Introduction
Methods
Methods part 1—Evaluation of the phantom
Structural integrity
“Reference” rT1 and rT2 data
Temperature sensitivity
B0 and B1 uniformity
Methods part 2—Multi-center phantom testing
Serial, multi-center T1 mapping data
Methods part 3—Statistical analysis
Temperature sensitivity
Correlation with rT1 times
T1 repeatability
Sources of T1 variation
Software upgrades
Results part 1—Evaluation of the phantom
Structural integrity
“Reference” rT1 and rT2 data
Temperature sensitivity
B0 and B1 uniformity
Results part 2—Multi-center phantom testing
T1 repeatability
Summary table ofx̅CoV ofT1(%) across 4 tubes | Order of tube IDs follows their orientation in the scanned bottle | |||||
Siemens | MOLLI 5s(3s)3s [448B] | 0.27 | ||||
Philips | MOLLI 3s(3s)5s | 0.54 | ||||
Siemens | SASHA | 0.56 | ||||
Siemens | ShMOLLI 5b(1b)1b(1b)1b [1041B] | 0.64 | ||||
Philips | SASHA | 0.92 | ||||
Philips | ShMOLLI 5b(1b)1b(1b)1bc | 1.04 | ||||
General Electric | MOLLI 5b(1b)1b(1b)1bd | 1.28 | ||||
General Electric | SMART | 3.00 | ||||
Platform | Sequence [prototype, #] | x̅’ CoV ofT1(%), globalx̅± SD ofT1according to native tube ID [tube rT1 by slow IR] | ||||
F [1090 ms] | E [1333 ms] | D [803 ms] | B [1489 ms] | |||
Siemens | MOLLI MyoMaps product 5s(3s)3s [5] | 1.07, 1041 ± 23 | 1.38, 1273 ± 35 | 0.85, 771 ± 6 | 1.45, 1538 ± 46 | |
MOLLI 3b(3s)3b(3s)5b [448, 3] | 0.87, 1019 ± 11 | 1.08, 1225 ± 17 | 0.65, 765 ± 6 | 1.19, 1462 ± 21 | ||
MOLLI 5b(3s)3b [448, 3] | 0.18, 1005 ± 13 | 0.53, 1217 ± 24 | 0.26, 762 ± 6 | 0.74, 1464 ± 23 | ||
MOLLI 5b(3s)3b [448B, 2] | 0.65, 1052 ± 7 | 0.82, 1289 ± 10 | 0.61, 784 ± 4 | 1.00, 1555 ± 16 | ||
MOLLI 5b(3s)3b [780, 2] | 0.70, 1048 ± 14 | 1.18, 1276 ± 20 | 0.61, 776 ± 6 sic. | 1.15, 1536 ± 24 | ||
MOLLI 5b(3s)3b [780B, 4] | 0.76, 1043 ± 9 | 0.99, 1279 ± 17 | 0.67, 776 ± 6 | 1.45, 1550 ± 37 | ||
MOLLI 5s(3s)3s [448, 2] | 0.89, 1035 ± 10 | 1.51, 1261 ± 25 | 1.04, 770 ± 9 | 1.44, 1538 ± 24 | ||
MOLLI 5s(3s)3s [448B, 1]a | 0.14, 1051 ± 1 | 0.38, 1282 ± 5 | 0.17, 783 ± 1 | 0.38, 1553 ± 6 | ||
MOLLI 5s(3s)3s [780B, 2] | 0.81, 1043 ± 15 | 0.69, 1277 ± 15 | 0.42, 771 ± 6 | 0.90, 1546 ± 25 | ||
MOLLI 5s(3s)3s [1041, 1] | 1.11, 1045 ± 12 | 0.69, 1285 ± 9 | 0.81, 772 ± 6 | 0.96, 1560 ± 15 | ||
MOLLI 5s(3s)3s [1041B, 1] | 1.27, 1033 ± 13 | 1.80, 1232 ± 22 | 0.66, 771 ± 5 | 1.47, 1494 ± 22 | ||
ShMOLLI 5b(1b)1b(1b)1b [448, 1] | 1.01, 989 ± 10 | 1.39, 1223 ± 17 | 0.60, 729 ± 4 | 1.50, 1510 ± 23 | ||
ShMOLLI 5b(1b)1b(1b)1b [448C, 1] | 0.61, 983 ± 6 | 0.64, 1218 ± 8 | 0.63, 727 ± 5 | 1.57, 1507 ± 22 | ||
ShMOLLI 5b(1b)1b(1b)1b [780B, 3] | 2.16, 945 ± 34 | 3.29, 1147 ± 62 | 0.98, 711 ± 11 | 3.70, 1392 ± 85 | ||
ShMOLLI 5b(1b)1b(1b)1b [1048, 1] | 0.70, 972 ± 7 | 0.71, 1208 ± 9 | 0.68, 715 ± 5 | 1.01, 1489 ± 15 | ||
ShMOLLI 5b(1b)1b(1b)1b [1041B, 1]a | 0.59, 978 ± 6 | 0.67, 1201 ± 8 | 0.43, 721 ± 3 | 0.86, 1501 ± 13 | ||
SASHA [4] | 0.39, 1104 ± 17 | 0.65, 1362 ± 27 | 0.47, 814 ± 8 | 0.73, 1522 ± 30 | ||
Philips | MOLLI CardiacQuant product 5s(3s)3s [2] | 0.53, 1025 ± 13 | 0.86, 1265 ± 19 | 0.50, 760 ± 7 | 1.04, 1516 ± 28 | |
MOLLI 3b(3s)3b(3s)5b [2] | 0.47, 1006 ± 17 | 0.94, 1247 ± 39 | 0.46, 751 ± 6 | 1.25, 1503 ± 46 | ||
MOLLI 3s(3s)3s(3s)5s [1] | 0.94, 1026 ± 10 | 1.66, 1251 ± 21 | 0.69, 765 ± 5 | 2.15, 1450 ± 31 | ||
MOLLI 3s(3s)5s [1]a | 0.20, 936 ± 2 | 1.11, 1100 ± 12 | 0.27, 722 ± 2 | 0.59, 1323 ± 8 | ||
MOLLI 5b(3s)3b(3s)2b [1] | 6.78, 1168 ± 79 | 6.32, 1374 ± 87 | 3.04, 816 ± 25 | 6.53, 1665 ± 109 | ||
MOLLI 5b(3s)3b [1] | 1.18, 1027 ± 12 | 1.48, 1265 ± 19 | 0.67, 758 ± 5 | 1.16, 1522 ± 18 | ||
MOLLI 5 s(3s)3s [3] | 0.89, 1012 ± 17 | 1.35, 1233 ± 31 | 1.20, 756 ± 9 | 1.76, 1456 ± 58 | ||
ShMOLLI 5b(1b)1b(1b)1bc [1] | 0.89, 1027 ± 9 | 1.37, 1270 ± 17 | 0.50, 753 ± 4 | 1.41, 1522 ± 21 | ||
SASHA [2] | 1.02, 1057 ± 44 | 0.89, 1336 ± 84 | 0.87, 798 ± 30 | 0.89, 1505 ± 33 | ||
General Electric | MOLLI 5b(3s)5b [1] | 11.18, 653 ± 73 | 7.69, 853 ± 66 | 8.45, 591 ± 50 | 6.68, 1319 ± 88 | |
MOLLI 5b(1b)1b(1b)1bd [1]a | 1.40, 575 ± 8 | 1.76, 642 ± 11 | 0.78, 543 ± 4 | 1.19, 1002 ± 12 | ||
SMART [1] | 2.05, 1063 ± 22 | 4.76, 1273 ± 61 | 0.95, 798 ± 8 | 4.22, 1430 ± 60 |
Summary table ofx̅CoV ofT1(%) across 4 tubes | Order of tube IDs follows their orientation in the scanned bottle | |||||
Philips | MOLLI 3b(3s)5b | 0.33 | ||||
Siemens | ShMOLLI 5b(1b)1b(1b)1b [780C] | 0.69 | ||||
Siemens | MOLLI 5b(3s)3b [448B] | 0.95 | ||||
Siemens | SASHA | 1.29 | ||||
Philips | SASHA | 3.76 | ||||
GE | SMART | 3.76 sic. | ||||
Platform | Sequence [prototype, #] | x̅’ CoV ofT1(%), globalx̅± SD ofT1according to native tube ID [tube rT1 by slow IR] | ||||
M [1260 ms] | L [1499 ms] | K [1010 ms] | J [1872 ms] | |||
Siemens | MOLLI MyoMaps product 5s(3s)3s [4] | 1.28, 1182 ± 20 | 1.61, 1367 ± 29 | 1.02, 955 ± 13 | 2.42, 1831 ± 62 | |
MOLLI 3b(3s)3b(3s)5b [448, 1] | 1.16, 1092 ± 13 | 1.92, 1252 ± 24 | 0.56, 889 ± 5 | 1.58, 1724 ± 27 | ||
MOLLI 5b(3s)3b [448B, 1]a | 0.36, 1233 ± 4 | 0.78, 1449 ± 11 | 0.76, 986 ± 7 | 0.86, 2012 ± 17 | ||
MOLLI 5b(3s)3b [780B, 2] | 0.99, 1204 ± 14 | 1.13, 1410 ± 18 | 0.74, 970 ± 9 | 1.50, 1912 ± 28 | ||
MOLLI 5s(3s)3s [780B, 1] | 1.46, 1222 ± 18 | 2.36, 1435 ± 34 | 1.76, 980 ± 17 | 3.04, 1916 ± 58 | ||
ShMOLLI 5b(1b)1b(1b)1b [780B, 2] | 1.17, 1155 ± 27 | 1.62, 1361 ± 35 | 0.80, 922 ± 17 | 2.45, 1882 ± 41 | ||
ShMOLLI 5b(1b)1b(1b)1b [780C, 1]a | 0.78, 1109 ± 9 | 0.89, 1322 ± 12 | 0.64, 888 ± 6 | 1.51, 1892 ± 29 | ||
SASHA [2] | 2.04, 1284 ± 30 | 1.15, 1534 ± 27 | 0.70, 1025 ± 13 | 1.27, 1949 ± 35 | ||
Philips | MOLLI 3b(3s)3b(3s)5b [4] | 1.64, 1147 ± 52 | 2.42, 1343 ± 80 | 1.05, 923 ± 29 | 3.90, 1785 ± 115 | |
MOLLI 3b(3s)5b [1]a | 0.01, 1055 ± 1 | 0.49, 1217 ± 6 | 0.05, 862 ± 1 | 0.78, 1710 ± 13 | ||
MOLLI 3s(3s)5s [1] | 5.41, 1175 ± 64 | 3.72, 1392 ± 52 | 7.13, 937 ± 67 | 3.56, 1833 ± 65 | ||
MOLLI 5b(3s)3b [3] | 4.18, 1098 ± 48 | 5.44, 1286 ± 73 | 4.37, 868 ± 42 | 6.33, 1784 ± 106 | ||
MOLLI 5s(3s)3s [6] | 2.40, 1131 ± 70 | 2.41, 1320 ± 108 | 2.60, 906 ± 34 | 2.71, 1727 ± 131 | ||
MOLLI 5b(1b)1b(1b)1bc [2] | 2.98, 948 ± 97 | 4.53, 1082 ± 137 | 0.96, 794 ± 60 | 6.29, 1286 ± 282 | ||
SASHA [2] | 3.11, 1290 ± 46 | 3.40, 1826 ± 171 | 2.11, 1054 ± 38 | 6.41, 2157 ± 267 | ||
General Electric | SMART [1] | 2.85, 1150 ± 33 | 4.13, 1227 ± 51 | 2.04, 973 ± 20 | 6.03, 1419 ± 86 |
Sources of T1 variation
Model fitting formulas | AIC | BIC | Log likelihood | χ2 | P value | Best model | ||
---|---|---|---|---|---|---|---|---|
1.5 T model | A1: T1 ~ sequence + (1|ID) | 5850.3 | 5920.8 | − 2909.1 | “ref” | “ref” | “ref” | |
A2: T1 ~ sequence + (1|ID/age) [worse fit with age] | 5852.3 | 5927.2 | − 2909.1 | 0.0 | 1.000 | – | ||
A3: T1 ~ sequence × software + (1|ID) | 5725.4 | 5883.9 | − 2826.7 | 164.9 | < 0.0001 | – | ||
A4: T1 ~ sequence × software + vendor + (1|ID) [vendor non-contributory to fit] | 5725.4 | 5883.9 | − 2826.7 | 0.0 | 1.000 | – | ||
A5: T1 ~ sequence × software + model + (1|ID) | 5723.4 | 5890.8 | − 2823.7 | 5.9 | 0.051 | A5 | ||
A6: T1 ~ sequence × software + vendor × model + (1|ID) [vendor non-contributory to fit] | 5723.4 | 5890.8 | − 2823.7 | 0.0 | 1.000 | – | ||
Final model (A5) parameters | Variance | SD | χ2 | Pvalue | ||||
Random effects | ID | 2547.2 | 50.47 | “ref” | “ref” | |||
Fixed effects | Sequence (β range − 32.3 to 455.6a) | / | / | 1101.0 | < 0.0001 | |||
Software (β range 385.1 to 522.3a) | / | / | 63.8 | < 0.0001 | ||||
Model (β range − 161.8 to 16.0a) | / | / | 11.2 | 0.011 | ||||
Sequence: Software (β range − 161.8 to − 6.9a) | / | / | 149.6 | < 0.0001 | ||||
3 T model | A1: T1 ~ sequence + (1|ID) | 4390.4 | 4447.0 | − 2181.2 | “ref” | “ref” | “ref” | |
A2: T1 ~ sequence + (1|ID/age) [worse fit with age] | 4392.4 | 4453.1 | − 2181.2 | 0.0 | 1.000 | – | ||
A3: T1 ~ sequence × software + (1|ID) | 4238.3 | 4339.4 | − 2094.2 | 174.1 | < 0.0001 | A3 | ||
A4: T1 ~ sequence × software + vendor + (1|ID) [vendor non-contributory to fit] | 4238.3 | 4339.4 | − 2094.2 | 0.0 | 1.000 | – | ||
A5: T1 ~ sequence × software + model + (1|ID) | 4239.2 | 4348.4 | − 2092.6 | 3.1 | 0.212 | – | ||
A6: T1 ~ sequence × software + vendor × model + (1|ID) [vendor non-contributory to fit] | 4239.2 | 4348.4 | − 2092.6 | 0.0 | 1.000 | – | ||
Final model (A3) parameters | Variance | SD | χ2 | P-value | ||||
Random effects | ID | 372.4 | 19.3 | “ref” | “ref” | |||
Fixed effects | Sequence (β range − 114.5 to 38.5a) | / | / | 732.8 | < 0.0001 | |||
Software (β range − 203.0 to 32.7a) | / | / | 46.0 | < 0.0001 | ||||
Sequence: Software (β range − 48.8 to 267.3a) | / | / | 189.0 | < 0.0001 |
Software upgrades
Discussion
Dedicated T1 device | Combined T1/T2 device | |||
---|---|---|---|---|
Phantom | HCMR (cardiac specific) | Brompton (cardiac specific) | T1MES (cardiac specific) | ISMRM/NIST system phantom (not cardiac specific) |
Field strength specificity | ☒ | ☒ | ☑ | ☒ |
Tube ingredients | NiCl2-doped agar + carrageenan | NiCl2-doped agar | NiCl2-doped agar | NiCl2-doped water, MnCl2-doped agar |
Structure | 9 T1/T2 tubes in an amber PVC sealed jar that is ~ 25% smaller in total volume compared to T1MES. | Air-filled box containing 4 duplicate T1/T2 glued glass tubes (total 8) requires cylindrical MRI test bottles on either side for B1 field/reference frequency calibration. | 9 plastic T1/T2 tubes in an amber PVC sealed jar with inter-tube gaps packed with a carefully specified agarose/ HDPE plastic macrobead/NiCl2-doped fill that flattened the B1 field and provided sufficient B0 homogeneity to obviate the need for side phantoms. | A layer of 14 T1 spheres and a separate layer of 14 T2 spheres so no unified T1/T2 compartment representing the relaxation parameters of the human heart. |
Cardiac T1/T2coverage | Health and disease, 9 biologies but limited T2 coverage (57 and 75 ms). | Health only, 4 biologies: (1) native myocardium, (2) native blood, (3) post-GBCA myocardium, and (4) post-GBCA blood. | Health and disease, 9 biologies and broad T2 coverage (1.5 T, 44, 48, 50, 155, 189, and 243 ms). Also see Supplementary Table 3. | Health and disease, but of the 14 T1 spheres (ranging from 21 to 2038 ms), half are not useful to cardiac T1 mapping as T1 times are too short for either native or post-GBCA myocardium/blood. Half the T2 spheres (ranging from 11 to 581 ms) are also anti-physiological for cardiac mapping. Also, the T1/T2 ratio is not representative of cardiac tissue. |
MT coverage | – | – | – | – |
Regulatory clearance | – | – | FDA, CE-mark | – |
Developers | Piechnik et al. | Vassiliou et al. | Captur et al. | NIST/ISMRM |