Robust EMI elimination for RF shielding‐free MRI through deep learning direct MR signal prediction

Abstract

PurposeTo develop a new electromagnetic interference (EMI) elimination strategy for RF shielding‐free MRI via active EMI sensing and deep learning direct MR signal prediction (Deep‐DSP).MethodsDeep‐DSP is proposed to directly predict EMI‐free MR signals. During scanning, MRI receive coil and EMI sensing coils simultaneously sample data within two windows (i.e., for MR data and EMI characterization data acquisition, respectively). Afterward, a residual U‐Net model is trained using synthetic MRI receive coil data and EMI sensing coil data acquired during EMI signal characterization window, to predict EMI‐free MR signals from signals acquired by MRI receive and EMI sensing coils. The trained model is then used to directly predict EMI‐free MR signals from data acquired by MRI receive and sensing coils during the MR signal‐acquisition window. This strategy was evaluated on an ultralow‐field 0.055T brain MRI scanner without any RF shielding and a 1.5T whole‐body scanner with incomplete RF shielding.ResultsDeep‐DSP accurately predicted EMI‐free MR signals in presence of strong EMI. It outperformed recently developed EDITER and convolutional neural network methods, yielding better EMI elimination and enabling use of few EMI sensing coils. Furthermore, it could work well without dedicated EMI characterization data.ConclusionDeep‐DSP presents an effective EMI elimination strategy that outperforms existing methods, advancing toward truly portable and patient‐friendly MRI. It exploits electromagnetic coupling between MRI receive and EMI sensing coils as well as typical MR signal characteristics. Despite its deep learning nature, Deep‐DSP framework is computationally simple and efficient.

Whole-body magnetic resonance imaging at 0.05 Tesla

“…The trained model was then used to predict EMI-free MR signal from data acquired during the MR signal acquisition window. This Deep-DSP strategy has been shown to yield superior performance ( 25 ) compared with all existing EMI reduction methods recently developed for brain ULF MRI ( 12 , 17 , 18 ). It is worth noting that, in practice, the EMI signal characterization window is not an absolute requirement for Deep-DSP ( 25 ).…”

“…1A). We developed and implemented a method termed deep learning direct signal prediction (Deep-DSP) ( 25 ) (Fig. 1C and fig.…”

“…As a low-cost, point-of-care, and patient-friendly device, whole-body ULF MRI should operate without any enclosed RF shielding. In this study, we have successfully developed and deployed the Deep-DSP approach ( 25 ), which directly predicts EMI-free MR signals even in the presence of very strong EMI signals from external environments and internal electronics. The Deep-DSP strategy ( 25 ) functions with or without dedicated EMI characterization data, considerably outperforming all existing analytical or deep learning methods that have been recently developed by our group ( 12 , 18 ) and others ( 17 ).…”

“…We utilized a deep learning method, Deep-DSP, developed by our group for mobile brain MRI scanners ( 25 ). Ten small EMI sensing coils (LC-resonant loops with 5 cm diameter) were placed near the patient bed and magnet and inside the electronic cabinet close to the gradient amplifier and console.…”

“…It features a homogeneous 0.05 T permanent magnet and linear imaging gradients, enabling us to implement ULF MRI protocols by building upon the methodologies developed for high-field MRI over the past five decades. To achieve robust EMI elimination for shielding-free scanning, we deployed a method to directly predict EMI-free MR signals through deep learning ( 25 ). We demonstrated the wide-ranging applicability of this scanner for imaging various anatomical structures, including the brain, spine, abdomen, heart, lungs, and extremities.…”