A Novel Approach to High Resolution Fetal Brain MR Imaging

Abstract

This paper describes a novel approach to forming high resolution MR images of the human fetal brain. It addresses the key problem of motion of the fetus by proposing a registration refined compounding of multiple sets of orthogonal fast 2D MRI slices, that are currently acquired for clinical studies, into a single high resolution MRI volume. A robust multi-resolution slice alignment is applied iteratively to the data to correct motion of the fetus that occurs between 2D acquisitions. This is combined with an intensity correction step and a super resolution reconstruction step, to form a single high isotropic resolution volume of the fetal brain. Experimental validation on synthetic image data with known motion types and underlying anatomy, together with retrospective application to sets of clinical acquisitions are included. Results indicate the method promises a unique route to acquiring high resolution MRI of the fetal brain in vivo allowing comparable quality to that of neonatal MRI. Such data is highly valuable in allowing a clinically applicable window into the process of normal and abnormal brain development.

Automatic Detection and Measurement of Structures in Fetal Head Ultrasound Volumes Using Sequential Estimation and Integrated Detection Network (IDN)

“…The segmentation of the skull bone content is performed by graph cuts, first in a midsagittal plane and then in 3-D. MRI fetal scanning is expensive and not approved for fetuses of gestational age below 20 weeks. Due to motion of the fetus during the scan, the motion correction must be applied [39]. Finally, more constrained scanning procedure and different imaging characteristics (higher signal-to-noise ratio, no signal drop outs, and no shadowing artifacts) make these techniques difficult to extend to the ultrasound domain.…”

Volume Reconstruction by Inverse Interpolation: Application to Interleaved MR Motion Correction

“…The advantage of our method over volume injection techniques [1,4] is more subtle, but still visually apparent (Fig. 4).…”

“…Image reconstruction from multiple co-registered views can be solved by push-forward interpolation and averaging [4]: Each pixel y is injected into the reconstructed image volume at all locations x within a radius d, weighted using a truncated Gaussian-shaped kernel,…”

“…This also includes data acquired on multiple regular grids, e.g., multiple images with general coordinate transformations between them. Such image reconstruction problems arise, for example, in freehand threedimensional (3D) ultrasound acquisition [1] or in the reconstruction of high-resolution motion-corrected magnetic resonance (MR) images [4].…”

Guidelines for Developing Automated Quality Control Procedures for Brain Magnetic Resonance Images Acquired in Multi-Centre Clinical Trials

“…These images would be expected to yield incorrect results for many image processing algorithms. While there exists algorithms to correct some of these artifacts (Ahmed et al, 2003;Blumenthal et al, 2002b;Forbes et al, 2001;Greenspan et al, 2001;Kim et al, 1999;Lötjönen et al, 2004;Malandain et al, 2004;Ourselin et al, 2000;Rousseau et al, 2005;Sled et al, 1998;Ward et al, 2000), these corrections may not be adequate to achieve the expected precision of the downstream image processing to quantify brain pathology metrics. Longitudinal inconsistencies tend to result from scanner software and hardware upgrades, scanner hardware deterioration and human errors/ inconsistencies.…”