JP2002257915A - Magnetic field uniformity coefficient adjusting method and adjusting device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically adjust a shim value in a short time regardless of a local peak without the need of a skilled person's high-accuracy manual adjustment. SOLUTION: A shim value is set as chromosome information of genetic algorithm, the shim value is optimized by the genetic algorithm to a group of individuals having the chromosome information, and the optimum solution of the shim value is obtained by the individual having highest fitness among the group of individuals when the next generation individual group satisfies the evaluation criteria.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a magnetic field uniformity adjusting method, an adjusting device and a program applicable to a nuclear magnetic resonance apparatus (NMR) and the like.

[0002]

BACKGROUND OF THE INVENTION NMR
In ESR and ESR, a sample placed in a static magnetic field is irradiated with a high-frequency magnetic field to detect the resonance state of sample atoms, but the uniformity of the static magnetic field around the sample is directly affected by the resolution. Therefore, a correction magnetic field is generated by arranging a magnetic field correction coil called a shim composed of an assembly of coils in a magnetic field.

There are dozens of shim coils, and the uniformity of the magnetic field largely depends on how the value of the current (shim value) passed through each shim coil is determined. In the past, this shim value was often determined by intuition, experience, and skill while observing the resonance signal to a certain extent.To improve this, however, an NMR lock signal was used as a signal to monitor the resolution, and the shim coil was used. The current to be supplied is sequentially changed in accordance with a predetermined procedure, and a combination of currents having the highest resolution is obtained (Japanese Patent Publication No. Sho 63-43131).
A device has been proposed in which a plurality of magnetic field components are generated in each coil to control the current so that the sum of the calorific values of all the coils is minimized (JP-A-8-316031).

However, even with the conventionally proposed method, the number of combinations of shim values is enormous.
The fact is that you must rely on human experience and skill.

[0005]

The inventor of the present invention has found that the deviation from the ideal shim correction value can be represented by a function F having the shim value as an argument. F
Focusing on the point that is equivalent to obtaining the optimal solution of the above, the present inventors have found that a genetic algorithm can be applied to adjustment of shim values, and have reached the present invention.

Therefore, the present invention provides a method for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field, wherein a shim value is set as chromosome information of a genetic algorithm, and a population of individuals having the chromosome information is set. , The shim value is optimized by a genetic algorithm, and an optimal solution of the shim value is obtained by an individual having high fitness from the individual population when the next generation individual population satisfies the evaluation criteria.

Further, the present invention provides a method for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field,
A shim value is set as the chromosome information of the genetic algorithm, the virtual shim value on the simulator is optimized by the genetic algorithm so that a signal having the same shape as the signal taken from the device is output from the magnetic field simulator, and the sign of the virtual shim value is The method is characterized in that a process of returning the inverted one to the apparatus is performed one or more times.

Further, the present invention provides an apparatus for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in the magnetic field,
Ideal data corresponding to the state where the magnetic field uniformity is achieved,
Fitness calculating means for calculating fitness based on actual detection data, and setting a shim value as chromosome information of the genetic algorithm, the genetic operation of the genetic algorithm for a population of individuals having the chromosome information A genetic manipulation processing means for performing, and a judging means for judging whether or not the next-generation individual population satisfies the evaluation criterion; To obtain the optimal solution of the shim value.

Further, the present invention provides an apparatus for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field,
A shim value is set as chromosome information of the genetic algorithm, the virtual shim value is optimized by the genetic algorithm so as to output a signal having the same shape as the signal acquired from the device, and the sign of the virtual shim value is inverted. And a magnetic field simulator for returning to the above.

Further, the present invention provides a program for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field, wherein a shim value is set as chromosome information of a genetic algorithm, and a population of individuals having the chromosome information is set. To generate a child individual by partially crossing the chromosomes of a predetermined number of parent individuals selected at random from the population to obtain a child individual, and to mutate the chromosome of the generated child individual to obtain a child individual. The fitness is calculated based on the difference between the response and the ideal response, and the fitness of the child individual and the parent individual is compared to replace the parent individual in the population with a higher fitness individual, the selection, crossover, Mutation and substitution are repeated, and an optimal solution of the shim value is obtained by an individual having a high fitness from the individual population when the next generation individual population satisfies the evaluation criteria.

According to the present invention, there is provided a program for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field, wherein a shim value is set as chromosome information of a genetic algorithm, and the same shape as a signal fetched from an apparatus is used. In this method, the virtual shim value on the magnetic field simulator is optimized by a genetic algorithm so as to output the signal of (i), and the process of returning the sign of the virtual shim value to the apparatus is repeated.

[0012]

Embodiments of the present invention will be described below. First, the genetic algorithm will be schematically described. References for genetic algorithms include:
For example, the publisher ADDISON-WESLEY PUB
LISHING COMPANY, INC. Is 1989
Published by David E. Goldberg, Genetic Algorithms in Se
arch, Optimization, and Mac
hine learning ". In addition, the genetic algorithm referred to in the present invention refers to an evolutionary calculation method, and an evolutionary tactic (Evolution Strategies).
gy: ES). References to evolutionary tactics include, for example, publisher John Wiley
& Sons, published in 1995, P. Sch
Wefel's "Evolution and Opt"
"im seeking".

In a general genetic algorithm, first, a group of virtual organisms having genes is set, and individuals adapting to a predetermined environment survive according to the degree of adaptation, and descendants To increase the probability of leaving Then, the gene of the parent is passed on to the child by a procedure called genetic manipulation. By performing such generational changes and evolving genes and biological populations, individuals with high fitness occupy the majority of the biological population. As the genetic manipulation at that time, gene crossover, mutation, and the like, which occur in the reproduction of actual organisms, are used.

FIG. 1 is a flowchart showing a schematic procedure of such a genetic algorithm. Here, first, in step S11, the chromosome of an individual is determined. That is,
In the generation change, what kind of data is to be transmitted from a parent individual to a descendant individual and in what format are determined. FIG. 2 illustrates a chromosome. Here, the variable vector x of the target optimization problem is represented by M symbols Ai (i = 1, 2,
.. M), which is regarded as a chromosome consisting of M genes. In the figure, Ch indicates a chromosome and Gs indicates a gene. In this example, the number M of genes is 5. As the gene value Ai, a certain set of integers, a certain range of real values, a simple symbol sequence, etc. are determined according to the problem. In the example of FIG. 2, the alphabets a to e are genes. The set of genes thus encoded is the chromosome of an individual.

In step S11, a method of calculating a fitness value indicating how much each individual has adapted to the environment is determined. At this time, the design is made such that the higher or lower the value of the evaluation function of the target optimization problem is, the higher the fitness of the individual corresponding to the variable is. In the subsequent generation change, individuals with higher fitness have a higher probability of surviving or producing offspring than individuals with lower fitness. Conversely, an individual with low fitness is regarded as an individual that is not well adapted to the environment and is extinguished. This reflects the principle of natural selection in evolution. That is, the fitness is a measure of how excellent each individual is from the viewpoint of the possibility of survival.

In the genetic algorithm, at the start of the search, the target problem is generally a completely black box, and it is completely unknown what individual is desirable. For this reason, the initial population of organisms is usually generated randomly using random numbers. Therefore, even in the procedure here, step S1 after the processing is started in step S12.
In 3, the initial population is randomly generated using random numbers. If there is some prior knowledge about the search space, processing such as generation of a biological population may be performed mainly on a portion considered to have a high evaluation value. Here, the total number of individuals to be generated is referred to as the number of individuals in a group.

Next, in step S14, the fitness of each individual in the biological population is calculated based on the calculation method previously determined in step S11. After the fitness is determined for each individual, in step S15, the individual that is the basis of the next generation individual is selected from the population. However, performing only selection will only increase the proportion of individuals having the highest fitness at the present time in the biological population, and will not generate new search points. Therefore, the following operations called crossover and mutation are performed.

That is, in the next step S16, a pair of two individuals is randomly selected with a predetermined frequency of occurrence from the next generation individuals generated by the selection and the chromosomes are rearranged to change the child chromosomes. Make (crossover). Here, the probability of occurrence of crossover is called a crossover rate. The offspring individuals generated by the crossover are individuals that have inherited the trait from each of the parents. This crossover process increases the chromosome diversity of the individual and causes evolution.

After the crossover process, in the next step S17, the gene of the individual is changed with a certain probability (mutation). Here, the probability of occurrence of mutation is called a mutation rate. The phenomenon that gene content is rewritten with low probability,
This is a phenomenon that can be seen in the genes of actual organisms. However, if the mutation rate is too high, the inheritance characteristic of the parent trait due to crossover is lost, which is similar to random search in the search space, so care must be taken.

The next generation population is determined by the above processing. Here, in step S18, it is checked whether or not the generated next generation organism population satisfies the evaluation criteria for ending the search. This evaluation criterion depends on the problem applied, but typical ones are as follows.・ The maximum fitness value in the population has exceeded a certain threshold. -The average fitness of the whole population has exceeded a certain threshold.・ Generations in which the rate of increase in the fitness of the biological population is below a certain threshold have continued for a certain period.・ The number of generation changes has reached a predetermined number.

If any of the above-mentioned end conditions is satisfied, the process proceeds to step S19 to end the search, and an individual having a high fitness in the population at that time is determined as a solution to the optimization problem. Specifically, it is a solution to an optimization problem that seeks the individual with the highest fitness. If the termination condition is not satisfied, the process returns to the process of calculating the fitness of each individual in step S14, and the search is continued. By repeating such generation alternation, it is possible to increase the fitness of individuals while keeping the number of individuals in the population constant. The above is an outline of a general genetic algorithm.

Next, an example in which the genetic algorithm is applied to magnetic field uniformity adjustment will be described. FIG. 3 is a diagram for explaining a main part of NMR to which the magnetic field uniformity adjustment is applied. NMR
Is to arrange a shim 2 for magnetic field correction in the magnetic field of the superconducting magnet 1 to generate a highly uniform static magnetic field, arrange a sample 3 in this, irradiate a high-frequency magnetic field from a probe 4, The resonance state in the signal detection space 6 is detected by the detection coil 5.

Now, the origin is set at the center of the magnet 1, the center axis is the z-axis, and the x-axis and the y-axis are set on a plane orthogonal to the z-axis.
Assuming that θ and φ are three coordinate axis components of polar coordinates, a magnetic field component Bz in the z-axis direction near the origin of the magnet 1 is expressed by Expression (1).

[0024]

(Equation 1)

In general, at the center of a highly uniform magnet, x
Since the magnetic field component in the direction and the magnetic field component in the y direction are sufficiently small and negligible as compared with the magnetic field in the z direction, in the following description, for convenience, only the z direction will be considered. Is also applicable. The genetic algorithm framework described above is a loose one that does not specify the actual programming details,
It does not specify a detailed algorithm for each problem. Therefore, in order to use the genetic algorithm for the adjustment of the magnetic field uniformity in the present embodiment, the following items need to be realized for the adjustment of the magnetic field uniformity. (A) Chromosome expression method (b) Individual evaluation function (c) Crossover method (d) Mutation method (e) Selective selection method (f) Search termination condition [Chromosome expression method] Genetic algorithm in this embodiment As a chromosome information, a set of current values (shim values) passed through the shim coil is directly used. In the example of FIG. 4, for example, a set of shim values for adjusting the z-direction magnetic field is Z1 = z1,
Z2 = z2, Z3 = z3, Z4 = z4, Z5 = z5 (z
When 1 to z5 are genes, the chromosome of the genetic algorithm is described as (z1, z2, z3, z4, z5). That is, the shim value and the genetic data of the genetic algorithm correspond one-to-one. [Individual evaluation function] The individual evaluation function F of the genetic algorithm, that is, the fitness is, for example, a mean square value of a difference between an ideal response calculated from a linear shape of a standard sample and an actual response. , The horizontal axis is frequency (magnetic field strength),
In FIG. 5 where the vertical axis represents the resonance signal detection intensity, FIG.
Assuming that (a) is an ideal response distribution and FIG. 5 (b) is an actual response distribution, the mean square value of the difference between them is used as an evaluation function. Note that the fitness is not limited to this, and for example,
The strength of the magnetic field lock signal may be evaluated.

The evaluation function thus determined has a number of local optimum solutions from the characteristics of the equation (1). Therefore, if the hill-climbing method or the simplex method, which is a general search method, is used, the search is caught by a local optimum solution, and as a result, sufficient magnetic field uniformity cannot be obtained. However, since the genetic algorithm is a stochastic search method and a parallel search method using a plurality of search points, it has a feature that it is hard to be caught by a local optimum solution. The search can proceed very effectively in. [Crossover Method] Crossover is an operation for partially exchanging chromosomal genes. Here, a crossover method called one-point crossover is used. For example, in FIG. 6, Ch1 and Ch2 are chromosomes of parent individuals A and B, respectively, of (+2.31 −6.45 −3.28 +1.05 −4.12) (+3.92 +1.08 −4.53 −0.73 +3.30) It consists of genes. In the crossover process here,
These chromosomes are cut at randomly selected crossover positions CP. In the illustrated example, the crossover position is between the second gene and the third gene from the left. Then, by replacing the cut partial genotypes (third to fifth genes), chromosomes Ch3 and C3 are replaced.
The child individuals A 'and B' each having h4 are generated. Chromosomes Ch3 and Ch4 are (+2.31 -6.45 -4.53 -0.73 +3.30) (+3.92 +1.08 -3.28 +1.05 -4.12), respectively. The crossover method may use a method called two-point crossover or uniform crossover instead of one-point crossover. [Mutation method] Mutation is an operation that changes the gene of an individual with a certain probability.
The operation of adding normal random numbers generated for each gene of the chromosome according to the Gaussian distribution N (0, σ). 7, the chromosome Ch5 is (+2.31 −6.45 −3.28 +3.55 −4.12), and the normal random number N generated according to the Gaussian distribution
Is + (0.12 -0.05 +0.02 -0.07 -0.09), this is added to change to chromosome Ch6 (+2.45 -6.50 -3.26 +3.48 -4.21). Note that another distribution other than the Gaussian distribution, such as a Cauchy distribution, may be used as the random number to be added. When a gene is represented by a binary number of a plurality of bits, a target bit may be determined by a random number, and the value may be inverted. [Selection Selection Method] The selection selection of the genetic algorithm is a process of selecting an individual to be selected from a group of individuals having a set of shim values as chromosomes. Here, as shown in the processing flow of FIG. 8, for example, two parent individuals A and B are selected at random from a group, and the individual individuals A 'and B'
And selects the top two individuals C and D having the highest fitness values of the four individuals A, B, A ′, and B ′ (step S21).
The parent individuals A and B in the group are replaced with the individuals C and D (step S22). In addition to this replacement method, a general selection method of a generation model or a known selection method called a neighborhood model GA or a non-generation model GA may be used. [Search End Condition] The search ends when the evaluation function (fitness) of the individual satisfies a predetermined condition. Here, the fitness is defined as the mean square value of the difference between the ideal response and the actual response. Therefore, for example, the maximum or average fitness in the group may exceed a certain threshold.

As described above, the processing flow for determining the chromosome expression method, the individual evaluation function, the crossover method, the mutation method, the selection method, the search termination condition, and adjusting the magnetic field uniformity using a genetic algorithm is shown in FIG. 9 will be described. FIG.
9 is a flowchart showing a processing procedure of magnetic field uniformity adjustment using a genetic algorithm in the present embodiment. The present embodiment is characterized in that a set of shim values is directly used as chromosome information of a genetic algorithm, thereby eliminating the need for processing for converting chromosome information to shim values. As an initial group of the genetic algorithm, a plurality of individuals having chromosomes as shown in FIG. 4 are created using uniform random numbers (step S31). In this case, the value of each gene of each chromosome in the initial population takes a random real value between the upper limit and the lower limit. However, when there is some prior knowledge about the tendency of the correct shim value, an individual considered to have higher fitness may be created as the initial group.

For each individual in the initial population, a fitness is set, and selection, crossover, mutation, and substitution are performed to sequentially search. A major feature of this method is that replacement of a part of individuals in a group is repeated, and the individuals in the entire group are not changed at the same time unlike a general genetic algorithm. Thereby, it is possible to perform a search with a small number of individuals in a group.

Next, in the individual selecting process in step S32, two parent individuals A and B are randomly selected from the group,
The following genetic operations are performed on these.

In the crossover process in step S33, as shown in FIG. 6, the chromosomes of the parent individuals A and B selected by the selection are partially divided into a set of coordinate values at random positions by the one-point crossover method. Replace it. That is, Ch
1 and Ch2 are chromosomes of parent individuals A and B, and these chromosomes are cut at a randomly selected crossover position CP,
Chromosome Ch3, by replacing the cut partial genotype,
The child individuals A 'and B' each having Ch4 are generated.

The mutation in step S34, which is executed following the crossover in step S33, is performed by adding a normal random number generated according to the Gaussian distribution N (0, σ) to each gene of the chromosomes of the offspring individuals A 'and B'. The chromosome Ch5 is changed to chromosome Ch6 by performing a mutation by performing the following operation.

After the completion of the mutation operation in step S34, the shim values are respectively changed with the obtained chromosome values of the child individuals A 'and B', and the NMR is operated to operate the child individuals A 'and B'. 'Is calculated (steps S35 to S35).
8). After the fitness of the child individuals A 'and B' is calculated in this manner, the individual is replaced in step S39 in accordance with the processing procedure shown in FIG. That is, of the four individuals, parent individuals A and B and child individuals A 'and B', the top two individuals C and D with the highest fitness values are selected, and then the parent individuals A and B in the group are: Individuals C and D are replaced. In this method, when the parent individual has higher fitness than the child individual, no replacement is performed, so that the selection pressure is high and the convergence time of the search is fast.

The above selection, crossover, mutation, evaluation, and replacement operations are repeated, and when the search termination condition is satisfied in the determination of step S40, the adjustment process ends, and the adjustment process is performed in the individual population at that time. An individual with a high degree of fitness is used as a solution to the optimization problem, and an individual with the highest degree of fitness is preferably employed. If the termination condition is not satisfied, the process returns to the process of calculating the fitness of each individual in step S14, and the search is continued. If a chromosome (shim value) that satisfies the search termination condition is not obtained even after performing the adjustment process repeatedly for a certain number of generations, the population to be adjusted is determined to be defective in step S41, and the initial population is generated again. Perform processing. Of course, instead of recreating the entire population, 50% of the individuals may be randomly generated and replaced, and the magnetic field is too distorted without retrying to find the optimal solution. The warning may be issued and the operation may be stopped.

FIG. 10 is a functional block diagram of the shim value optimizing device of the above embodiment. A shim value optimizing device 10 including a data processing device having a memory and a CPU includes a fitness calculating means 11 for calculating fitness of an individual, selecting an individual from a population of organisms, and generating genes for crossover, mutation, substitution, etc. A genetic manipulation processing unit 12 for performing an operation, and an evaluation unit 13 for evaluating whether or not the biological population satisfies the search processing condition,
The signal from the NMR device 20 is taken in, and a process of supplying a correction current to the shim coil of the NMR device to generate a response is performed.

In the genetic algorithm, the search speed may decrease at the end of the search. This is because, since it is a global search method, the speed at which a certain local peak is reached is not comparable to the hill-climbing method. Therefore, in order to further reduce the search time, the hill-climbing method may be combined with the hill-climbing method as the fine-tuning method after performing the adjustment by the genetic algorithm. The criterion for switching from the genetic algorithm to the hill-climbing method may be, for example, the time when 90% of the initial target value is achieved by the genetic algorithm.

In the above embodiment, it takes several seconds to several minutes until the output value is actually stabilized using the standard sample, and this is repeated many times, so that the adjustment may take time. A method of using a magnetic field forward simulator as a device for further reducing the adjustment time will be described with reference to FIG.

The simulator (computer) 30 shown in FIG.
It has a shim value generating means 31, for example, as shown in FIG.
The response waveform generation means 32 generates a waveform having an ideal distribution as shown in FIG. Now, when a signal waveform as shown in FIG. 5B is obtained by the NMR apparatus 20, the signal is fetched and the simulator shown in FIG. 5A is made to match the waveform of FIG. Shim value above (virtual shim value)
Make adjustments. Although the same genetic algorithm as described above is used in the adjustment of the virtual shim value, this processing is performed on a simulator, so that high-speed processing is possible. The evaluation means 33 checks whether the two waveforms coincide with each other, and inverts the sign of the shim value when the two waveforms coincide with each other.
Returning to the R apparatus, the NMR apparatus 20 uses FIG.
Thus, a signal having the following waveform is obtained. If a satisfactory result cannot be obtained by one adjustment of the shim value, the above procedure is repeated.

The method of performing shim adjustment using this forward simulator may be implemented by this method alone.
The method may be used in combination with a method of actually operating R and applying a genetic algorithm. For example, after achieving about 70% of the target by a method using a simulator, the NMR may be operated and shim adjustment may be performed by a genetic algorithm so as to approximate an ideal distribution.

When performing shim adjustment, since parameters relating to waveform symmetry are known in equation (1), first, symmetry is set to fitness and a genetic algorithm is applied to obtain symmetry. If the property can be achieved, the parameter may be excluded and the other parameters may be adjusted by applying a genetic algorithm. Further, in the above description, the number of shims is assumed to be five, but it is clear that the above method is applicable even when adjusting a larger number of shims.

In the above embodiment, a genetic algorithm is used as a search method. However, in the fitness (evaluation function F) in the genetic algorithm,
When the number of local optimal solutions is small, an algorithm called annealing can be used instead of the genetic algorithm. Even when the number of local optimal solutions is large, the performance obtained as a result of the adjustment is lower than that of the genetic algorithm, but the search can be performed at high speed. For details of the annealing method, see, for example, JOHNWILE
Y. & Sons, published in 1989, E. Aar
ts and J.J. "Simulated" by Korst
Annealing andBoltzmann M
facilities. " The annealing method is a kind of hill-climbing method, in which a control parameter called a temperature is used so that the search is not trapped in the local optimum solution.

[0041]

As described above, according to the present invention, a real response is searched for according to a genetic algorithm so as to be an ideal response, so that a highly accurate manual adjustment by an expert is not required. Further, the shim value can be automatically adjusted in a short time without being bound by a local peak.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a schematic procedure of a genetic algorithm.

FIG. 2 is a diagram showing an example of a chromosome in a genetic algorithm.

FIG. 3 is an explanatory view of a main part of an NMR apparatus to which magnetic field uniformity adjustment is applied.

FIG. 4 is a diagram showing an example of a chromosome in a genetic algorithm.

FIG. 5 is a diagram illustrating an individual evaluation function of the genetic algorithm.

FIG. 6 is a diagram illustrating a crossover method.

FIG. 7 is a diagram illustrating a mutation.

FIG. 8 is a diagram showing a processing flow of selection selection.

FIG. 9 is a diagram illustrating a processing flow for performing magnetic field uniformity adjustment using a genetic algorithm.

FIG. 10 is a functional block diagram of the shim value optimizing device.

FIG. 11 is a functional block diagram using a forward simulator.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Superconducting magnet, 2 ... Shim for magnetic field correction, 3 ... Sample, 4
... probe, 5 ... detection coil, 6 ... signal detection space, 10
... Shim value optimization device, 11 ... Fitness calculation means, 12 ... Gene manipulation processing means, 13 ... Evaluation means, 30 ... Simulator, 31 ... Shim value generation means, 32 ... Response waveform generation means
33 ... Evaluation means.

 ──────────────────────────────────────────────────続 き Continuation of the front page (71) Applicant 592187165 Koji Suzuki 1-7-1 Kokura, Saiwai-ku, Kawasaki-shi, Kanagawa A-705 (71) Applicant 000004271 3-1-2-2 Musashino, Akishima-shi, Tokyo, Japan 72) Inventor Masahiro Murakawa 1-4-1 Umezono, Tsukuba, Ibaraki Pref., Ministry of Economy, Trade and Industry, National Institute of Advanced Industrial Science and Technology (72) Inventor Tetsuya Higuchi 1-4-1, Umezono, Tsukuba, Ibaraki (72) Inventor Koji Suzuki 1-A705, Kokura, Koyuki-ku, Kawasaki-shi, Kanagawa Prefecture (72) Inventor Hikonari Ishikawa 3-1-2 Musashino, Akishima-shi, Tokyo Japan Electronics Co., Ltd.

Claims (6)

[The claims] 1. A method for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field, wherein a shim value is set as chromosome information of a genetic algorithm, and a shim value is set for a population of individuals having the chromosome information. Magnetic field homogeneity adjustment method characterized by optimizing the value with a genetic algorithm and obtaining an optimal solution of shim value from individuals with high fitness from the population when the next generation population satisfies the evaluation criteria . 2. A method for uniformly adjusting a magnetic field by means of a magnetic field correcting shim arranged in a magnetic field, wherein a shim value is set as chromosome information of a genetic algorithm, and a signal having the same shape as a signal fetched from an apparatus is converted to a magnetic field. A method for adjusting the uniformity of a magnetic field, comprising optimizing a virtual shim value on a simulator by a genetic algorithm so as to output the same from a simulator and returning the virtual shim value with its sign inverted to a device at least once. 3. An apparatus for uniformly adjusting a magnetic field by a magnetic field correcting shim disposed in a magnetic field, wherein the fitness is determined based on ideal data corresponding to a state where the magnetic field uniformity is achieved and actual detection data. Calculating the fitness value, and setting a shim value as chromosome information of the genetic algorithm,
Genetic manipulation processing means for performing genetic manipulation of a genetic algorithm on the population of individuals having the chromosome information, and determination means for determining whether the next-generation population satisfies the evaluation criteria, A magnetic field homogeneity adjusting apparatus characterized in that an optimal solution of shim value is obtained by an individual having a high fitness from an individual population when an evaluation criterion is satisfied. 4. A device for uniformly adjusting a magnetic field by a magnetic field correcting shim disposed in a magnetic field, wherein a shim value is set as chromosome information of a genetic algorithm, and a signal having the same shape as a signal taken from the device is output. A magnetic field simulator for optimizing a virtual shim value by a genetic algorithm and returning the virtual shim value with its sign inverted to the device. 5. In a program for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field, a shim value is set as chromosome information of a genetic algorithm, and a population of individuals having the chromosome information is generated. A child individual is generated by partially crossing over chromosomes of a predetermined number of parent individuals selected at random from the population, and the chromosome of the generated child individual is mutated to obtain a response obtained by the child individual and an ideal response. The fitness is calculated based on the difference between the above, the fitness of the child individual and the parent individual are compared, and the parent individual in the population is replaced with a higher fitness individual, and the selection, crossover, mutation, and replacement are performed. A program for obtaining an optimal solution of the shim value by an individual having high fitness from the individual population when the next generation individual population satisfies the evaluation criteria. 6. In a program for uniformly adjusting a magnetic field by a magnetic field correcting shim arranged in a magnetic field, a shim value is set as chromosome information of a genetic algorithm, and a signal having the same shape as a signal taken in from a device is output. A program for optimizing a virtual shim value on a magnetic field simulator by a genetic algorithm, and returning the virtual shim value with its sign inverted to the apparatus.