Method of optimizing NMR measurement condition

Abstract

There is disclosed a method of finding a reliable optimum value of a measurement condition in an NMR measurement. This method starts with gaining NMR measurement data while varying the value of the measurement condition to be optimized. Then, a certain property is extracted as a numerical value from the measurement data. A plot is made of the measurement data versus the value of the measurement condition to create a curve. A model equation coincident with the measurement condition, its range, and the certain property extracted as a numerical value is established. Curve fitting is done in which constants of the model equation are varied such that the equation agrees the created curve. The constant values of the model equation and their standard deviation are obtained by curve fitting. An optimum value of the measurement condition is obtained from the results.

Description

BACKGROUND OF THE INVENTION

[0001] 1. Field of the Invention

[0002] The present invention relates to a method of optimizing a measurement condition in an NMR measurement.

[0003] 2. Description of Related Art

[0004] FIGS. 1(a) and 1(b) illustrate the prior art method of optimizing an NMR measurement condition. FIG. 1(a) is a flowchart illustrating a general procedure for finding an optimum value of a measurement condition. FIG. 1(b) is a flowchart illustrating a procedure for finding optimum values of RF pulse widths. FIG. 1(c) is a diagram illustrating a pulse sequence for measurement of an RF pulse width.

[0005] The general procedure for finding an optimum value of a measurement condition is described by referring to FIG. 1(a). In step 1, NMR measurements are carried out while varying the value of a measurement condition to be optimized in given increments. In step 2, a graph for finding the optimum value from the obtained measurement data by appropriate processing is created. At this time, the varying value of the measurement condition is plotted on one axis of the graph. In step 3, the optimum value is found from the shape of the graph by visual estimation.

[0006] As a specific example, a procedure for finding an optimum value of an RF pulse width as a measurement condition is next described by referring to FIGS. 1(b) and 1(c). In step 1, NMR measurements are performed while varying the pulse width from 0 to 70 is using a pulse sequence shown in FIG. 1(c), for example, under measurement conditions listed in Table 1.

TABLE 1
Measurement condition set 1 under which measurements
are performed with varying RF pulse width
Item                            Value
Sample                          15 mM copper dichloride/1% H2O,
                                99% D2O
Magnetic field intensity        14.09636928 T
Observed nucleus                1H
Observe frequency               600.1723046 MHz
Center frequency of observation about 4.7 ppm
                                (resonance frequency of water)
Number of data points           16384
Sweep width                     9.00252071 kHz
Number of accumulations         1
B1 pulse intensity              about 25 kHz
Observation time                1.81993472 s
Relaxation_delay                1 s
Temperature                     25° C.

[0007] In the pulse sequence of FIG. 1(c), “[relaxation_delay]” indicates the wait time of each repetition pulse. In this example, the time is 1 s.

[0008] “[x_pulse]” indicates an RF pulse. In this example, NMR measurements are performed using a pulse width varied from 0 to 70 μs.

[0009] “[acquisition]” indicates an observation. In this example, the time necessary for an observation is 1.81993472 s as shown in Table 1.

[0010] Data obtained from a measurement is shown in FIG. 2, where one-dimensional (1D) NMR data obtained using a certain pulse width are arrayed in the order of values of pulse widths at intervals of 2 μs.

[0011] In step 2, the obtained NMR data are first Fourier-transformed. The resulting data are shown in FIG. 3, where the data are arrayed in the order of values of pulse widths at intervals of 2 μs in the same way as in FIG. 2. Then, with respect to each set of 1D NMR data, a range from 4 to 5.5 ppm in the signal region is displayed. The data are arrayed horizontally in the order of values of pulse widths. The obtained graph is shown in FIG. 4, where the horizontal axis indicates the pulse width, while the vertical axis indicates the intensity of the NMR spectrum.

[0012] In step 3, a waveform formed by connecting the vertices of spectral intensities of FIG. 4 by means of straight lines is regarded as a sinusoidal (SIN) wave. Visual estimation of a pulse width of 360° reveals that it is about 28 μs. Since it has been already known that the optimum pulse width is 90°, the optimum pulse width is equal to the pulse width of 360° divided by 4, i.e., 28 μs/4=7 μs.

[0013] An NMR instrument designed to quantitatively indicate the nonuniformities in transmit and receive magnetic fields is shown in Japanese Patent Laid-Open No. H3-139330. In particular, NMR scans are made with RF exciting field intensities of different arrays. A curve is applied to each set of corresponding data elements in one set of intensity arrays. The peaks of the applied curves are determined. Corresponding data in the transmit and receive arrays are generated from the determined peaks. Thus, a magnetic field map indicating nonuniformities in an RF magnetic field by means of the magnitudes of data elements is created.

[0014] However, the prior art method of finding the optimum RF pulse width has the problem that the reliability of the optimum value is low because the value is found from a created graph by visual estimation. In order to obtain an optimum value with high reliability, it is necessary to increase the number of measurement data items. This prolongs the measurement time. If the number of measurement data items is reduced to shorten the measurement time, the reliability of the obtained optimum value deteriorates.

[0015] Furthermore, Japanese Patent Laid-Open No. H3-139330 does not disclose a technique for optimizing measurement conditions, though the reference discloses a technique quantitatively indicating nonuniformities in transmit and receive fields.

SUMMARY OF THE INVENTION

[0016] It is an object of the present invention to provide a technique for finding a reliable optimum value of a measurement condition with a reduced number of measurement data items, i.e., in a short time.

[0017] A method of optimizing an NMR measurement condition in accordance with the present invention starts with gaining NMR measurement data while varying the value of the measurement condition to be optimized. Then, a certain property is extracted as a numerical value from the measurement data. A plot is made along the varied value of the measurement condition to create a curve. A model equation coincident with the measurement condition, its range, and the certain property extracted as a numerical value is established. Curve fitting in which constants of the model equation are varied is done such that the equation agrees with the created curve. Constant values of the model equation and their standard deviation are obtained by curve fitting. An optimum value of the measurement condition is obtained from the results.

[0018] In the present invention, curve fitting is used and so a good result can be obtained if the number of data items is reduced as long as they characterize a waveform. A reliable optimum value of a measurement condition can be found in a short time.

[0019] Other objects and features of the invention will appear in the course of the description thereof, which follows.

BRIEF DESCRIPTION OF THE DRAWINGS

[0020] FIGS. 1(a), 1(b), and 1(c) are flowcharts and a diagram illustrating the prior art method of optimizing an NMR measurement condition;

[0021] FIG. 2 is a graph of one-dimensional(1D) NMR data obtained by the method illustrated in FIGS. 1(a), 1(b), and 1(c);

[0022] FIG. 3 is a graph illustrating data obtained by Fourier-transforming the 1D NMR data shown in FIG. 2;

[0023] FIG. 4 is a graph consisting of pulses having different widths and obtained from the 1 D NMR data shown in FIG. 2, and in which the pulse widths are arrayed in the order of their values in a range from 4 to 5.5 ppm;

[0024] FIG. 5 is a flowchart illustrating Embodiment 1 of the present invention;

[0025] FIG. 6 is a graph of curves obtained by plotting the integrated values in a given signal region;

[0026] FIG. 7 is a flowchart illustrating Embodiment 2 of the present invention;

[0027] FIG. 8 is a graph showing NMR measurement data close to the 360° pulse width;

[0028] FIG. 9 is a diagram showing data obtained after processing;

[0029] FIG. 10 is a diagram showing a curve obtained by plotting integrated values;

[0030] FIG. 11 is a flowchart illustrating Embodiment 3 of the present invention;

[0031] FIG. 12 is a diagram showing data obtained by a measurement in which an RF pulse width was varied;

[0032] FIG. 13 is a diagram showing data obtained after processing;

[0033] FIG. 14 is a graph showing a curve obtained by plotting integrated values;

[0034] FIG. 15 is a flowchart illustrating Embodiment 4 of the present invention;

[0035] FIG. 16 is a diagram showing data obtained by a measurement in which an RF pulse width was varied;

[0036] FIG. 17 is a diagram showing data obtained after processing;

[0037] FIG. 18 is a diagram showing a curve obtained by plotting integrated values;

[0038] FIG. 19 is a flowchart illustrating Embodiment 5 of the present invention;

[0039] FIG. 20 is a diagram showing data obtained by a measurement in which an RF pulse width was varied;

[0040] FIG. 21 is a diagram showing data obtained after processing;

[0041] FIG. 22 is a diagram showing a curve obtained by plotting integrated values;

[0042] FIG. 23 is a flowchart illustrating Embodiment 6 of the present invention;

[0043] FIG. 24 is a graph showing a curve obtained by plotting peak top intensities;

[0044] FIG. 25 is a flowchart illustrating Embodiment 7 of the present invention;

[0045] FIG. 26 is a graph showing a curve obtained by plotting peak top intensities;

[0046] FIG. 27 is a flowchart illustrating Embodiment 8 of the present invention;

[0047] FIG. 28 is a graph showing a curve obtained by plotting peak top intensities;

[0048] FIG. 29 is a flowchart illustrating Embodiment 9 of the present invention;

[0049] FIG. 30 is a graph showing a curve obtained by plotting peak top intensities;

[0050] FIG. 31 is a flowchart illustrating Embodiment 10 of the present invention;

[0051] FIG. 32 is a graph showing a curve obtained by plotting peak top intensities;

[0052] FIG. 33 is a flowchart illustrating Embodiment 11 of the present invention;

[0053] FIGS. 34(a) and 34(b) illustrate a first specific example of Embodiment 11;

[0054] FIGS. 35(a) and 35(b) illustrate a second specific example of Embodiment 11; and

[0055] FIGS. 36(a) and 36(b) are diagrams illustrating a third specific example of Embodiment 11.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0056] The preferred embodiments of the present invention are hereinafter described. Of the following embodiments, Embodiments 1-5 pertain to a method of finding an optimum value of an RF pulse width using integrated values in the signal region. Embodiments 6-10 pertain to a method of finding an optimum value of an RF pulse width using peak top intensities of a signal. In the following description, the measurement conditions of Table 1 and the pulse sequence of FIG. 1(c) are used. Of course, the invention is not restricted to the measurement conditions of Table 1 and pulse sequence of FIG. 1(c). Embodiment 11 pertains to a general method of optimizing an NMR measurement condition. In the following description, the unit name “abn” in the various tables is identical with the unit name “au” and means a unitless intensity or magnitude.

[0057] FIGS. 5(a) and 5(b) are flowcharts illustrating Embodiment 1 of the present invention. FIG. 5(a) illustrates a procedure for finding an optimum value of an RF pulse width. FIG. 5(b) illustrates a procedure for finding an initial value in a DFP method (described later). Embodiment 1 is a method of finding an optimum RF pulse width by measuring the pulse width over the whole range.

[0058] In FIG. 5(a), in step 1, NMR measurements are performed while varying an RF pulse width from 0 to 70 μs in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. As a result, measurement data shown in FIG. 2 are obtained.

[0059] In step 2, the measurement data are Fourier-transformed and a plot is done along the values of the RF pulse width using integrated values in the signal region. In this way, a curve is created. This processing results in data as shown in FIG. 3. Finding the integrated values from a signal range from 4 to 5.5 ppm produces results listed in Table 2. Plotting the results gives rise to the curve indicated by the solid line in FIG. 6.

TABLE 2
Results 1 of calculations of integrated values
Index	x_90_width	Intensity
1	0 μs	−3.9417	kabn
2	2 μs	50.35382	Mabn
3	4 μs	96.74335	Mabn
4	6 μs	126.74086	Mabn
5	8 μs	134.78419	Mabn
6	10 μs	119.99778	Mabn
7	12 μs	84.80437	Mabn
8	14 μs	37.11067	Mabn
9	16 μs	−14.37721	Mabn
10	18 μs	−60.44845	Mabn
11	20 μs	−93.60701	Mabn
12	22 μs	−106.62882	Mabn
13	24 μs	−98.13594	Mabn
14	26 μs	−70.18078	Mabn
15	28 μs	−28.46294	Mabn
16	30 μs	18.92854	Mabn
17	32 μs	61.74452	Mabn
18	34 μs	93.01865	Mabn
19	36 μs	105.83569	Mabn
20	38 μs	98.68427	Mabn
21	40 μs	72.78549	Mabn
22	42 μs	34.93452	Mabn
23	44 μs	−7.80636	Mabn
24	46 μs	−45.95444	Mabn
25	48 μs	−76.29246	Mabn
26	50 μs	−90.08089	Mabn
27	52 μs	−84.6685	Mabn
28	54 μs	−62.78815	Mabn
29	56 μs	−28.93477	Mabn
30	58 μs	10.14414	Mabn
31	60 μs	47.04411	Mabn
32	62 μs	73.08897	Mabn
33	64 μs	85.13906	Mabn
34	66 μs	80.43171	Mabn
35	68 μs	60.11846	Mabn
36	70 μs	29.17292	Mabn

[0060] In step 3, cure fitting described later) is performed in which the constants A, B, C, D, and ω of model equation (1) (given later) are varied such that model equation (1) agrees with the curve indicated by the solid line in FIG. 6. Model equations (1) (Eq. (1)), Eq. (2) for finding a 360° RF pulse width, and Eq. (3) for finding a 90° RF pulse width are given below. 1$\begin{array}{cc}y=A\sin \left(\omega x+B\right)\exp \left(-\frac{t}{C}\right)+D& \left(1\right)\\ PW360=\frac{2\pi }{\omega }& \left(2\right)\\ PW90=\frac{PW360}{4}& \left(3\right)\end{array}$

[0061] where t is the RF pulse width, A, B, C, D, and ω are constants, and y is the theoretical value of the intensity at t.

[0062] Curve fitting using model equation (1) produces the results shown in Table 3.

TABLE 3
Results 1 of calculations of curve fitting
	Constant	Value
	A	120149.01181	kabn
	B	0	rad
	C	207.85512	μs
	D	5381.92103	kabn
	ω	0.21696	rad/μs
	σ	9792.45099	kabn

[0063] Plotting of the values of Table 3 using model equation (1) produces the curve indicated by the broken line in FIG. 6, thus obtaining the constant values A, B, C, D, and ω and their standard deviation σ. Substituting the results into Eq. (2) gives rises to 360° pulse width (PW360)=28.96043 μs. This is substituted into Eq. (3), resulting in PW90=7.24011 μs that is an optimum RF pulse width. The standard deviation σ in Table 3 gives an index of the reliability of the obtained RF pulse width.

[0064] Curve fitting is to find the A, B, C, D, and ω which minimize the following evaluation formula (4) using a multi-dimensional variable metric method (Davidon-Fletcher-Powell (DFP) method) described in “Numerical Recipes in C: THE ART OF SCIENTIFIC COMPUTING”, Second Edition (ISEN 0-521-43108-5), 1992, pp. 425-430. The DFP method needs an evaluation equation and initial values. Eq. (4) using Eq. (5) identical with Eq. (1) is employed as this evaluation formula. 2$\begin{array}{cc}DIFF=\sum _{i=\mathrm{PWstart}}^{\mathrm{PWend}}{\left\{g\left(t\right)-f\left(t\right)\right\}}^2& \left(4\right)\\ f\left(t\right)=A\sin \left(\omega x+B\right)\exp \left(-\frac{t}{C}\right)+D& \left(5\right)\end{array}$

[0065] In Eq. (4), t is an RF pulse width, PWstart is the start value of the RF pulse width, PWend is the end value of the width, and g(t) is the actually measured value of the intensity at the RF pulse width of t. The theoretical value f(t) of intensity is defined by Eq. (5) and equal to the right side of Eq. (1).

[0066] A method of finding the initial values used in the DFP method is next described with reference to FIG. 5(b). In step 1, the initial values of ω and C of model equation (1) are computed using the LPSVD (Linear Prediction Singular Value Deconvolution) method that indicates a method of finding the initial values in the Journal of Magnetic Resonance, 61, 1985, pp. 465-481. If calculations using the LPSVD method are successful, initial values ω=0.21703 rad/μs and C=231.59569 μs are obtained.

[0067] In step 2, if calculations using the LPSVD method are unsuccessful, the program goes to step 3. If the calculations are successful, the program goes to step 7.

[0068] In step 3, the number n, of passes of the curve indicated by the solid line in FIG. 6 across straight line y=0 and the RF pulse widths zp[1], . . . , zp[n1] at this time are found. The results are shown in Table 4.

TABLE 4
Intersection 1 with y = 0
Item  Value
n1    5
zp[1] 0.00016 μs
zp[2] 15.44153 μs
zp[3] 29.20118 μs
zp[4] 43.63471 μs
zp[5] 57.48084 μs

[0069] In step 4, the results shown in Table 4 are substituted into Eq. (6). As an initial value of ω, we obtain 3$\begin{array}{cc}\omega =\frac{\pi \times \left(n_1-1\right)}{zp\left[n_1\right]-zp\left[1\right]}& \left(6\right)\end{array}$

[0070] In step 5, the number of positive-going (upward) peaks n2 of the curve indicated by the solid line in FIG. 6, the RF pulse widths pp[1], . . . , pp[n2] at the tops of the peaks, and the intensities pi[1], . . . , pi[n2] of the tops of the peaks are found. The results are listed in Table 5.

TABLE 5
Tops 1 of peaks
Item  Value
n2    3
pp[1] 8 μs
pp[2] 36 μs
pp[3] 64 μs
pi[1] 134.78419 kabn
pi[2] 105.83569 kabn
pi[3] 85.13906 kabn

[0071] In step 6, the results shown in Table 5 are substituted into Eq. (7). We obtain 121.90106 μs as an initial value of C (C=121.90106 μs). 4$\begin{array}{cc}C=\frac{pp\left[n_2\right]-pp\left[1\right]}{\ln \left(\frac{pi\left[1\right]}{pi\left[n_2\right]}\right)}& \left(7\right)\end{array}$

[0072] In step 7, the values of ω and C calculated in step 1 or steps 4 and 6 are substituted into Eq. (8) that is an expansion of Eq. (5). Using each item as a basis function and employing the curve indicated by the solid line in FIG. 6, the linear least squares method shown in “Numerical Recipes in C: THE ART OF SCIENTIFIC COMPUTING”, Second Edition (ISEN 0-521-43108-5), 1992, pp. 671-681 is implemented. The results (optimum constant values D, E, and F) listed in Table 6 are obtained. Table 6 uses the results of step 1. 5$\begin{array}{cc}y=E\sin \left(\omega x\right)\exp \left(-\frac{t}{C}\right)+F\cos \left(\omega x\right)\exp \left(-\frac{t}{C}\right)+D& \left(8\right)\end{array}$ A={square root}{square root over (E2+F2)}  (9)

TABLE 6
Results 1 of calculations of linear least squares method
	Item	Value
	D	5381.95203	kabn
	E	119968.83441	kabn
	F	−6577.86175999999	kabn

[0073] In step 8, the results shown in Table 6 are substituted into calculation formulas (9), (10-1), and (10-2), resulting in A=120149.03035 kabn and B=0.05477 rad. 6$\begin{array}{cc}\cos \left(B\right)=\frac{E}{\sqrt{E^2+F^2}}& \left(10\text{-}1\right)\\ \sin \left(B\right)=\frac{F}{\sqrt{E^2+F^2}}& \left(10\text{-}2\right)\end{array}$

[0074] Then, ω and C calculated in step 1 or steps 4 and 6, D calculated in step 7, and A and B calculated in step 8 are used as initial values in the DFP method. These values are varied. The values of A, B, C, D, and ω that minimize the value of evaluation formula (4) are found. Where the initial values of the DFP method are used, if measurement conditions as listed in Table 1 and the pulse sequence shown in FIG. 1(c), for example, are used, it is known that B=0 rad. Therefore, the DFP method is so implemented that the constants A, C, D, and ω are varied while the constant B of model equation (1) is kept at 0 rad.

[0075] In the present embodiment, the step of obtaining an optimum RF pulse width contains no manual operation and so reproducible results can be obtained. In the above description, the number of data items is set to the number of data items used in the prior art procedure for convenience of illustration. In curve fitting, in a case where there are data points characterizing a waveform, if the number of data points is reduced, similar results can be obtained. Therefore, good results can be obtained if there are a reduced number of data items. This similarly applies to embodiments described later.

[0076] Embodiment 2 of the present invention is described next with reference to FIG. 7. Embodiment 2 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the pulse width in the neighborhood of 360° pulse width. In the following description, it is assumed that it is previously known that the 360° pulse width is 28 μs because of the technique described as the prior art or information obtained from another technique.

[0077] In step 1, NMR measurements are performed while varying an RF pulse width from 24 to 34 μs (i.e., around 360° pulse width) in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 8 are obtained.

[0078] In step 2, the measurement data are Fourier-transformed, and integrated values in the signal region are plotted along the values of the RF pulse width. In this way, a curve is created. This processing results in data shown in FIG. 9. Finding the integrated values from a signal range from 4 to 5.5 ppm produces results listed in Table 7. Plotting the results gives rise to the curve indicated by the solid line in FIG. 10.

TABLE 7
Results 2 of calculations of integrated values
	Index	x_90_width	Intensity
	1	24 μs	−49.11459 Mabn
	2	26 μs	−35.15117 Mabn
	3	28 μs	−14.27532 Mabn
	4	30 μs	9.45965 Mabn
	5	32 μs	30.95036 Mabn
	6	34 μs	46.61799 Mabn

[0079] In step 3, the linear least squares method is implemented using model equation (11). The results shown in Table 8 are obtained. Plotting of the contents of Table 8 using model equation (11) results in the curve indicated by the broken line in FIG. 10.

y=At+B   (11)

TABLE 8
Results 2 of calculations of curve fitting
	Constant	Value
	A	10.01003	Mabn/μs
	B	−292.20986	Mabn
	σ	2.68816	Mabn

[0080] Then, the values in Table 8 are substituted into Eq. (12), producing PW360=29.19169 μs as a 360° pulse. This is substituted into Eq. (3), giving rise to an optimum RF pulse width PW90=7.29792 μs. The standard deviation σ in Table 8 gives an index of the reliability of the obtained RF pulse width. 7$\begin{array}{cc}\mathrm{PW}360=-\frac{B}{A}& \left(12\right)\end{array}$

[0081] In the present embodiment, the step of obtaining the RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0082] Embodiment 3 of the present invention is next described with reference to FIG. 11. Embodiment 3 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the pulse width in the neighborhood of 90° pulse width. In the following description, it is assumed that it is previously known that the 90° pulse width is 7 μs because of the technique described as the prior art or information obtained from another technique.

[0083] In step 1, NMR measurements are performed while varying an RF pulse width from 4 to 12 μs (i.e., around 90° pulse width) in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 12 are obtained.

[0084] In step 2, the measurement data are Fourier-transformed, and integrated values in the signal region are plotted along the values of the RF pulse width. In this way, a curve is created. This processing results in data shown in FIG. 13. Finding the integrated values from a signal existing region from 4 to 5.5 ppm produces results listed in Table 9. Plotting the results gives rise to the curve indicated by the solid line in FIG. 14.

TABLE 9
Results 3 of calculations of integrated values
	Index	x_90_width	Intensity
	1	4 μs	48.48201 Mabn
	2	6 μs	63.54521 Mabn
	3	8 μs	67.35346 Mabn
	4	10 μs	60.167 Mabn
	5	12 μs	42.88578 Mabn

[0085] In step 3, the linear least squares method is implemented using model equation (13). The results shown in Table 10 are obtained. Plotting of the contents of Table 10 using model equation (13) results in the curve indicated by the broken line in FIG. 14.

y=At 2 +Bt+C   (13)

TABLE 10
Results 3 of calculations of curve fitting
	Constant	Value
	A	−1.35149	Mabn/μs2
	B	20.89533	Mabn/μs
	C	−13.36858	Mabn
	σ	0.18738	Mabn

[0086] Then, the values in Table 10 are substituted into Eq. (12), producing PW90=7.73047 μs as an optimum RF pulse width. The standard deviation σ in Table 10 gives an index of the reliability of the obtained RF pulse width. 8$\begin{array}{cc}\mathrm{PW}90=-\frac{B}{2A}& \left(14\right)\end{array}$

[0087] In the present embodiment, the step of obtaining the RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0088] Embodiment 4 of the present invention is next described with reference to FIG. 15. Embodiment 4 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the pulse width in the neighborhood of 180° pulse width. In the following description, it is assumed that it is previously known that the 180° pulse width is 14 μs because of the technique described as the prior art or information obtained from another technique.

[0089] In step 1, NMR measurements are performed while varying an RF pulse width from 12 to 18 μs (i.e., around 180° pulse width) in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 16 are obtained.

[0090] In step 2, the measurement data are Fourier-transformed, and integrated values in the signal region are plotted along the values of the RF pulse width. In this way, a curve is created. This processing results in data shown in FIG. 17. Finding the integrated values from a signal existing range from 4 to 5.5 ppm produces results listed in Table 11. Plotting the results gives rise to the curve indicated by the solid line in FIG. 18.

TABLE 11
Results 4 of calculations of integrated values
	Index	x_90_width	Intensity
	1	12 μs	42.88578 Mabn
	2	14 μs	18.76097 Mabn
	3	16 μs	−7.33509 Mabn
	4	18 μs	−30.59374 Mabn

[0091] In step 3, the linear least squares method is implemented using model equation (11). The results shown in Table 12 are obtained. Plotting of the contents of Table 12 using model equation (11) results in the curve indicated by the broken line in FIG. 19.

TABLE 12
Results 4 of calculations of curve fitting
	Constant	Value
	A	−12.32673	Mabn/μs
	B	190.83045	Mabn
	σ	0.66926	Mabn

[0092] Then, the values in Table 12 are substituted into Eq. (14), producing PW90=7.74051 μs as an optimum RF pulse width. The standard deviation a in Table 12 gives an index of the reliability of the obtained RF pulse width. In the present embodiment, the step of obtaining the RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0093] Embodiment 5 of the present invention is next described by referring to FIG. 19. Embodiment 5 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the pulse width in the neighborhood of 270° pulse width. In the following description, it is assumed that it is previously known that the 270° pulse width is 21 μs because of the technique described as the prior art or information obtained from another technique.

[0094] In step 1, NMR measurements are performed while varying an RF pulse width from 18 to 26 μs (i.e., around 270° pulse width) in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 20 are obtained.

[0095] In step 2, the measurement data are Fourier-transformed, and integrated values in the signal region are plotted along the values of the RF pulse width. In this way, a curve is created. This processing results in data shown in FIG. 21. Finding the integrated values from a signal existing region 4 to 5.5 ppm produces results listed in Table 13. Plotting the results gives rise to the curve indicated by the solid line in FIG. 22.

TABLE 13
Results 5 of calculations of integrated values
	Index	x_90_width	Intensity
	1	18 μs	−30.59374 Mabn
	2	20 μs	−47.13169 Mabn
	3	22 μs	−53.43675 Mabn
	4	24 μs	−49.11459 Mabn
	5	26 μs	−35.15117 Mabn

[0096] In step 3, the linear least squares method is implemented using model equation (13). The results shown in Table 14 are obtained. Plotting of the contents of Table 14 using model equation (13) results in the curve indicated by the broken line in FIG. 22.

TABLE 14
Results 5 of calculations of curve fitting
	Constant	Value
	A	1.27911	Mabn/μs2
	B	−56.83557	Mabn/μs
	C	577.97661	Mabn
	σ	0.12472	Mabn

[0097] Then, the values in Table 14 are substituted into Eq. (15), producing PW90=7.40563 μs as an optimum RF pulse width. The standard deviation σ in Table 14 gives an index of the reliability of the obtained RF pulse width. 9$\begin{array}{cc}\mathrm{PW}90=-\frac{B}{6A}& \left(15\right)\end{array}$

[0098] In the present embodiment, the step of obtaining the RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0099] Embodiment 6 of the present invention is described with reference to FIG. 23. Embodiment 6 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the pulse width in a range from 0 to 360° and using peak top intensities of a signal.

[0100] In step 1, NMR measurements are performed while varying an RF pulse width from 0 to 70 μs in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 2 are obtained.

[0101] In step 2, the measurement data are Fourier-transformed, and the top intensities of maximum peaks in the signal region are plotted along the values of the RF pulse width. This processing results in data as shown in FIG. 3. Finding the maximum peak top intensities in the signal existing region from 4 to 5.5 ppm produces the results listed in Table 15. Plotting the results gives rise to the curve indicated by the solid line in FIG. 24.

TABLE 15
Results 1 of calculations of peak top intensities
Index	x_90_width	Intensity
1	0 μs	−686.46028	Mabn
2	2 μs	1.432	Mabn
3	4 μs	2.91449	Mabn
4	6 μs	4.23296	Mabn
5	8 μs	5.27167	Mabn
6	10 μs	5.79026	Mabn
7	12 μs	5.3959	Mabn
8	14 μs	2.91626	Mabn
9	16 μs	−2.4162	Mabn
10	18 μs	−4.82156	Mabn
11	20 μs	−5.28899	Mabn
12	22 μs	−4.76431	Mabn
13	24 μs	−3.77162	Mabn
14	26 μs	−2.49942	Mabn
15	28 μs	−1.0764	Mabn
16	30 μs	491.43929	Mabn
17	32 μs	1.87553	Mabn
18	34 μs	3.17844	Mabn
19	36 μs	4.16982	Mabn
20	38 μs	4.73415	Mabn
21	40 μs	4.49485	Mabn
22	42 μs	2.71549	Mabn
23	44 μs	−1.14449	Mabn
24	46 μs	−3.60914	Mabn
25	48 μs	−4.42519	Mabn
26	50 μs	−4.20325	Mabn
27	52 μs	−3.41928	Mabn
28	54 μs	−2.343	Mabn
29	56 μs	−1.08743	Mabn
30	58 μs	339.72675	Mabn
31	60 μs	1.5567	Mabn
32	62 μs	2.65856	Mabn
33	64 μs	3.52513	Mabn
34	66 μs	3.94934	Mabn
35	68 μs	3.63341	Mabn
36	70 μs	2.05146	Mabn

[0102] In step 3, a curve fitting DFP method is performed in which the constants A, B, C, D, and ω of model equation (1) are varied such that model equation (1) agrees with the curve. Consequently, the results shown in Table 16 are obtained.

TABLE 16
Results 6 of calculations of curve fitting
	Constant	Value
	A	5780247.97088	abn
	B	0	rad
	C	165.63334	μs
	D	12119.0114	abn
	ω	0.21641	rad/μs
	σ	964882.75674	abn

[0103] Plotting of the values of Table 16 using model equation (1) produces the curve indicated by the broken line in FIG. 24. Substituting the values in Table 16 into Eq. (2) gives rises to PW360=29.03433 μs. This is substituted into Eq. (3), resulting in PW90=7.25858 μs as an optimum RF pulse width. The standard deviation σ in Table 16 gives an index of the reliability of the obtained RF pulse width.

[0104] The method of finding the initial values of A, B, C, D, and ω in the curve fitting DFP method is the same as the method illustrated in FIG. 5(b). Accordingly, the procedure is described with reference to the flowchart of FIG. 5(b).

[0105] In step 1, the LPSVD method is used but calculations are unsuccessfully performed with no results. The program goes to step 3 where the number of times n, that the curve indicated by the solid line in FIG. 24 crosses straight line y=0 is found. Also, the RF pulse widths zp[1], . . . , zp[n1] produced at this time are found. The results shown in Table 17 are obtained.

TABLE 17
Intersection 2 with y = 0
Item  Value
n1    5
zp[1] 0.00096 μs
zp[2] 15.09378 μs
zp[3] 29.3731 μs
zp[4] 43.407 μs
zp[5] 57.52391 μs

[0106] In step 4, the results shown in Table 17 are substituted into Eq. (6). As an initial value of ω, we obtain ω=0.21846 rad/μs.

[0107] In step 5, the number of positive-going (upward) peaks n2 of the curve indicated by the solid line of FIG. 24, the RF pulse widths pp[1], . . . , pp[n2] at the tops of the peaks, and the intensities pi[1], . . . , pi[n2] of the tops of the peaks are found. The results listed in Table 18 are obtained.

TABLE 18
Tops 2 of peaks
Item  Value
n2    3
pp[1] 10 μs
pp[2] 38 μs
pp[3] 66 μs
pi[1] 5.79026 Mabn
pi[2] 4.73415 Mabn
pi[3] 3.94934 Mabn

[0108] In step 6, the results shown in Table 18 are substituted into Eq. (7). We obtain C=146.35619 μs as an initial value of C.

[0109] In step 7, the values of ω and C calculated in steps 4 and 6 are substituted into Eq. (8). Using each item as a basis function and employing the curve indicated by the solid line in FIG. 24, the linear least squares method is implemented. The results (optimum constant values D, E, and F) listed in Table 19 are obtained.

TABLE 19
Results 2 of calculations of linear least squares method
	Item	Value
	D	46209.1885	abn
	E	5919590.1939	abn
	F	−796281.22262	abn

[0110] In step 8, the results shown in Table 19 are substituted into Eqs. (9), (10-1), and (10-2), resulting in A=5972906.48254 abn and B=0.13371 rad.

[0111] The values calculated in this way are taken as initial values in the DFP method. These values are varied. The values of A, B, C, D, and ω that minimize the value of evaluation formula (4) are found. Where the initial values of the DFP method are used, in a case where measurement conditions as listed in Table 1 and the pulse sequence shown in FIG. 1(c) are used, for example, it is known that B=0 rad. Therefore, the DFP method is so implemented that the constants A, C, D, and ω are varied while the constant B of model equation (1) is kept at 0 rad.

[0112] In the present embodiment, the step of obtaining an optimum RF pulse width contains no manual operation and so reproducible results can be obtained.

[0113] Embodiment 7 of the present invention is described by referring to FIG. 25. Embodiment 7 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the RF pulse width in the neighborhood of 360° pulse width. In the following description, it is assumed that it is previously known that the 360° pulse width is 28 μs because of the technique described as the prior art or information obtained from another technique.

[0114] In step 1, NMR measurements are performed while varying an RF pulse width from 24 to 34 μs (i.e., around 360° pulse width) in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 8 are obtained.

[0115] In step 2, the measurement data are Fourier-transformed, and the top intensities of maximum peaks in the signal region are plotted along the values of the RF pulse width. This processing results in data as shown in FIG. 9. Finding the top intensities of maximum peaks in a signal existing region from 4 to 5.5 ppm produces results listed in Table 20. Plotting the results gives rise to the curve indicated by the solid line in FIG. 26.

TABLE 20
Results 2 of calculations of peak top intensities
Index	x_90_width	Intensity
1	24 μs	−5.96076	Mabn
2	26 μs	−3.78569	Mabn
3	28 μs	−1.6393	Mabn
4	30 μs	873.45073	kabn
5	32 μs	2.8408	Mabn
6	34 μs	4.99155	Mabn

[0116] In step 3, the linear least squares method is implemented using model equation (11). The results shown in Table 21 are obtained. Plotting of the contents of Table 21 using model equation (11) results in the curve indicated by the broken line in FIG. 26.

TABLE 21
Results 7 of calculations of curve fitting
	Constant	Value
	A	1.1022	Mabn/μs
	B	−32.41038	Mabn
	σ	0.11171	Mabn

[0117] Then, the values in Table 21 are substituted into Eq. (12), producing PW360=29.40524 μs. This is substituted into Eq. (3), giving rise to PW90=7.35131 μs as an optimum RF pulse width. The standard deviation σ in Table 21 gives an index of the reliability of the obtained RF pulse width.

[0118] In the present, the step of obtaining an optimum RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0119] Embodiment 8 of the present invention is next described with reference to FIG. 27. Embodiment 8 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the RF pulse width in the neighborhood of 90° pulse width and by using the peak top intensities of the signal. In the following description, it is assumed that it is previously known that the 90° pulse width is 7 μs because of the technique described as the prior art or information obtained from another technique.

[0120] In step 1, NMR measurements are performed while varying an RF pulse width from 4 to 12 μs (i.e., around 90° pulse width) in increments of 2 μs, using the measurement conditions show in Table 1 and pulse sequence shown in FIG. 1(c). The measurement data shown in FIG. 12 are obtained.

[0121] In step 2, the measurement data are Fourier-transformed, and the top intensities of the maximum peaks in the signal region are plotted along the values of the RF pulse width. This processing results in data shown in FIG. 13. Finding the top intensities of the maximum peaks in a signal existing range from 4 to 5.5 ppm produces results listed in Table 22. Plotting the results gives rise to the curve indicated by the solid line in FIG. 28.

TABLE 22
Results 3 of calculations of peak top intensities
Index	x_90_width	Intensity
1	4 μs	4.27768	Mabn
2	6 μs	6.43286	Mabn
3	8 μs	8.34666	Mabn
4	10 μs	9.39558	Mabn
5	12 μs	10.03132	Mabn

[0122] In step 3, the linear least squares method is implemented using model equation (13). The results shown in Table 23 are obtained. Plotting of the contents of Table 23 using model equation (13) results in the curve indicated by the broken line in FIG. 28.

TABLE 23
Results 8 of calculations of curve fitting
	Constant	Value
	A	−0.069719	Mabn/μs2
	B	1.83886	Mabn/μs
	C	−1.99495	Mabn
	σ	0.06976	Mabn

[0123] Then, the values in Table 23 are substituted into Eq. (14), producing PW90=13.18934 μs as an optimum RF pulse width. The standard deviation σ in Table 12 gives an index of the reliability of the obtained RF pulse width.

[0124] In the present embodiment, the step of obtaining the optimum RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0125] Embodiment 9 of the present invention is described by referring to FIG. 29. Embodiment 9 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the RF pulse width in the neighborhood of 180° pulse width and using the peak top intensities of the signal. In the following description, it is assumed that it is previously known that the 180° pulse width is 14 μs because of the technique described as the prior art or information obtained from another technique.

[0126] In step 1, NMR measurements are performed while varying an RF pulse width from 12 to 18 μs (i.e., around 180° pulse width) in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 16 are obtained.

[0127] In step 2, the measurement data are Fourier-transformed and the top intensities of the maximum peaks in the signal region are plotted along the values of the RF pulse width. This processing results in data as shown in FIG. 17. Finding the peak top intensities of the maximum peaks in a signal existing range from 4 to 5.5 ppm produces results listed in Table 24. Plotting the results gives rise to the curve indicated by the solid line in FIG. 30.

TABLE 24
Results 4 of calculations of peak top intensities
Index	x_90_width	Intensity
1	12 μs	10.03132	Mabn
2	14 μs	7.52695	Mabn
3	16 μs	−6.83314	Mabn
4	18 μs	−9.34702	Mabn

[0128] In step 3, the linear least squares method is implemented using model equation (11). The results shown in Table 25 are obtained. Plotting of the contents of Table 25 using model equation (11) results in the curve indicated by the broken line in FIG. 30.

TABLE 25
Results 9 of calculations of curve fitting
	Constant	Value
	A	−3.62476	Mabn/μs
	B	54.71586	Mabn
	σ	3.05991	Mabn

[0129] Then, the values in Table 25 are substituted into Eq. (14), producing PW90=7.54752 μs as an optimum RF pulse width. The standard deviation σ in Table 25 gives an index of the reliability of the obtained RF pulse width. In the present embodiment, the step of obtaining the optimum RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0130] Embodiment 10 of the present invention is described by referring to FIG. 31. Embodiment 10 is a method of finding an optimum RF pulse width by performing NMR measurements while varying the RF pulse width in the neighborhood of 270° pulse width and by using the peak top intensities of the signal. In the following description, it is assumed that it is previously known that the 270° pulse width is 21 μs because of the technique described as the prior are or information obtained from another technique.

[0131] In step 1, NMR measurements are performed while varying an RF pulse width from 18 to 26 μs (i.e., around 270° pulse width) in increments of 2 μs, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c), for example. The measurement data shown in FIG. 20 are obtained.

[0132] In step 2, the measurement data are Fourier-transformed, and the top intensities of the maximum peaks in the signal region are plotted along the values of the RF pulse width. This processing results in data as shown in FIG. 21. Finding the top intensities of the maximum peaks in a signal existing region from 4 to 5.5 ppm produces results listed in Table 26. Plotting the results gives rise to the curve indicated by the solid line in FIG. 32.

TABLE 26
Results 5 of calculations of peak top intensities
Index	x_90_width	Intensity
1	18 μs	−9.34702 Mabn
2	20 μs	−9.03598 Mabn
3	22 μs	−7.89824 Mabn
4	24 μs	−5.96076 Mabn
5	26 μs	−3.78569 Mabn

[0133] In step 3, the linear least squares method is implemented using model equation (13). The results shown in Table 27 are obtained. Plotting of the contents of Table 27 using model equation (13) results in the curve indicated by the broken line in FIG. 32.

TABLE 27
Results 10 of calculations of curve fitting
	Constant	Value
	A	0.08085	Mabn/μs2
	B	−2.84765	Mabn/μs
	C	15.66296	Mabn
	σ	0.09848	Mabn

[0134] Then, the values in Table 27 are substituted into Eq. (15), producing PW90=5.87 μs as an optimum RF pulse width. The standard deviation σ in Table 27 gives an index of the reliability of the obtained RF pulse width.

[0135] In the present embodiment, the step of obtaining the optimum RF pulse width does not use visual estimation. Therefore, reproducible results can be obtained.

[0136] Embodiment 11 of the present invention is next described with reference to FIG. 33. Embodiment 11 is a general method of optimizing NMR measurement conditions. Its specific examples are shown in FIGS. 34-36.

[0137] In step 1, NMR measurements are performed while varying a measurement condition to be optimized as shown in Table 28, using the measurement conditions shown in Table 1 and pulse sequence shown in FIG. 1(c).

TABLE 28
Candidates for measurement condition
Name of measurement condition    Definition
RF pulse width                   Time width of certain RF pulse
40                               Intensity of certain RF pulse
                                 Phase of certain RF pulse
RF pulse shape                   Shape of certain RF pulse
RF offset                        Frequency of RF pulse
FG pulse width                   Time width of certain magnetic field gradient pulse
FG pulse intensity               Intensity of certain magnetic field gradient pulse
FG pulse shape                   Shape of certain magnetic field gradient pulse
Direction of application of FG pulse Sense of certain magnetic field gradient pulse in
                                 absolute coordinate system
Wait time                        Wait time between certain pulses
Repeat delay                     Repetition time
Acquisition time                 Observation time
Scan                             Number of repetitions
Sweep width                      Width of observe frequency
Data points                      Number of sampling points of observed data
Total measurement time           Time from start to end of measurement
Lock offset                      Frequency of lock signal
Angle of axis of rotation of sample Angle between axis of rotation of sample tube and
                                 static magnetic field in measurement of sample being
                                 rotated
Sample temperature               Temperature of sample under measurement
Pressure                         Pressure inside sample tube
Spin rate                        Spin rate of sample tube
Volume                           Volume of sample
pH                               pH of sample
Concentration                    Concentration of sample
Position                         Position of sample tube
Static magnetic field intensity  Intensity of static magnetic field on sample
Flow rate                        Flow rate of sample through cell in LC-NMR probe
Q value                          Q value of transmit/receive circuit of probe
Shim value                       Current value of shim coil
Room temperature                 Temperature of measuring chamber
Coil length                      Length of probe coil
Filter characteristics           Parameters determining characteristics of digital and
                                 analog filters
Length of cable                  Length of cable used in portions associated with NMR
                                 measurement such as NMR spectrometer and probe
Thickness of cable               Thickness of cable used in portions associated with
                                 NMR measurement such as NMR spectrometer and
                                 probe

[0138] In step 2, the measurements data are Fourier-transformed. Then, properties as shown in Table 29 are extracted as numerical values. A curve is created by plotting them along the value varied in step 1.

TABLE 29
Candidates for certain property
Property of signal  Definition
Area                Integrated value of signal region
Intensity of vertex Intensity of peak top
Position of vertex  Position of peak top
Half-height width   Line width at half-peak position
Symmetry            Left-and-right symmetry, e.g., area ratio of left
                    and right halves on the opposite sides of a peak top

[0139] In step 3, curve fitting is done by varying constants of a model equation as shown in Table 30 such that the model equation coincident with the parameter varied in step 1, its range, and the property extracted in step 2 agrees with the curve created in step 2. In Table 30, x represents the measurement condition varied on the horizontal axis of the graph, and y represents the horizontal axis of the graph and indicates the value of intensity extracted from measurement results under certain measurement conditions. A, B, C, D, and E are constants, and they are varied such that the standard deviation σ decreases during curve fitting.

TABLE 30
Candidates for model equation
Model equation                   Definition
y = Ax + B                       Used when the graph varies
                                 linearly
y = Ax2 + Bx + C                 Used when the graph varies
                                 quadratically
y = Ax3 + Bx2 + Cx + D           Used when the graph varies
                                 cubically
10y=A⁢ ⁢sin⁢ ⁢(Bx+C)⁢exp⁡(-xD⁢ )+E Used when the graph varies periodically
y = −Acos(2x + B){1 − cos(2x + B)}2 + C Used when the graph varies
                                 periodically

[0140] Values of the constants and their standard deviation are obtained by this curve fitting. From these results, an optimum value of a measurement condition as shown in Table 31 is obtained. The standard deviation σ gives an index of the reliability of the obtained optimum value.

TABLE 31
Candidates for optimum value
Optimum value             Definition
Maximum x                 Value of x when the intensity of graph (value on vertical axis)
                          maximizes
Minimum x                 Value of x when the intensity of graph (value on vertical axis)
                          minimizes
x equal to 0              Value of x when the intensity of graph (value on vertical axis)
                          approaches closest to 0
Results of calculation of Value obtained from calculation formula, using an optimum
constant value            constant value derived as a result of curve fitting as described in
                          Embodiments 1-10

[0141] In the present embodiment, the step of obtaining an optimum value of a measurement condition does not use visual estimation. Therefore, reproducible results can be obtained.

[0142] FIGS. 34(a) and 34(b) illustrate a first specific example of Embodiment 11. This first specific example is a method of optimizing the irradiation center frequency in an NMR measurement.

[0143] In step 1, NMR measurements are performed using measurement conditions shown in Table 32 and pulse sequence shown in FIG. 34(a), for example. The center frequency of irradiation is varied about the frequency of signal derived from light water. Plural sets of 1D NMR measurement data are obtained.

TABLE 32
Measurement conditions for optimization
of center frequency of irradiation
Item                         Value
Sample                       1 mM albumen lysozyme/90% H20,
                             10% D2O
Magnetic field intensity     14.09636928 T
Observed nucleus             1H
Observe frequency            600.1723046 MHz
Observation center frequency Center frequency of irradiation
Number of data points        16384
Sweep width                  9.00252071 kHz
Number of accumulations      1
Pulse width                  1 μs
B1 pulse intensity           25 kHz
Saturation power             0.25 kHz
Saturation delay             2 s
Observation time             1.81993472 s
relaxation_delay             1 s
Temperature                  25° C.

[0144] In step 2, the residual signal derived from light water and indicating the 1D NMR data is Fourier-transformed and integrated values are plotted against the center frequency of irradiation. As a result, a curve as shown in FIG. 34(b) is obtained.

[0145] In step 3, the linear least squares method using model equation (16) from which a graph is created is implemented to obtain optimum A, B, and C. Since the optimum center frequency of irradiation is the value of x at the minimum value of the graph of FIG. 34(b), the constants A and B obtained by curve fitting are substituted into Eq. (17), thus obtaining optimum values. In the present specific example, the pulse sequence is not limited to the sequence shown in FIG. 34(a).

y=Ax 2 +Bx+C   (16) 11 $$ optimum ⁢   ⁢ value = - B 2 ⁢ A ( 17 )

[0146] where x is the center frequency of the irradiation and y indicates the integrated value of the residual signal derived from light water at the center frequency x of irradiation.

[0147] FIGS. 35(a) and 35(b) illustrate a second specific example of Embodiment 11. A method of optimizing the wait time δ in 15N—1H HSQC is described.

[0148] In step 1, ID NMR measurements are performed using measurement conditions as shown in Table 33 and pulse sequence shown in FIG. 35(a), for example, see Chemical Physics Letters, 69, 1980, pp. 185-188, and using varying wait time δ. Thus, plural sets of 1D NMR measurement data are obtained.

TABLE 33
Measurement conditions for optimization
of wait time δin 15N−1H HSQC
Item                             Value
Sample                           1 mM 15N labeled ubiquitin/90% H20,
                                 10% D2O
Magnetic field intensity         14.09636928 T
Observed nucleus                 1H
Irradiating nucleus              15N
Observe frequency                600.1723046 MHz
Observation center frequency     about 4.7 ppm (resonance frequency
                                 of water)
Number of data points            512
Sweep width                      9.00252071 kHz
Number of accumulations          2
Pulse width                      10 μs
1HB1 pulse intensity             25 kHz
15NB1 pulse intensity            6.25 kHz
15N decoupled B1 pulse intensity 0.78 kHz
t1                               1 μs
Observation time                 57 ms
relaxation_delay                 1 s
Temperature                      25° C.

[0149] In step 2, a certain signal indicating 1D NMR data and derived by Fourier transform is integrated. The integrated value is plotted against wait time δ. As a result, a curve as shown in FIG. 35(b) is obtained.

[0150] In step 3, the linear least squares method is implemented using model equation (16) from which a graph is created to obtain optimum A, B, and C. Since the optimum wait time δ is the value of x at the minimum value of the graph of FIG. 35(b), the constants A and B obtained by curve fitting are substituted into Eq. (17), thus obtaining optimum values. In this case, in Eq. (16), x indicates the wait time δ and y indicates the integrated value of signal at the wait time x. Also, in the present specific example, the pulse sequence is not limited to the sequence shown in FIG. 35(a).

[0151] FIGS. 36(a) and 36(b) illustrate a third specific example of Embodiment 11. This third example is a method of optimizing magnetic field gradient pulse intensity Gz2 in a 15N—1H HSQC (heteronuclear single quantum coherence) measurement (hereinafter simply referred to as SE-HSQC) using coherence selection utilizing magnetic field gradient pulses.

[0152] In step 1, NMR measurements are performed using measurement conditions as shown in Table 34 and pulse sequence shown in FIG. 36(a), for example, see Journal of the American Chemical Society, 114, 1992, pp. 10663-10665, and using varying magnetic field gradient pulse intensity Gz2. Thus, plural sets of 1D NMR measurement data are obtained.

TABLE 34
Measurement conditions for optimization
of magnetic field gradient pulse intensity
Gz2 in SE-HSQC measurement
Item                             Value
Sample                           0.1 mM 13C labeled methanol 0.1 mM
                                 15N labeled urea/dimethyl sulfoxide-d6
Magnetic field intensity         14.09636928 T
Observed nucleus                 1H
Irradiating nucleus              15N
Observe frequency                600.1723046 MHz
Observation center frequency     about 4.7 ppm (resonance frequency
                                 of water)
Number of data points            512
Sweep width                      9.00252071 kHz
Number of accumulations          2
Pulse width                      10 μs
1HB1 pulse intensity             25 kHz
15NB1 pulse intensity            6.25 kHz
15N decoupled B1 pulse intensity 0.78 kHz
Gz1                              0.25 T/m
Observation time                 57 ms
relaxation_delay                 1 s
Temperature                      25° C.

[0153] In step 2, 1D NMR data are obtained by Fourier transform and an integrated value derived from a certain signal contained in the data is obtained. The integrated value is plotted against Gz2. As a result, a curve as shown in FIG. 36(b) is obtained.

[0154] In step 3, the linear least squares method is implemented using model equation (16) from which a graph is created to obtain optimum A, B, and C. Since the optimum magnetic field gradient pulse intensity Gz2 is the value of x at the maximum value of the graph of FIG. 36(b), the constants A and B obtained by curve fitting are substituted into Eq. (17), thus obtaining optimum values. In this case, in Eq. (16), x indicates the Gz2 and y indicates the integrated value of signal at the magnetic field gradient pulse intensity x. Also, in the present specific example, the pulse sequence is not limited to the sequence shown in FIG. 36(a).

[0155] As described so far, according to the present invention, the step of obtaining an optimum value of a measurement condition in an NMR measurement does not include visual estimation. Therefore, reproducible results can be obtained. Furthermore, the use of a curve fitting method produces good results even if there are a limited number of data items, as long as they characterize a waveform. Consequently, a reliable optimum value of the measurement condition can be found in a short time.

[0156] Having thus described our invention with the detail and particularity required by the Patent Laws, what is desired protected by Letters Patent is set forth in the following claims.

Claims

1. A method of optimizing an NMR measurement condition, comprising the steps of: gaining NMR measurement data while varying the value of the NMR measurement condition to be optimized; extracting a certain property as a numerical value from the measurement data; plotting the numerical value of the measurement data versus the varied value of the measurement condition to create a curve; setting a model equation coincident with the measurement condition, its range, and the certain property extracted as a numerical value; performing curve fitting in which constants of the model equation are varied such that the model equation agrees with the created curve; and gaining constant values of the model equation and their standard deviation by curve fitting and obtaining an optimum value of the measurement condition from the gained constant values and standard deviation.

2. A method of optimizing an NMR measurement condition as set forth in claim 1, wherein said measurement condition is an RF pulse width, and wherein said certain property extracted as a numerical value is an integrated value of a region in which a signal obtained by Fourier-transforming measurement data obtained using a given pulse sequence is present.

3. A method of optimizing an NMR measurement condition as set forth in claim 1, wherein said measurement condition is an RF pulse width, and wherein said certain property extracted as a numerical value is a peak top intensity of a region in which a signal obtained by Fourier-transforming measurement data obtained using a given pulse sequence is present.

4. A method of optimizing an NMR measurement condition as set forth in any one of claims 2 and 3, wherein said measurement data are obtained by varying the RF pulse width in the neighborhood of total range of RF pulse width, 360° pulse width, 90° pulse width, 180° pulse width, or 270° pulse width.

5. A method of optimizing an NMR measurement condition as set forth in claim 1, wherein: (A) said measurement condition is a center frequency of irradiation, (B) said certain property extracted as a numerical value is an integrated value of a signal obtained by Fourier-transforming measurement data obtained using a given pulse sequence, and (C) a quadratic function is used as said model equation.

6. A method of optimizing an NMR measurement condition as set forth in claim 1, wherein: (A) said measurement condition is a wait time in 15N—1H HSQC, (B) said certain property extracted as a numerical value is an integrated value of a signal obtained by Fourier-transforming measurement data obtained using a given pulse sequence, and (C) a quadratic function is used as said model equation.

7. A method of optimizing an NMR measurement condition as set forth in claim 1, wherein: (A) said measurement condition is a magnetic field gradient pulse intensity in a 15N—1H HSQC measurement using coherence selection that employs magnetic field gradient pulses, (B) said certain property extracted as a numerical value is an integrated value of a signal obtained by Fourier-transforming measurement data obtained using a given pulse sequence, and (C) a quadratic function is used as said model equation.