healthcare GmbH, Erlangen, Germany
2Pattern Recognition Lab, Department of Computer Science Friedrich‐Alexander‐Universität Erlangen‐Nürnberg Erlangen Germany
3Institute of Radiology, University Hospital Erlangen Friedrich‐Alexander‐Universität Erlangen‐Nürnberg Erlangen Germany

Tóm tắt

PurposeTo develop an algorithm for the retrospective correction of signal dropout artifacts in abdominal DWI resulting from cardiac motion.MethodsGiven a set of image repetitions for a slice, a locally adaptive weighted averaging is proposed that aims to suppress the contribution of image regions affected by signal dropouts. Corresponding weight maps were estimated by a sliding‐window algorithm, which analyzed signal deviations from a patch‐wise reference. In order to ensure the computation of a robust reference, repetitions were filtered by a classifier that was trained to detect images corrupted by signal dropouts. The proposed method, named Deep Learning–guided Adaptive Weighted Averaging (DLAWA), was evaluated in terms of dropout suppression capability, bias reduction in the ADC, and noise characteristics.ResultsIn the case of uniform averaging, motion‐related dropouts caused signal attenuation and ADC overestimation in parts of the liver, with the left lobe being affected particularly. Both effects could be substantially mitigated by DLAWA while preventing global penalties with respect to SNR due to local signal suppression. Performing evaluations on patient data, the capability to recover lesions concealed by signal dropouts was demonstrated as well. Further, DLAWA allowed for transparent control of the trade‐off between SNR and signal dropout suppression by means of a few hyperparameters.ConclusionThis work presents an effective and flexible method for the local compensation of signal dropouts resulting from motion and pulsation. Because DLAWA follows a retrospective approach, no changes to the acquisition are required.

Từ khóa


Tài liệu tham khảo

10.1593/neo.81328

10.1148/radiol.09090021

10.1002/jmri.23947

10.1016/j.mric.2014.04.009

10.1002/mp.15246

10.2214/AJR.06.1403

10.1007/s10334-009-0183-1

10.1148/radiol.2241011117

10.1002/jmri.23796

10.1002/mrm.26166

10.1002/mrm.27462

10.1002/mrm.27735

VanAT CervantesB OginoT et al.Partial velocity‐compensated diffusion encoding for combined motion compensation and residual vessel signal suppression in liver DWI. In Proceedings of the 26th Annual Meeting of ISMRM Paris France 2018.

10.1002/jmri.22816

10.2463/mrms.mp.2018-0085

10.1038/nature14539

10.1016/0893-6080(91)90009-T

Krizhevsky A, 2012, Imagenet classification with deep convolutional neural networks. In: Pereira F, Burges CJ, Bottou, Weinberger KQ, eds. Advances in Neural Information Processing Systems (NeurIPS 2012), vol 25

TamadaD MotosugiU OnishiH.Improving the image quality of liver DWI using the convolutional neural network‐based selection algorithm. In Proceedings of the 26th Annual Meeting of ISMRM Paris France 2018;2723.

GadjimuradovF BenkertT NickelMD MaierA.Deep learning‐based adaptive image combination for signal‐dropout suppression in liver DWI. In Proceedings of the 29th Annual Meeting of the ISMRM Virtual Conference 2021.

IoffeS SzegedyC.Batch normalization: accelerating deep network training by reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine Learning (PMLR) 2015;37:448‐456.

ZaheerM KotturS RavanbhakhshS PóczosB SalakhutdinovR SmolaAJ.Deep sets. In Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS) 2017.

10.1002/cmr.a.21223

10.1002/mrm.10171

Paszke A, 2019, Advances in Neural Information Processing Systems (NeurIPS 2019)

KukarM KononenkoI.Cost‐sensitive learning with neural networks. In Proceedings of the 13th European Conference on Artificial Intelligence (ECAI‐98) 1998.

10.1002/jmri.24792

KimJ GadjimuradovF BenkertT VahleT MaierA.Deep learning‐based respiratory motion correction in free‐breathing abdominal diffusion‐weighted imaging. In Proceedings of the Joint Annual Meeting ISMRM‐ESMRMB & ISMRT 31st Annual Meeting London UK 2022.

10.1002/mrm.21127

10.1097/RLI.0b013e31822438e8

10.1002/nbm.3311

10.1002/mrm.21132