SWI or T2*: Which MRI Sequence to Use in the Detection of Cerebral Microbleeds? The Karolinska Imaging Dementia Study

Article Figures & Data

Download figure
Open in new tab
Download powerpoint
Fig 1.
CMBs on tSWI but not T2*. For all images: A, T2*. B, tSWI. 1, Only Rater 1 identified the CMBs on tSWI. 2, Raters 1 and 2 identified the CMBs on tSWI.
Download figure
Open in new tab
Download powerpoint
Fig 2.
CMBs on both T2* and tSWI. For all images: A, T2*. B, tSWI. 1, All raters identified the single CMB on both T2* and tSWI. 2, The pale CMB on T2* was only identified by Rater 2. On tSWI, the CMB was clearly delineated and was identified by all raters. 3 and 4, Disagreement on the exact number of CMBs occurred on T2*. On tSWI, the CMBs are more clearly outlined.

Table 1:

CMBs for each sequence and rater^{^a}

	All Raters	1	2	3
Prevalence of CMBs on T2* (No.) (%)	43 (17)^{^b}	42 (17)^{^b}	45 (18)	38 (15)^{^b,^c}
Prevalence of CMBs on TSWI (No.) (%)	50 (20)	49 (20)	46 (19)	51 (21)^{^c}
Prevalence of CMBs on tSWI (No.) (%)	51 (21)^{^b}	51 (21)^{^b}	50 (20)	51 (21)^{^b}
T2* sum CMBs (No.)	365^{^b,^c}	373^{^b,^c}	461^{^b}	260^{^b,^c}
TSWI sum CMBs (No.)	528^{^c}	570^{^c}	450^{^d}	576^{^c}
tSWI sum CMBs (No.)	699^{^b}	619^{^b}	806^{^b,^d}	672^{^b}
Times more CMBs on TSWI than T2*	1.5	1.5	1.0	2.2
Times more CMBs on tSWI than T2*	1.9	1.7	1.7	2.6
Times more CMBs on tSWI than TSWI	1.3	1.1	1.8	1.2

↵a McNemar and Wilcoxon signed rank tests were used to determine the difference in prevalence and sum of CMBs, respectively; analysis was done separately between T2* and TSWI, T2* and tSWI, and TSWI and tSWI. “All Raters” represents the median value of all separate ratings for all patients (ie, not the final values). The whole cohort is included in the analysis (N = 246). For “Prevalence,” values are given as the number of patients (percentage). All significant values are designated in the Table. All P values have been Bonferroni-corrected.
↵b P < .05 between T2* and tSWI.
↵c P < .05 between T2* and TSWI.
↵d P < .05 between TSWI and tSWI.

Table 2:

CMBs on T2*, TSWI, and tSWI in the separate dementia diagnoses^{^a}

Diagnosis	AD (n = 62)	MCI (n = 80)	Other (n = 33)	SCI (n = 71)
Prevalence of CMBs on T2* (No.) (%)	20 (32)	13 (16)	6 (16)	4 (6)
Prevalence of CMBs on TSWI (No.) (%)	22 (32)	16 (20)	7 (22)	5 (7)
Prevalence of CMBs on tSWI (No.) (%)	22 (35)	16 (20)	7 (22)	6 (8)
T2* sum CMBs (No.)	258^{^b,^c}	69^{^c}	47	6
TSWI sum CMBs (No.)	344^{^b}	101	73	10
tSWI sum CMBs (No.)	459^{^c}	116^{^c}	79	12
Times more CMBs on TSWI than T2*	1.3	1.5	1.6	1.7
Times more CMBs on tSWI than T2*	1.8	1.7	1.7	2.0
Times more CMBs on tSWI than TSWI	1.3	1.1	1.1	1.2

Note:—AD indicates Alzheimer disease; Other, other dementias; MCI, mild cognitive impairment; SCI, subjective cognitive impairment.
↵a Differences in prevalence and sum of CMBs were determined with the McNemar and Wilcoxon signed rank test, respectively; analysis was made between T2* and TSWI, T2* and tSWI, and TSWI and tSWI. Ratings represent the median value of all separate ratings for all patients. All significant values are designated in the Table. All P values have been Bonferroni-corrected.
↵b P < .05 between T2* and TSWI.
↵c P < .05 between T2* and tSWI.

Table 3:

Interrater agreement for CMBs on T2*, TSWI, and tSWI^{^a}

Sequence^{^b}	Patients (No.)	All Raters^{^c}	Raters 1 and 2	Raters 1 and 3	Raters 2 and 3
Analysis 1
T2*	55	0.942 (0.915–0.961)	0.990 (0.981–0.994)	0.858 (0.758–0.917)	0.818 (0.689–0.894)
TSWI	55	0.982 (0.974–0.988)	0.963 (0.942–0.977)	0.994 (0.990–0.996)	0.959 (0.936–0.974)
tSWI	55	0.991 (0.986–0.994)	0.979 (0.964–0.988)	0.997 (0.995–0.998)	0.984 (0.972–0.991)
Analysis 2
T2*	25	0.959 (0.795–0.944)	0.992 (0.982–0.996)	0.911 (0.798–0.961)	0.882 (0.733–0.948)
TSWI	37	0.984 (0.972–0.991)	0.968 (0.938–0.984)	0.994 (0.988–0.997)	0.963 (0.929–0.981)
tSWI	33	0.991 (0.985–0.995)	0.983 (0.966–0.992)	0.997 (0.994–0.999)	0.983 (0.966–0.992)

↵a Interrater agreement was determined with the intraclass correlation coefficient, and the 95% confidence intervals are given in parenthesis.
↵b Analysis was made twice: 1) for all patients (n = 55) with CMBs, on the basis of the initial KIDS CMB analysis; 2) for all patients with >1 CMB according to Rater 1.
↵c “All Raters” represents analysis among Raters 1, 2, and 3.

Table 4:

Clinical parameters and association with CMBs depending on sequence^{^a}

Clinical Parameters	T2* CMB+ (n = 43)	T2* CMB− (n = 203)	TSWI CMB+ (n = 50)	TSWI CMB− (n = 196)	tSWI CMB+ (n = 51)	tSWI CMB− (n = 195)
Women (No.) (%)	19 (44)	113 (56)	22 (44)	110 (56)	22 (43)	109 (56)
Age (yr) (mean) (SD)	68 (11)^{^b}	63 (10^{)^b}	68 (11)^{^b}	63 (10)^{^b}	68 (11)	63 (10)
Hypertension (No.) (%)	20 (47)	74 (37)	23 (46)	71 (36)	24 (47)	70 (36)
Hyperlipidemia (No.) (%)	6 (14)	31 (16)	8 (16)	29 (15)	9 (18)	28 (15)
Diabetes (No.) (%)	7 (16)	21 (10)	8 (16)	20 (10)	7 (14)	21 (11)
Current smoking (No.) (%)	4 (12)	35 (23)	6 (15)	33 (22)	6 (15)	33 (23)
Alcohol (No.) (%)	30 (91)	131 (82)	35 (90)	126 (82)	36 (92)	125 (82)
Heredity (No.) (%)	21 (72)	82 (56)	23 (67)	80 (57)	23 (68)	80 (57)
Anticoagulant medication (No.) (%)	11 (26)	36 (18)	14 (28)	33 (17)	13 (26)	34 (18)
MMSE (mean) (SD)	25 (4)	25 (5)	25 (4)	25 (5)	25 (4)	25 (5)
BMI (mean) (SD)	27 (5)	25 (3)	25 (4)	27 (5)	25 (4)	27 (5)
WMH ≥1 (No.) (%)	38 (88)^{^b}	138 (68)^{^b}	45 (90)^{^b}	131 (74)^{^b}	45 (88)^{^b}	130 (67)^{^b}
WMH ≥2 (No.) (%)	19 (44)^{^c}	20 (10)^{^c}	21 (42)^{^c}	18 (9)^{^c}	21 (41)^{^c}	18 (9)^{^c}
WMH = 3 (No.) (%)	12 (28)^{^c}	8 (4)^{^c}	14 (28)^{^c}	6 (3)^{^c}	14 (28)^{^c}	6 (3)^{^c}

Note:—MMSE indicates Mini-Mental State Examination; BMI, body mass index; CMB+, patients with CMBs; CMB−, patients without CMBs.
↵a χ² and Mann-Whitney U tests were done to determine in-sequence differences; significance testing was only done within sequences, and significant values are designated in the Table. All analyses were made on the basis of the median rating value of all raters on all patients. All P values have been Bonferroni-corrected.
↵b P < .05.
↵c P < .001.