Zero-Shot Self-Supervised Learning of Single Breath-Hold Magnetic Resonance Cholangiopancreatography (MRCP) Reconstruction.

Authors

Abstract

To investigate the feasibility of zero-shot self-supervised learning reconstruction for reducing breath-hold times in magnetic resonance cholangiopancreatography (MRCP). Breath-hold MRCP was acquired from 11 healthy volunteers on 3T scanners using an incoherent k-space sampling pattern, leading to a 14-s acquisition time and an acceleration factor of R = 25. Zero-shot reconstruction was compared with parallel imaging of respiratory-triggered MRCP (338 s, R = 3) and compressed sensing reconstruction. For two volunteers, breath-hold scans (40 s, R = 6) were additionally acquired and retrospectively undersampled to R = 25 to compute peak signal-to-noise ratio (PSNR). To address long zero-shot training time, the <math xmlns="http://www.w3.org/1998/Math/MathML"> <semantics><mrow><mi>n</mi> <mo>+</mo> <mi>m</mi></mrow> <annotation>$$ n+m $$</annotation></semantics> </math> full stages of the zero-shot learning were divided into two parts to reduce backpropagation depth during training: (1) <math xmlns="http://www.w3.org/1998/Math/MathML"> <semantics><mrow><mi>n</mi></mrow> <annotation>$$ n $$</annotation></semantics> </math> frozen stages initialized with <math xmlns="http://www.w3.org/1998/Math/MathML"> <semantics><mrow><mi>n</mi></mrow> <annotation>$$ n $$</annotation></semantics> </math> -stage pretrained network and (2) <math xmlns="http://www.w3.org/1998/Math/MathML"> <semantics><mrow><mi>m</mi></mrow> <annotation>$$ m $$</annotation></semantics> </math> trainable stages initialized either randomly or <math xmlns="http://www.w3.org/1998/Math/MathML"> <semantics><mrow><mi>m</mi></mrow> <annotation>$$ m $$</annotation></semantics> </math> -stage pretrained network. Efficiency of our approach was assessed by varying initialization strategies and the number of trainable stages using the retrospectively undersampled data. Zero-shot reconstruction significantly improved visual image quality over compressed sensing, particularly in SNR and ductal delineation, and achieved image quality comparable to that of successful respiratory-triggered acquisitions with regular breathing patterns. Improved initializations enhanced PSNR and reduced reconstruction time. Adjusting frozen/trainable configurations demonstrated that PSNR decreased only slightly from 38.25 dB (0/13) to 37.67 dB (12/1), while training time decreased up to 6.7-fold. Zero-shot learning delivers high-fidelity MRCP reconstructions with reduced breath-hold times, and the proposed partially trainable approach offers a practical solution for translation into time-constrained clinical workflows.

Tags

Topics