Hybrid spatial and circuit optimization for targeted performance of MRI coils
A method of operating a multi-coil magnetic resonance imaging system is disclosed which includes a controller performing a simulation using a predefined tissue model, determining output values of a variable of interest (VOI) associated with operation of two or more coils of an MRI system based on the simulation, comparing the simulated output values of the VOI to an a priori target values of the VOI, if the simulated output values of the VOI are outside of a predetermined envelope about the a priori target values of the VOI, then performing an optimization, wherein the optimization includes iteratively adjusting the circuit values until the simulated output values of the VOI are within the predetermined envelope about the a priori target values of the VOI thereby establishing VOI optimized values, and loading the established VOI optimized values and operating the magnetic resonance imaging system on the tissue to be imaged.
1. A method of operating a multi-coil magnetic resonance imaging system, comprising:
a controller having a processor and software loaded on tangible memory performing a simulation using a predefined tissue model for a tissue to be imaged;
the controller determining output values of a variable of interest associated with operation of two or more coils of a magnetic resonance imaging system based on the simulation;
the controller comparing the simulated output values of the variable of interest to an a priori target values of the variable of interest;
if the simulated output values of the variable of interest are outside of a predetermined envelope about the a priori target values of the variable of interest, then the controller performing a first optimization, wherein the first optimization includes:
iteratively adjusting the circuit values until the simulated output values of the variable of interest are within the predetermined envelope about the a priori target values of the variable of interest thereby establishing variable of interest optimized values; and
loading the established variable of interest optimized values and operating the magnetic resonance imaging system on the tissue to be imaged.
2. The method of claim 1 , wherein the first optimization is based on a cost function defined as:
f
(
x
)
=
w
1
❘
"\[LeftBracketingBar]"
diag
(
S
(
x
)
)
❘
"\[RightBracketingBar]"
-
S
i
i
+
w
2
❘
"\[LeftBracketingBar]"
S
r
(
x
)
❘
"\[RightBracketingBar]"
-
S
i
j
+
w
3
S
td
(
B
1
+
)
Mean
(
B
1
+
)
-
target
wherein
ƒ(x) is the cost function of a real number vector x whose entries are coil parameters,
∥ ∥ denotes the Euclidean distance,
| | is the elementwise absolute values,
w 1-3 are weights,
B 1 + represents field associated with each coil of the two or more coils, and
target represents target values for the variable of interest.
3. The method of claim 2 , wherein the variable of interest is cross-talk between any two neighboring coils.
4. The method of claim 2 , wherein variable of interest is B 1 + homogeneity.
5. The method of claim 2 , wherein variable of interest is B 1 + power efficiency.
6. The method of claim 2 , wherein variable of interest is a combination of cross-talk between any two neighboring coils and B 1 + homogeneity.
7. The method of claim 2 , further comprising:
exporting one or more of the circuit values, magnetic and electric field values as well as S-parameters of each of the two or more coils;
performing an initial scan of a tissue of interest associated with the tissue model;
obtaining actual output values of the variable of interest;
if the actual output values of the variable of interest of the initial scan are outside of a predetermined envelope about the a priori target values of the variable of interest, then performing a second optimization, wherein the second optimization includes:
establishing an error parameter based on the actual measured variable of interest from the initial scan vs. the loaded target values, and
iteratively adjusting the circuit values using a first gradient descent optimization process,
re-computing the variable of interest, and
re-comparing the recomputed values of the variable of interest to the loaded target values of the variable of interest until the recomputed variable of interest values are within the predetermined envelope about the a priori target values of the variable of interest.
8. The method of claim 7 , wherein the variable of interest is one of cross-talk, B 1 + homogeneity, B 1 + power efficiency, and any combination thereof.
9. The method of claim 7 , further comprising:
exporting one or more of the post circuit values, magnetic and electric field values as well as S-parameter of each of the two or more coils;
continuously operating the two or more coil;
obtaining continuous actual output values of the variable of interest;
if the continuous actual output values of the variable of interest are outside of a predetermined operational envelope about the a priori target values of the variable of interest, then performing a third optimization, wherein the third optimization includes:
iteratively adjusting the circuit values using a second gradient descent optimization process,
continuously operating the two or more coils, and
re-comparing the actual output values of the variable of interest to the a priori target values of the variable of interest until the actual output values are within the predetermined operational envelope.
10. The method of claim 9 , wherein the variable of interest is one of cross-talk, B 1 + homogeneity, B 1 + power efficiency, and any combination thereof.
11. A drive system for a multi-coil magnetic resonance imaging system, comprising:
two or more coils utilized for imaging a tissue of interest;
a drive circuit for driving the two or more coils;
a controller having a processor and software loaded on tangible memory adapted to:
perform a simulation using a predefined tissue model for a tissue to be imaged
determine output values of a variable of interest associated with operation of two or more coils of a magnetic resonance imaging system based on the simulation;
compare the simulated output values of the variable of interest to an a priori target values of the variable of interest;
if the simulated output values of the variable of interest are outside of a predetermined envelope about the a priori target values of the variable of interest, then perform a first optimization, wherein the first optimization includes:
iteratively adjust the circuit values until the simulated output values of the variable of interest are within the predetermined envelope about the a priori target values of the variable of interest thereby establishing variable of interest optimized values; and
load the established variable of interest optimized values and operate the magnetic resonance imaging system on the tissue to be imaged.
12. The system of claim 11 , wherein the first optimization is based on a cost function defined as:
f
(
x
)
=
w
1
❘
"\[LeftBracketingBar]"
diag
(
S
(
x
)
)
❘
"\[RightBracketingBar]"
-
S
i
i
+
w
2
❘
"\[LeftBracketingBar]"
S
r
(
x
)
❘
"\[RightBracketingBar]"
-
S
i
j
+
w
3
S
td
(
B
1
+
)
Mean
(
B
1
+
)
-
target
wherein
ƒ(x) is the cost function of a real number vector x whose entries are coil parameters,
∥ ∥ denotes the Euclidean distance,
| | is the elementwise absolute values,
w 1-3 are weights,
B 1 + represents field associated with each coil of the two or more coils, and
target represents target values for the variable of interest.
13. The system of claim 12 , wherein the variable of interest is cross-talk between any two neighboring coils.
14. The system of claim 12 , wherein variable of interest is B 1 + homogeneity.
15. The system of claim 12 , wherein variable of interest is B 1 + power efficiency.
16. The system of claim 12 , wherein variable of interest is a combination of cross-talk between any two neighboring coils and B 1 + homogeneity.
17. The system of claim 12 , the controller further adapted to:
export one or more of the circuit values, magnetic and electric field values as well as S-parameter of each of the two or more coils;
perform an initial scan of a tissue of interest associated with the tissue model;
obtain actual output values of the variable of interest;
if the actual output values of the variable of interest of the initial scan are outside of a predetermined envelope about the a priori target values of the variable of interest, then performing a second optimization, wherein the second optimization includes:
establish an error parameter based on the actual measured variable of interest from the initial scan vs. the a priori target values, and
iteratively adjust the circuit values using a first gradient descent optimization process,
re-compute the variable of interest, and
re-compare the recomputed values of the variable of interest to the a priori target values of the variable of interest until the recomputed variable of interest values are within the predetermined envelope about the a priori target values of the variable of interest.
18. The system of claim 17 , wherein the variable of interest is one of cross-talk, B 1 + homogeneity, B 1 + power efficiency, and any combination thereof.
19. The system of claim 17 , the controller further adapted to:
export one or more of the post circuit values, magnetic and electric field values as well as S-parameter of each of the two or more coils;
continuously operate the two or more coil;
obtain continuous actual output values of the variable of interest;
if the continuous actual output values of the variable of interest are outside of a predetermined operational envelope about the a priori target values of the variable of interest, then performing a third optimization, wherein the third optimization includes:
iteratively adjust the circuit values using a second gradient descent optimization process,
continuously operate the two or more coils, and
re-compare the actual output values of the variable of interest to the a priori target values of the variable of interest until the actual output values are within the predetermined operational envelope.
20. The system of claim 19 , wherein the variable of interest is one of cross-talk, B 1 + homogeneity, B 1 + power efficiency, and any combination thereof.