Diagnostic Performance of Ultrafast Brain MRI for Evaluation of Abusive Head Trauma

Article Figures & Data

Download figure
Open in new tab
Download powerpoint
Fig 1.
A 4-month-old infant with suspected abusive head trauma found to have bilateral subdural collections identified on coronal T2 TSE (A); however, the right subdural collection was not prospectively identified on ultrafast coronal T2 HASTE (B).
Download figure
Open in new tab
Download powerpoint
Fig 2.
A 31-month-old child with a suspected abusive head trauma with a subdural hematoma (not shown) found to have subarachnoid hemorrhage in the sulci of the left superior frontal and parietal lobes on axial SWI (A), which was prospectively detected by only 1 reviewer on ultrafast axial EPI T2* (B).
Download figure
Open in new tab
Download powerpoint
Fig 3.
A 10-month-old child with suspected abusive head trauma found to have subtle parenchymal edema identified in the left parietal lobe on axial and coronal T2 TSE (A and B), which was not prospectively identified on ultrafast axial or coronal HASTE (C and D).

Table 1:

Ultrafast MRI brain protocols

Table 2:

stMRI brain protocols

Table 3:

Prevalence of imaging findings per patient on stMRI

Finding	Prevalence
Subdural collection	11/24 (46%)
Bilateral subdural collection	10/11 (44%)
Subarachnoid hemorrhage	8/24 (33%)
Intraparenchymal hemorrhage	7/24 (29%)
Intraventricular hemorrhage	1/24 (4%)
Epidural hemorrhage	3/24 (13%)
Cytotoxic edema	4/24 (17%)
Parenchymal laceration	0/24 (0%)
Vasogenic edema	2/24 (8%)
Herniation or midline shift	0/24 (0%)
Hydrocephalus	0/24 (0%)
Encephalomalacia	2/24 (8%)
Large subarachnoid spaces	5/24 (21%)
Total No. of patients with any abnormal finding	20/24 (83%)

Table 4:

Discrepancy rates for consensus ufMRI, nHCT, and combined versus stMRI

	Ultrafast vs stMRI	nHCT vs stMRI	Ultrafast + nHCT vs stMRI
Subdural collection	0/24 (0%)	0/24 (0%)	0/24 (0%)
Bilateral subdural collection	3/24 (13%)	1/24 (4%)	1/24 (4%)
Tentorial subdural hemorrhage	3/24 (13%)	3/24 (13%)	3/24 (13%)
Subdural membrane formation	0/24 (0%)	2/24 (8%)	0/24 (0%)
Subdural fluid-fluid level	2/24 (8%)	2/24 (8%)	2/24 (8%)
Subarachnoid hemorrhage	4/24 (17%)	4/24 (17%)	4/24 (17%)
Intraparenchymal hemorrhage	0/24 (0%)	6/24 (25%)^{^a}	0/24 (0%)
Intraventricular hemorrhage	0/24 (0%)	1/24 (4%)	0/24 (0%)
Epidural hemorrhage	0/24 (0%)	0/24 (0%)	0/24 (0%)
Cytotoxic edema	0/24 (0%)	4/24 (17%)	0/24 (0%)
Parenchymal laceration	0/24 (0%)	0/24 (0%)	0/24 (0%)
Vasogenic edema	0/24 (0%)	1/24 (4%)	0/24 (0%)
Herniation or midline shift	0/24 (0%)	0/24 (0%)	0/24 (0%)
Hydrocephalus	0/24 (0%)	0/24 (0%)	0/24 (0%)
Encephalomalacia	0/24 (0%)	0/24 (0%)	0/24 (0%)
Large subarachnoid spaces	0/24 (0%)	1/24 (4%)	0/24 (0%)
Any discrepancy	10/24 (42%)^{^a}	15/24 (63%)^{^a}	8/24 (33%)^{^a}

Table 5:

Diagnostic performance of consensus ufMRI, nHCT, and combined ufMRI and nHCT compared with stMRI^{^a}

	Sensitivity	Specificity	PPV	NPV
ufMRI	50%	100%	100%	31%
	(27%–73%)	(40%–100%)	(69%–100%)	(8%–58%)
nHCT	25%	100%	100%	21%
	(9%–49%)	(40%–100%)	(48%–100%)	(6%–46%)
Combined ultrafast with nHCT	60%	100%	100%	33%
	(36%–81%)	(40%–100%)	(74%–100%)	(10%–65%)