Fig 1.
A. Our 1minute dual-echo Echo-Planar Imaging (DE-EPI) DWI pulse sequence corresponding to the acquisition of one slice, with timing and relative gradient and pulse amplitudes drawn approximately to scale. The diffusion (tetrahedral encoding) diffusion gradients, GD, are shown following slice selection with a 90° spectral spatial radiofrequency (RF) pulse. After the diffusion preparation period, Echo1 is acquired at TE1, followed by a 90° refocusing pulse and Echo2 at TE2. Note that the total acquisition time for one slice varies from 102–108ms depending on the FOV. B. Equivalent coil-sensitivity imprinted on Echo1 and Echo2 (here DWI) images allows one to extract and remove the contribution of receive coil and RF sensitivity, B0 susceptibility differences, T1, and proton density to the signal. Note that the resulting intensity-corrected images (cEcho1 and cEcho2) resemble that of an exponential ADC image with added T2 contrast. Data is acquired on a 9day old male infant with diffusion abnormality associated with hypoglycemia.
Table 1.
Patient demographics, clinical presentation, and specific neuropathology with reduced diffusion.
Fig 2.
Examples of diffusion echoes before (baseline) and after intensity correction.
A. Note that intensity variation (arrow) before correction is markedly improved after correction in this 7-day old female infant. B, C. Similar improvements are shown along the posterior fossae after intensity correction in a 4-month female infant (A) and a 17-month old male infant (B).
Fig 3.
Examples of improved high intensity artifact associated with susceptibility after intensity correction.
The columns represent baseline (Echo1 or Echo2); after correction (cEcho1 or cEcho2); and corresponding GRE images. A. High signal in the right temporal lobe cortex suggested possible pathology on Echo1 (shown here, arrow) and Echo2 (not shown) in this 6-year old girl but was considered an artifact from the temporal bone. Note improved more homogeneous brain signal on its intensity-corrected complement, cEcho1 (shown here) in both the temporal lobe and posterior brain regions. B. Similar improved susceptibility artifact is shown on cEcho1 compared to Echo1 (arrow) in a 6-year old boy. C. Susceptibility resulting from old hemosiderin is also mitigated on cEcho2 compared to Echo2 (arrow) in this 5-month old male infant with complicated birth history and remote intraventricular hemorrhage. D. Susceptibility artifact on Echo2 (arrow) related to subdural hemorrhage shown on GRE (short arrow) is improved after intensity correction on cEcho2 (shown here) in this 11 month old male infant with congenital heart disease and left cerebral infarction.
Table 2.
Mean and median (in brackets) ratings for reader assessment for Echoes 1 and 2, and their corrected variants.
Fig 4.
Stacked bar graph showing the grade value and frequency of Readers’ A and B assessment of lesion conspicuity and diagnostic confidence for Echo1 and Echo2 and their intensity corrected variants (cEcho1 and cEcho2) for patients that had reduced diffusivity.
Fig 5.
Examples of superior lesion conspicuity on Echo2 or its intensity corrected complement, cEcho2 compared to Echo1 or cEcho1.
A. Note improved conspicuity of cerebral contusions (arrows) in this 3-year old boy who fell from a 2-story building on Echo2 compared to Echo1. B. Reduced diffusivity from hypoxic-ischemic encephalopathy is more distinctly depicted on Echo2 (long arrow) compared to Echo1 in a 7-day old male infant who was cooled for neonatal encephalopathy. Note also focal reduced diffusivity of right cerebral peduncle (short arrow) is also better delineated on Echo2. C. Reduced diffusivity from hypoglycemia is more conspicuous on the cEcho2 image (arrows) compared to cEcho1 in a 9-day old male infant. D. Similar findings are also noted for this 8-week old male infant presenting with brain injury from non accidental trauma (arrows).
Fig 6.
Examples of false negatives, or lesions missed by either of the reviewers.
A. While both reviewers identified various lobar brain regions of reduced diffusivity affected by hypoglycemia in this 9-day old male infant, right hippocampal involvement was noted on Echo2 and cEcho2 (not shown) but not for Echo1 or cEcho1. In retrospect, subtle high signal in right hippocampus on Echo1 (arrow) is identified but less conspicuous compared to Echo2. B. Same patient as Fig 5B where focal lesion in right cerebral peduncle was overlooked by both reviewers, perhaps due to small size, but identified by Reviewer1 on both Echo2 (arrow) and cEcho2 (not shown). C. Focal lesion in the left pons was identified by both reviewers on cEcho2 (arrow) and Echo2 (not shown) and but missed on Echo1 and cEcho1 in this 9-week old infant with cardiac arrest. D. Focal embolic infarct in the left caudate of a 63-week old female infant was missed on Echo1 by both reviewers but detected on cEcho2 (arrow), as well as Echo2 (not shown) and cEcho1.
Fig 7.
Examples of false positives lesions.
The majority of these cases resulted from coil-induced brain signal variations in the posterior regions and/or susceptibility-related high signal present in baseline echoes (Echo1 and Echo2). A. Left temporal lobe cortical high signal (arrow) seen on multiple axial images on Echo1 raised possibility of pathology given other scattered lesions in the brain in this 16-month female with a biotin-thiamine responsive basal ganglia disease, a neurometabolic disorder. This was not associated with low signal on ADC and was considered to reflect combination of temporal bone artifact and T2-shine through from an underlying T2/FLAIR high intensity present in this child. Note high signal intensity associated with bilateral temporal bone susceptibility is markedly mitigated on its intensity-corrected complement, cEcho1. B. Asymmetric high signal was seen at baseline on Echo1 (arrow) and Echo2 (not shown) images, raising suspicion of pathology in this 7-day old infant cooled for neonatal encephalopathy. However, this represented erroneous intensity modulations and was without associated low ADC signal. cEcho1 showed improved signal variation in the posterior brain region. C. The reviewers initially thought the right occipital bone tumor (sarcoma metastasis) in this 16-year old girl showed reduced diffusivity on both Echo2 (arrow) and Echo1 (not shown), but this was found to represent a T2-shine through effect from tumor high water content. Note that cEcho2 better reflects the non-cellular nature of the tumor. D. High signal along the left posterior brain on Echo2 (arrow) and Echo1 (not shown) was suspected to represent either hemorrhage or acute parenchymal lesion in this 13-day old male infant with coagulopathy and liver disease and hemorrhagic collection elsewhere in the brain (short arrow). However, the left posterior brain high signal arose from inhomogeneous signal modulations and not true pathology, and was found to be corrected on both cEcho1 (not shown) and cEcho2 images.
Table 3.
A summary of false negative and positive cases stratified by the individual echo dataset.