JP5395888B2 - Image generation apparatus, image generation system, and image generation method - Google Patents

Description

  The present invention relates to an optical tomographic image generation apparatus and an optical tomographic image generation method.

  Currently, an imaging apparatus (hereinafter also referred to as an OCT apparatus) using an optical coherence tomography (OCT) using interference by low-coherence light has been put into practical use. In this case, since a tomographic image can be acquired with a depth resolution of several micrometers, a tomographic image of the object to be inspected can be taken at a high resolution.

  Patent Document 1 discloses an optical image measurement device that improves the image quality of a formed image. In this apparatus, a plurality of tomographic images of the fundus are formed and the formed images are stored. A new tomographic image is formed by calculating using one of the tomographic images and a tomographic image adjacent thereto. As a result, the image quality of the formed image can be improved.

JP 2008-237238 A

  However, even if a final tomographic image is formed by forming a plurality of tomographic images as in Patent Document 1 and then calculating using the pixel values of these tomographic images, the noise component is not efficiently removed. Therefore, the improvement in image quality may be limited.

  The present invention has been made to address the above-described problems, and an object thereof is to further improve the image quality of a tomographic image by effectively removing noise components.

  When performing calculations using a plurality of tomographic images as in Patent Document 1, the present inventors have conventionally performed the calculation after converting the pixel values of the tomographic images into real numbers, so noise removal is efficiently performed. I found that I couldn't do it. Therefore, the present invention was completed by verifying that noise is more efficiently removed by calculating as a complex number and then converting to a real number.

In other words, the image generating apparatus of the present invention includes means for obtaining a plurality of complex data by Fourier transforming each of a plurality of frames of signals obtained by irradiating the inspection object with light, and the plurality of complex data. Means for converting to a plurality of real number data; means for determining whether or not there is a measurement error for the plurality of frames using the plurality of real number data; and the measurement error of the plurality of frames. Means for averaging data of a plurality of complex numbers corresponding to a plurality of frames other than the frame determined to be .
In addition, the image generation system of the present invention includes means for obtaining a plurality of complex number data by Fourier-transforming signals of a plurality of frames obtained by irradiating light on the object to be inspected, and the plurality of complex number data. Means for converting to a plurality of real number data; means for determining whether or not there is a measurement error for the plurality of frames using the plurality of real number data; and the measurement error of the plurality of frames. Means for averaging data of a plurality of complex numbers corresponding to a plurality of frames other than the frame determined to be .

Further, the image generation method of the present invention includes a step of obtaining a plurality of complex number data by performing Fourier transform on each of a plurality of frame signals obtained by irradiating light on the object to be inspected, and the plurality of complex number data. Converting to a plurality of real number data; determining whether or not there is a measurement error for the plurality of frames using the plurality of real number data; and measuring error among the plurality of frames And averaging a plurality of complex data corresponding to a plurality of frames other than the frame determined to be .

  According to the present invention, further improvement in image quality can be expected by effectively removing noise components.

The figure explaining the signal processing in Example 1 of this invention. The figure explaining the Michelson type OCT apparatus in Example 1 of this invention. The figure explaining the time chart in Example 1 of this invention. The figure explaining the Mach-Zehnder type OCT apparatus in Example 2 of this invention. The figure explaining the signal processing in Example 2 of this invention.

  In an optical tomographic image generation method for generating a tomographic image of an inspection object, an embodiment according to the present invention is a first step of acquiring a plurality of frames of signals obtained by irradiating the inspection object with light, A second step of Fourier-transforming signals of a plurality of frames to obtain complex number data, a third step of combining the plurality of frames with complex numbers using the respective complex number data, and the synthesized data And a fourth step of generating a tomographic image. As shown in FIGS. 1 and 5, as a fifth step, a frame movement amount may be calculated before the signals are combined, and a plurality of frames may be combined based on the information.

Example 1
Next, Example 1 of the present invention will be described. In the present embodiment, a tomographic image is generated by an imaging apparatus using a Michelson interferometer, but the imaging apparatus that can be used in the present invention is not limited to this. Further, the signal processing of the present embodiment is characterized in that the movement amount is calculated based on real number data, unlike the second embodiment described later.

(Michelson interferometer)
An optical coherence tomographic imaging apparatus (hereinafter also referred to as an OCT apparatus) according to the first embodiment will be described with reference to FIG. FIG. 2 is a schematic diagram for explaining an imaging apparatus using the Michelson-type optical system (Michelson interferometer) of this embodiment.

  Light emitted from the light source 201 is split into measurement light 214 and reference light 213 by a beam splitter 204 via a fiber 202 and a lens 203-1. The measurement light 214 is incident on the eye 217 that is an object to be inspected via the XY scanner 208 and the objective lenses 205-1 and 205-2. Then, the measurement light 214 incident on the eye passes through the cornea 216 and further reaches the retina 218.

  The return light 215 reflected and scattered by the retina 218 of the eye 217 returns in the order of the objective lenses 205-2 and 205-1, the XY scanner 208, and the beam splitter 204. Further, the light is guided to the spectroscope 211 via the lens 203-2. The spectroscope 211 includes a lens, a grating, an image sensor, and the like. A CCD or CMOS type line sensor is used as the image sensor. A signal obtained by the line sensor of the spectroscope 211 is transmitted to the computer 212 and stored in the memory, and processing described later is performed.

  On the other hand, the reference light 213 is reflected by the reference mirror 209 via the dispersion compensation glass 207, passes through the dispersion compensation glass 207 again, and returns to the beam splitter 204. The dispersion compensation glass 207 is for compensating dispersion of the eye 217 and the objective lenses 205-1 and 205-2. The reference mirror 209 can adjust the optical path length of the reference optical path by the mirror adjustment mechanism 210. These reference light 213 and return light 215 are combined by the beam splitter 204. Then, this combined light is guided to the spectroscope 211 and detected. In the measurement optical path, the point where the optical path length matches the reference optical path is called a coherence gate. When measuring the retina 218 of the eye 217, the position of the reference mirror 209 is adjusted so that this coherence gate is close to the retina 218.

  As the light source 201, an SLD (Super Luminescent Diode) which is a typical low coherent light source is used. The wavelength is, for example, a center wavelength of 840 nm and a bandwidth of 50 nm. The bandwidth is an important parameter because it affects the resolution in the optical axis direction of the obtained tomographic image. Although the type of the light source 201 is SLD here, it is only necessary to emit low-coherent light, and ASE (Amplified Spontaneous Emission) or the like can also be used. Of course, other light sources such as a halogen lamp may be used depending on the contents of the inspection object. However, since the wavelength affects the resolution in the horizontal direction of the obtained tomographic image, it is desirable that the wavelength is short when importance is attached to the resolution in the horizontal direction.

  The computer 212 controls the spectroscope 211, the XY scanner 208, the mirror adjustment mechanism 210, and the focus adjustment mechanism 206 in addition to performing arithmetic processing and control described later. Of course, the computer 212 can also perform data input, image processing, image display, data storage, and the like.

(Signal processing process)
The signal processing executed by the OCT apparatus shown in FIG. 2 will be described with reference to FIG.
Measurement is started in step A1. In this state, the OCT apparatus is activated and the eye to be examined is placed. Furthermore, adjustment necessary for the measurement is performed by the operator, and the measurement is started.

  In step A2, a signal is acquired. Here, a case where a three-dimensional image is acquired by irradiating the retina 218 of the fundus oculi with the measurement light 214 and two-dimensionally scanning the retina 218 with the XY scanner 208 will be described. The X direction perpendicular to the optical axis of the eye in FIG. 2 is Fast Scan, and the Y direction is Slow Scan. FIG. 3 (a) shows an X position, FIG. 3 (b) shows a Y position, and FIG. 3 (c) shows a timing chart showing photographing timing. When photographing continuously 512 times in the X direction while fixing the Y position, if the number of pixels of the line sensor in the spectroscope 211 is 1024, data of one frame of 1024 × 512 elements is obtained. An XZ plane tomographic image can be obtained from one frame of data. Furthermore, while moving 500 steps in the Y direction, continuous shooting in the X direction is repeated for each step, resulting in 500 frames of data. FIG. 3 shows a state in which the first frame 301, the second frame 302, and the third frame 303, which are three frames to be combined, are acquired. The time between frames is used to move the scanner in the X direction and Y direction to capture the next position and to transfer data to the computer 212.

  Of course, the scan pattern is not limited, and a plurality of frames having the same Y position may be combined and imaged. Further, it may be data acquired while rotating as in a circle scan as well as movement in a linear direction such as the X direction and the Y direction.

  In the present embodiment, data having different positions in the Y direction between frames is acquired. Here, when data having a large positional difference between acquired frames is combined, signal components may be lost (data to be combined cancel each other and the original signal components are lost). Therefore, it is desirable that the difference in position is several times or less the lateral resolution of the OCT apparatus (generally determined by the beam diameter of the measurement light on the object to be measured). In the case of the present embodiment, if the beam diameter of the measurement light on the object to be measured is about 5 micrometers, the difference in position is about several tens of micrometers in the Y direction.

  In step A3, wavelength wave number conversion is performed. In general, data from the spectroscope 211 includes a wavelength and an intensity at the wavelength. The wavelength is sampled at equal intervals. First, a function of intensity data with respect to wavelength is created. Next, each wavelength is converted into a wave number to create a function of intensity data. Since the wave number is the reciprocal of the wavelength, a new wave number is assigned so that the 1024 wave numbers are equally spaced. Then, intensity data corresponding to the wave number is calculated. The calculation method is, for example, interpolation, and may be general linear interpolation, spline interpolation, or the like. However, it is desirable that it is a linear operation. As a result, for each frame, a two-dimensional array of 1024 × 512 elements composed of equally spaced wave numbers and intensities for the wave numbers is obtained. Naturally, when the spectroscope 211 can sample at equal intervals with respect to the wave number, or when the error due to wavelength wave number conversion does not become a problem, the step A3 can be skipped.

  Fourier transform is performed in step A4. Here, intensity data equally spaced with respect to the wave number is subjected to discrete Fourier transform for each column. As a result, a 1024 × 512 complex two-dimensional array is obtained in each frame. However, due to the nature of the Fourier transform, the m-th row and the (1024-m) -th row in each column have the same intensity. Therefore, lines 0 to 511 are extracted to obtain a two-dimensional array of 512 × 512 complex numbers, and data is sent to the next process.

In step A5, a movement amount between frames for combining signals is calculated. Here, the movement amount calculation of the images of the second frame 302 and the third frame 303 with respect to the image of the first frame 301 will be described. First, the 512 × 512 complex data of each frame is converted to a real number. Here, with respect to the 128th column (384 elements of 65 to 448 rows) of the first frame 301, the i column (512 elements) of the image of the second frame 302 is selected, and continuous from j rows therein. A one-dimensional array is obtained by calculating a difference from 384 elements. The elements of the one-dimensional array are square-averaged. This is performed for the necessary i and j, and i ₁ and j ₁ are determined in which the mean square is minimized. Next, the same operation is performed on the 384th column (384 elements in 65th to 448th rows) of the first frame 301 to determine i ₂ and j _{2 at} which the mean square is minimized. As a result, the movement amount of the image of the second frame 302 is obtained with respect to the image of the first frame 301. If the movement amount is zero, i ₁ = 128, j ₁ = 65, i ₂ = 384, and j ₂ = 65, respectively. Further, the movement amount of the image of the third frame 303 is obtained with respect to the image of the first frame 301. Then, information on the calculated movement amount is sent to the process M.

  Note that this method may be repeated to increase the accuracy of the information regarding the movement amount, or each column may be interpolated to calculate the movement amount in the subpixel. Of course, other methods may be used as the moving amount calculation method.

  In step M, signals are synthesized. The addition average is performed on the complex two-dimensional data obtained in step A4 based on the information regarding the movement amount from step A5. The complex number data of the second frame 302 as a result of translation and rotation are added to the complex number data of the first frame 301. Of course, since the arrangement is discrete, the necessary interpolation processing is applied to the image of the second frame 302. Similarly, the complex number data of the third frame 303 as a result of translation and rotation are added. Then, the synthesized complex number data is converted into a real number and sent to step A6.

  By the way, the average may be a weighted average, but if the weight is different, the noise removal effect may change. It should be noted that the frame data may be thinned if it is determined that a measurement error has occurred.

In step A6, the image is displayed on the display screen of the computer 212 as one tomographic image.
And it complete | finishes by process A7.
Here, description has been given from measurement using an OCT apparatus to image display. However, for example, an image with reduced noise can be obtained by applying the above processing to data of a plurality of frames obtained via a network.

  In this embodiment, when combining a plurality of frame data, the movement amount calculation is performed based on the real number data, and the combination is performed based on the complex number data, so that noise can be effectively removed and further image quality can be improved. Improvement can be expected.

(Description of principle)
The process M of the present embodiment is characterized in that it is converted to a real number after adding in a complex number state, compared to the conventional method of combining signals by adding images between frames after converting to a real number. These fundamental differences result in the difference between adding two complex numbers after conversion to real numbers and adding them before conversion to real numbers and then converting them to real numbers.

Here, using the imaginary unit i, the complex element 1 and the element 2 are expressed by Formula 1-1 and Formula 1-2, respectively.

The result obtained by adding element 1 and element 2 by complex numbers and then converting them to real numbers is expressed by Equation 2.

The result obtained by adding element 1 and element 2 to real numbers is expressed by Equation 3.

When the mathematical formulas 2 and 3 are squared and the common part is subtracted, the relationship of the mathematical formula 4 is obtained. (Further, you can easily prove it by squaring both sides.)

  That is, the value of Equation 2 is equal to or less than the value of Equation 3. This is important in the noise removal method. That is, in the case of random noise, there are plus and minus. Addition in the complex state cancels out the positive and negative components of the noise, so that averaging can be performed more efficiently. If linear transformation is used for each step of signal processing, the same result can be obtained even if they are overlapped with data before being converted into complex numbers.

  Table 1 is a comparison of signal to noise ratio (SNR). This result is slightly different from the result obtained by the process of this embodiment because the lines in the frame are synthesized as in Embodiment 2 described later. However, the data is sufficient to compare the SNR according to the timing at which the pixel value is converted to a real number. The subject to be imaged was measured in the range of about 6 mm of the retina centered on the macular of the normal eye. The number of lines is 2048 lines. When the same original data is used to (1) extract a tomographic image from 2048 lines every 4 lines to 512 lines, and generate a tomographic image that does not synthesize one frame, (2) 1024 lines are extracted every 2 lines to 1024 When two tones are combined to generate a 512-line one-frame tomographic image, (3) When 2048 lines are combined every four lines to generate a 512-line one-frame tomographic image It is. In each case, the SNR when the synthesis is performed before the conversion from the complex number to the real number is compared with the SNR when the synthesis is performed after the conversion to the real number. In the case of combining before conversion to a real number (corresponding to the frame combination with complex numbers in this embodiment), it can be seen that the SNR is improved as the number of combinations increases. On the other hand, it can be seen that the SNR is generally constant when synthesized after being converted to a real number. However, regardless of when the real number conversion is performed, a smoother image can be obtained by performing the combining process when comparing the case where the combining process (averaging) is not performed and the case where the combining process is performed after the real number. Here, the SNR is the ratio of the lowest value of the RMS (Root Mean Square) of noise in each row to the maximum value of each pixel.

(Example 2)
Next, a second embodiment of the present invention will be described. Here, differences from the first embodiment will be particularly described. In this embodiment, a tomographic image is generated by an imaging apparatus using a Mach-Zehnder interferometer, but the imaging apparatus that can be used in the present invention is not limited to this. In addition, the signal processing of this embodiment includes a sixth step of synthesizing the frame signals within the frame, and performing a movement amount calculation including the phase component. In addition, although the aspect containing both of these characteristics is described below, this invention is not limited to the following description, The aspect containing only one of the characteristics may be sufficient.

(Mach-Zehnder interferometer)
An imaging apparatus using the optical coherence tomography according to the second embodiment will be described with reference to FIG. FIG. 4 is a schematic diagram for explaining an imaging apparatus using the Mach-Zehnder type optical system of the present embodiment. In addition, the same number is attached | subjected to the same structure as FIG. 2, and the description is partially abbreviate | omitted. Light emitted from the light source 201 is split into measurement light 214 and reference light 213 through the fiber 202 and the fiber coupler 401-1.

  Measurement light 214 enters port 1 of circulator 402-2, exits port 2, and reaches lens 403-2. Furthermore, it reaches the retina 218 via the XY scanner 208, the objective lenses 205-1 and 205-2, and the cornea 216. The return light 215 scattered and reflected by the retina 218 returns to the objective lenses 205-2 and 205-1 and the XY scanner 208, enters the port 2 of the circulator 402-2, exits the port 3, and the fiber coupler 401- 2 is reached.

  On the other hand, reference light 213 enters port 1 of circulator 402-1 and exits from port 2, and is reflected by reference mirror 209 via lens 403-1 and dispersion compensation glass 207. The reflected reference light 213 returns to the lens 403-1 and the port 2 of the circulator 402-1 through the dispersion compensation glass 207, exits from the port 3, and reaches the fiber coupler 401-2. The reference mirror 209 can adjust the optical path length by the mirror adjustment mechanism 210. The reference light 213 and the return light 215 are combined by the fiber coupler 401-2, guided to the spectroscope 211, detected, and transmitted to the computer 212.

(Signal processing process)
The signal processing in the second embodiment executed by the OCT apparatus shown in FIG. 4 will be described with reference to FIG. The same reference numerals as those in FIG. 1 represent the same processing.
Measurement is started in step A1.
In step A2, a signal is acquired. Here, an image of 2048 lines is acquired for one frame, and a total of 500 frames are acquired. As a result, 1024 × 2048 element data is obtained per frame. Here, one of the features of the present embodiment is to synthesize signals within each frame.

  When a signal is synthesized within a frame and when a signal is synthesized between frames, the noise removal effect is the same in principle if the object to be inspected does not move. However, it is not simple when the object to be inspected moves like an eye. If the line rate is 20 kHz, the time required to acquire one frame of data is 26 msec for 512 lines and 102 msec for 2048 lines. When 4 msec which is the return time of the scanner is added, the frame intervals are about 30 msec and 106 msec, respectively. That is, when the adjacent lines in the frame are combined, the movement amount calculation is not necessary. On the other hand, when the measurement time of the frame becomes long, the probability that the distortion in the frame or the loss of data due to the eye closing will occur increases. Also in this embodiment, for the same reason as in the first embodiment, it is desirable that the difference in the position of the acquired line is several times the beam diameter or less.

In step S, adjacent lines in the frame are synthesized. The data is synthesized every four adjacent 2048 lines, and data of 1024 × 512 elements is obtained in each frame.
In step A3, wavelength wave number conversion is performed.
Fourier transform is performed in step A4. Then, by extracting necessary portions, two-dimensional array data of 512 × 512 complex numbers is obtained in each frame, and the data is sent to the next step.
The movement amount is calculated in step A5-1. Here, the movement amount is calculated in the complex state in the present embodiment, whereas the movement amount is calculated by converting the first embodiment into a real number. That is, when calculating the data difference, it is performed in a complex state, and the case where the mean square of the result is minimized is extracted. As a result, it is possible to calculate the movement amount including the phase.
Note that complex data may be calculated after being converted into polar coordinates.
In step M, an addition average is performed on the complex two-dimensional array data obtained in step A4 based on information relating to movement including the phase obtained in step A5-1. Then, the result is converted to a real number and data is sent to step A6.
In step A6, one tomographic image is obtained.
The process ends at step A7.
In this embodiment, the movement amount calculation including the phase can be performed, and further improvement in image quality can be expected.

(Example 3)
The present invention can also be realized by executing the following processing. That is, software (computer program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and the computer of the system or apparatus (or CPU, MPU, etc.) reads the program. To be executed.
That is, the computer program, a first procedure for performing a first step of obtaining signals of a plurality of frames obtained by irradiating light to the inspection object, and Fourier transform the signals of the plurality of frames, A second procedure for performing a second step for obtaining complex data respectively; a third procedure for performing a third step for combining the plurality of frames with complex numbers using the respective complex data; And a fourth procedure for performing a fourth step of generating a tomographic image based on the synthesized data. The computer program further includes a fifth procedure for performing a fifth step of acquiring frame movement information and a sixth procedure for performing a sixth step of synthesizing the signals of the frame within the frame. You may prepare.

DESCRIPTION OF SYMBOLS 201 Light source 202 Optical fiber 203 Lens 204 Beam splitter 205 Objective lens 206 Focus adjustment mechanism 207 Dispersion compensation glass 208 XY scanner 209 Reference mirror 210 Mirror adjustment mechanism 211 Spectrometer 212 Computer 213 Reference light 214 Measurement light 215 Return light 216 Cornea 217 Eye 218 Retina 301 first frame 302 second frame 303 third frame 401 photocoupler 402 circulator 403 lens

Claims (23)

Means for obtaining a plurality of complex number data by Fourier-transforming signals of a plurality of frames obtained by irradiating the inspection object with light;
Means for converting the plurality of complex data to a plurality of real data;
Means for determining whether there is a measurement error for the plurality of frames using the plurality of real number data;
Means for averaging a plurality of complex data corresponding to a plurality of frames other than the frame determined to be the measurement error among the plurality of frames ;
An image generation apparatus comprising: Means for thinning out the frames determined to be the measurement error from the plurality of frames;
Said averaging means, the image generating apparatus according to claim 1, characterized in that averaging the data of a plurality of complex numbers corresponding to a plurality of frames after which the decimated. Said means for determining the image generating apparatus according omission of data by distortion in a frame in claim 1 or 2, characterized in that it is determined that the said measurement error. The inspected object is an eye;
The image generation apparatus according to claim 1, wherein the determination unit determines that the missing data due to closing eyes is the measurement error.   The averaging means averages the data of the plurality of complex numbers based on information of at least one of parallel movement and rotation movement between the plurality of frames. The image generating apparatus according to item 1. The inspected object is an eye;
The image generation apparatus according to claim 5, wherein the movement is caused by a fine movement of the eye.   The image generating apparatus according to claim 1, wherein the averaging unit performs either addition averaging or square averaging on the plurality of complex data. The inspected object is an eye;
The plurality of frames are obtained at different positions in the direction intersecting the optical axis of the eye,
The image generation apparatus according to claim 1, wherein the difference in position between the frames is several times less than a horizontal resolution. 9. The image generation apparatus according to claim 1, further comprising means for generating an image based on the averaged complex number data. The generating means generates a tomographic image as the image ,
The image generating apparatus according to claim 9 , further comprising means for displaying the tomographic image on a display device. Means for converting the averaged complex number data into a real number;
The image generating apparatus according to claim 9 , wherein the generating unit generates the data converted into the real number as the image. A program that causes a computer to be executed as each unit of the image generation apparatus according to any one of claims 1 to 11. Means for obtaining a plurality of complex number data by Fourier-transforming signals of a plurality of frames obtained by irradiating the inspection object with light;
Means for converting the plurality of complex data to a plurality of real data;
Means for determining whether there is a measurement error for the plurality of frames using the plurality of real number data;
Means for averaging a plurality of complex data corresponding to a plurality of frames other than the frame determined to be the measurement error among the plurality of frames ;
An image generation system comprising: Means for thinning out the frames determined to be the measurement error from the plurality of frames;
14. The image generating system according to claim 13, wherein the averaging means averages a plurality of complex data corresponding to the plurality of frames after the thinning. The image generation system according to claim 13 or 14, wherein the determination unit determines that data loss due to distortion in a frame is the measurement error. The inspected object is an eye;
The image generation system according to claim 13 or 14, wherein the determination unit determines that the missing data due to closing eyes is the measurement error. Means for transferring the plurality of complex data to a computer;
The image generation system according to any one of claims 13 to 16, wherein the averaging means averages the plurality of complex data after being transferred to a computer. A program that causes a computer to be executed as each unit of the image generation system according to any one of claims 13 to 17. A process of obtaining a plurality of complex data by Fourier-transforming each of a plurality of frames of signals obtained by irradiating the inspection object with light;
Converting the plurality of complex number data into a plurality of real number data;
Determining whether there is a measurement error for the plurality of frames using the plurality of real number data;
Averaging a plurality of complex data corresponding to a plurality of frames other than the frame determined to be the measurement error among the plurality of frames ;
An image generation method characterized by comprising: Thinning out frames determined to be the measurement error from the plurality of frames,
20. The image generation method according to claim 19, wherein in the averaging step, a plurality of complex data corresponding to the plurality of frames after the thinning are averaged. 21. The image generation method according to claim 19 or 20, wherein in the determination step, it is determined that the missing data due to distortion in a frame is the measurement error. The inspected object is an eye;
21. The image generation method according to claim 19 or 20, wherein, in the determination step, it is determined that the missing data due to closing eyes is the measurement error. A program that causes a computer to execute each step of the image generation method according to any one of claims 19 to 22 .

Images