Julian Hossbach1,2,3, Daniel Nicolas Splitthoff2, Stephen Farman Cauley4, and Andreas Maier1
1Pattern Recognition Lab, Friedrich-Alexander-University Erlangen-Nuremberg, Erlangen, Germany, 2Siemens Healthcare GmbH, Erlangen, Germany, 3Erlangen Graduate School in Advanced Optical Technologies, Erlangen, Germany, 4Martinos Center for Biomedical Imaging, Charlestown, MA, United States
Synopsis
We exploit the data redundancy and the locality
of motion in k-space for an estimation of the motion parameters using a Deep
Learning approach. The exploratory Motion Parameter Estimation
DenseNet (MoPED) extracts the in-plane motion parameters between echo
trains of a TSE sequence. As input, the network receives the center patch of
the k-space from multiple coils; the network’s output can serve multiple
purposes. While an image rejection/reacquisition can be triggered by the motion
guess, we show that motion aware reconstruction can be accelerated using MoPED.
Introduction
Despite a wide range of proposed motion correction
methods1, patient movements during scanning remain as one of the
main sources of artifacts in MR imaging2. Detecting if3,4
and when5 motion occurred as well as its severity6 is
useful for rejecting or adapting the image reconstruction and shows the
potential of Machine- and especially Deep Learning (DL) for treating motion. However,
DL-based tomographic image reconstruction might be unstable7 and can
also lead to undesired results.
Combining a DL approach with model-based
reconstruction8 takes an additional step to ensures data consistency
and can avoid this pitfall. Building on motion detection5 and our previous work9, we
present a new exploratory DL network that regresses the in-plane motion
parameter for multiple echo trains (ET) from only the raw multi-coil k-space to
allow an informed rejection of misaligned ETs or an improved initialized optimization
of a motion-aware reconstruction. Method
Dataset generation:
Images and their sensitivity maps $$$\textbf{C}$$$ of 25
subjects and 4 phantoms (35 slices; transversal, sagittal and coronal) were
acquired with a TSE sequence (TR=6100 ms, TE=103 ms, R=2, FOV=220 mm, resolution
448x448; on various Siemens MAGNETOM scanners, Siemens Healthcare, Erlangen,
Germany at 1.5/3 T) and used for simulating in-plane motion (Translation Tx,y,l, Rotation Rz,l)
for a given ET l.
Therefore, a regular cartesian sampling trajectory $$$\textbf{S}_l$$$ with L=9 ETs (undersampling R = 2,
80% phase resolution, turbo factor = 20) was used to apply random integer
motion trajectories (between -6- and +6 pixels translation/degree rotation) to a
random and varying subset of ETs (between 0 and 6 ETs) of the Fourier
transformed image kGT (Eq. 1). Prior to the simulation, noise and
random flips were applied to the images for data augmentation.
$$\textbf{k}_{Mot}=∑_{l=0}^{L-1}S_lCT_{x,y,l}R_{z,l}k_{GT}\qquad\text{Eq. 1}$$
Processing pipeline:
The processing pipeline, depicted in Figure 1
a), utilizes the coil sensitivity maps to reduce the coil dimensionality of kMot by removing coils with less than the median mean absolute value and selecting the 4 most distant coils w.r.t. their
sensitivity center of the remaining maps. The center is determined by the
maximum value of the projection onto the axes. The remaining corresponding
non-zero filled k-space is cropped to a 64x64x4 center patch and the complex
values are split into additional channels. Each ET samples at least 7 lines in
that crop and ensures that inter-ET motion is encoded within the input.
Neural Network and Training:
The Motion Parameter Estimation
DenseNet (MoPED, see Figure 1b)) applies a DenseNet for feature
extraction and finally maps them into the 3 motion parameters per ET. The multitask
mapping of the feature vector consists of 2 fully connected layers separately for
each ET for smoothing the training. Thus, each task is to learn the 3
parameters of one ET, which are then combined in a Lx3 (i.e. 27 total
parameters) matrix for the loss function. The first fully connected layer before
the multitask layers is added to avoid that mutual processing is learned redundantly
for each ET.
The loss function is split into a) detection
loss $$$\mathcal{L}_{detect}$$$ (Eq.2) being the sigmoid cross entropy between the thresholded network
guess $$$\textbf{θ}_{Net} = \{\textbf{T}_x, \textbf{T}_y, \textbf{R}_z\}$$$ and ground truth motion $$$\overline{\textbf{θ}}_{GT}$$$ and b) $$$\mathcal{L}$$$mot, which calculates the L1
loss for all non-zero motion parameters. $$$\mathcal{L}$$$detect is weighted with
λ=0.1.
$$\mathcal{L}_{detect}=D(\textbf{θ}_{GT} )(-\log(sig(D(\textbf{θ}_{Net}))))+(1-D(\textbf{θ}_{G}))(\log(-sig(D(\textbf{θ}_{Net}))))\qquad\text{Eq. 2}$$
$$D(x)=\begin{cases}0&|x|\geq0.4\\1&|x|<0.4\end{cases}$$
Following our previous work, the input is standardized
followed by a trainable batch normalization before the first layer. The Xavier initialized network was
trained employing the Adam optimizer.
Reconstruction:
The motion parameter estimation can trigger a threshold-based
rejection of the acquisition or can be used as initial guess to obtain an improved image using for example a target voxel SENSE reconstruction combined
with a motion model10.
We
compare the reconstruction results after 20 iterations with and without MoPED estimation.
Using the motion-aware
reconstruction technique NAMER8 and assuming a perfect image guess, we compare the convergence speed for
minimizing the residual motion estimation error against zero initialized
motion.Results
The loss curves during the training process are
depicted in Fig. 2. In Fig 3, boxplots of the absolute error of the motion
estimation over a validation data set are depicted, showing our desired focus
on non-zero motion parameters.
The images in Fig. 4 depict the reconstruction
results after 20 iterations with and without MoPED initialization.
As the joint optimization of the image and motion parameter is highly non-linear, convergence can be slow. In Fig. 4, the MoPED initialization helps to overcome the stagnating convergence of the optimization.
Almost 80 samples are processed within one
second, thus the MoPED allows a fast initial guess of the motion parameters (on a Nvidia Geforce GTX 1080 Ti).
The optimization of the residual parameters using the NAMER approach converged by an average of 18% faster. Discussion and conclusion
The results show that our approach can retrospectively
estimate motion solely from the reduced raw k-space as an initial guess without
no external hardware or changes on the sequence. Nonetheless, an increased
accuracy and focusing more on no/smaller motion is crucial and addressed, for
example using Oksuz et al5, in future work.
Regarding the convergence speed for a full
reconstruction a thorough evaluation has yet to be done.Acknowledgements
No acknowledgement found.References
1 Zaitsev M., Maclaren J., Herbst M. Motion
artifacts in MRI: A complex problem with many partial solutions. J Magn Reson
Imaging 2015; 42(4); 887-901
2 Andre J., Bresnahan B., Mossa-Basha M. et al. Toward Quantifying the Prevalence,
Severity, and Cost Associated With Patient Motion During Clinical MR
Examinations. J Am Coll Radiol 2016; 12(7); 689-95
3 Lorch B., Vaillant G., Baumgartner C. et al. Automated Detection of Motion Artefacts in MR Imaging Using Decision Forests. Journal of Medical Engineering 2017, 1-9
4 Arroyo-Camejo S., Odry B., Chen X. et al. Towards
Contrast-Independent Automated Motion Detection Using 2D Adversarial DenseNets.
Proc. ISMRM 2019 4868
5 Oksuz I., Clough J., Ruijsink B. et al. Detection
and Correction of Cardiac MR Motion Artefacts during Reconstruction from
K-space. Proc MICCAI 2019; 695-703
6 Hasiuk M., Pawar K., Zhong S. et al. Deep
learning based motion estimation from highly under-sampled EPI volumetric
navigators. Proc. ISMRM 2019 4448
7
Antun V., Renna F., Poon C. et al. On instabilities of deep learning in image
reconstruction - Does AI come at a cost? CoRR 2019
8 Haskell M., Cauley S., Bilgic B. et al. Network
Accelerated Motion Estimation and Reduction (NAMER): Convolutional neural
network guided retrospective motion correction using a separable motion model.
MRM 2019; 82(4)
9 Hossbach J., Splitthoff D., Haskell M. et al. Deep
Neural Networks for Motion Estimation in k-space: Applications and Design.
Proc. ISMRM 2019 4851
10 Haskell M., Cauley S., Wald L. TArgeted Motion
Estimation and Reduction (TAMER): Data Consistency Based Motion Mitigation for
MRI Using a Reduced Model Joint Optimization. TMI 2018; 37(5); 1253-65