Signal processing circuit, photodetector, and signal processing method

Abstract

The present invention provides a signal processing circuit, an optical detection device, and a signal processing method that can remove high-output noise while improving real-time performance and data accuracy. [Solution] The noise reduction system includes a processing unit that processes a target signal corresponding to the output signal from a vacuum tube sensor, which has a photocathode that emits electrons in response to incident light and outputs an output signal corresponding to the amount of incident light, and derives an output value for each predetermined measurement time. The processing unit is configured to perform abnormal value handling processing, which divides the measurement period into n (where n is an integer of 2 or more) intervals, derives the signal value in each of the n intervals from the target signal, and derives the output value for the measurement period from the signal values ​​in the n intervals in such a way that the magnitude of the largest maximum signal value among the signal values ​​does not affect the magnitude of the output value.

Description

[Technical Field]

[0001] The present invention relates to a signal processing circuit, a photodetector, and a signal processing method. [Background technology]

[0002] One type of optical sensor is the vacuum tube sensor, which has a photocathode that emits electrons in response to incident light, and detects light by multiplying the electrons emitted from the photocathode. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2015-141705 [Overview of the Initiative] [Problems that the invention aims to solve]

[0004] Vacuum tube sensors can be used, for example, in the light detection section of a laser scanning microscope. The inventors have found that in such cases, due to various noise components, abnormal values ​​with outputs larger than the normal values ​​due to the signal light may be detected, making it difficult to use the information at the time the abnormal output value was detected in the acquired image.

[0005] On the other hand, there is a noise reduction technique, for example, described in Patent Document 1. The method described in Patent Document 1 identifies pixels where noise has occurred in the acquired X-ray image and replaces the value of the pixel with an alternative value calculated from the values ​​of surrounding pixels, or an alternative value calculated from the values ​​of the pixel acquired at the time before and after the pixel. However, the method described in Patent Document 1 processes the data, which has been configured as image data, in the manner described above and reconstructs it into an image. Therefore, there were problems, especially in cases where high-speed processing of noise is required, such as in real-time observation.

[0006] Therefore, an object of the present invention is to provide a signal processing circuit, an optical detection device, and a signal processing method capable of performing high-speed processing on an event indicating an abnormal value.

Means for Solving the Problems

[0007] The signal processing circuit of the present invention is a signal processing circuit including a processing unit that processes a target signal corresponding to an output signal from a vacuum tube sensor having a photocathode that emits electrons in response to incident light and outputs an output signal corresponding to the amount of the incident light, and derives an output value for each predetermined measurement period. The processing unit divides the measurement period into n (n is an integer of 2 or more) intervals, derives signal values in each of the n intervals from the target signal, and when deriving the output value in the measurement period from the signal values in the n intervals, is configured to be able to execute an abnormal value corresponding process for deriving the output value from the signal values in the n intervals so that the magnitude of the maximum signal value, which is the largest among the signal values, does not affect the magnitude of the output value.

[0008] In this signal processing circuit, the processing unit divides the measurement period into n intervals, derives signal values in each of the n intervals from the target signal, and derives an output value in the measurement period from the signal values in the n intervals. And when the processing unit derives the output value in the measurement period, it is configured to be able to execute an abnormal value corresponding process for deriving the output value from the signal values in the n intervals so that the magnitude of the maximum signal value, which is the largest among the signal values in the n intervals, does not affect the magnitude of the output value. Thereby, it is possible to derive the output value in the measurement period after performing the abnormal value corresponding process as necessary. That is, since the correspondence to an event indicating an abnormal value is performed at the stage of numerical data called the output value rather than image data, high-speed processing is possible. Thereby, high-speed processing for an event indicating an abnormal value becomes possible.

[0009] The signal processing circuit of the present invention may be the signal processing circuit described in [2], "In the abnormal value corresponding process, the processing unit derives the average value of the signal values other than the maximum signal value among the signal values in the n intervals as the output value". In this case, the output value can be derived from the signal values in the n intervals so that the magnitude of the maximum signal value does not affect the magnitude of the output value.

[0010] The signal processing circuit of the present invention may be the signal processing circuit described in [3], "In the abnormal value corresponding process, the processing unit derives the median value of the signal values in the n intervals as the output value". In this case, the output value can be derived from the signal values in the n intervals so that the magnitude of the maximum signal value does not affect the magnitude of the output value.

[0011] The signal processing circuit of the present invention may be the signal processing circuit described in [4], "The signal processing circuit further includes a current-voltage conversion unit that converts a current signal corresponding to the amount of incident light into a voltage signal at a predetermined cut-off frequency, and the cut-off frequency is set so that the pulse width of the voltage signal corresponding to the maximum signal value is narrower than the width of each of the n intervals". In this case, it is possible to suppress the complexification of the abnormal value corresponding process due to an event indicating an abnormal value spanning multiple intervals.

[0012] The signal processing circuit of the present invention may be the signal processing circuit described in [5], "When the processing unit derives the output value during the measurement period, a first value based on the maximum signal value is the number of electrons P emitted from the photoelectric surface when the incident light corresponding to the maximum signal value is incident on the photoelectric surface" enoise and the fluctuation σ of the number of electrons, and the signal amount Q that appears in the target signal when one photoelectron is emitted from the photoelectric surface singleThe signal processing circuit may be one of the [1] to [4] described above, which executes the abnormal value handling process if the first value is greater than or equal to the second value based on the above, and executes a derivation process to derive the average value of the signal values ​​in the n intervals as the output value if the first value is smaller than the second value. In this case, if an event indicating an abnormal value occurs during the measurement period, the abnormal value handling process can be executed, and if no event indicating an abnormal value occurs, a derivation process to derive the average value of the signal values ​​in the n intervals as the output value can be executed.

[0013] The signal processing circuit of the present invention may also be [6] "the signal processing circuit according to [5], wherein the first value is the value obtained by subtracting from the maximum signal value any one of the following signal values: the smallest minimum signal value among the signal values ​​in the n intervals, the average value of the remaining signal values ​​in the n intervals excluding the maximum signal value, the median value of the remaining signal values ​​in the n intervals excluding the maximum signal value, or the maximum signal value." In this case, it is possible to suitably determine whether or not an event indicating an abnormal value occurred during the measurement period.

[0014] The signal processing circuit of the present invention is [7] "The second value is Q single (P enoise The signal processing circuit may be based on -mσ (where m is 2 or greater) as described in [5] or [6]. In this case, it is possible to suitably determine whether or not an event indicating an abnormal value occurred during the measurement period.

[0015] The signal processing circuit of the present invention may also be [8] "the signal processing circuit according to any one of [1] to [7], wherein the vacuum tube sensor further comprises an avalanche diode that multiplies electrons from the photocathode." In such a vacuum tube sensor, since a solid element such as an avalanche diode is used, it becomes easier to appropriately perform the above-mentioned abnormal value handling processing.

[0016] The photodetector of the present invention is [9] "a photodetector comprising the vacuum tube sensor and a signal processing circuit described in any of [1] to [8]". This photodetector makes it possible to suppress the impact on time response characteristics, improve real-time performance and data accuracy in responding to events showing abnormal values, and remove high-output noise.

[0017] The light detection device of the present invention may also be the "light detection device according to [9], wherein the vacuum tube sensor further comprises an avalanche diode that multiplies electrons from the photocathode." In such a vacuum tube sensor, a solid element such as an avalanche diode is used, making it easier to appropriately perform the above-mentioned abnormal value handling process.

[0018] The signal processing method of the present invention is

[11] "a signal processing method that includes processing a target signal corresponding to the output signal from a vacuum tube sensor having a photocathode that emits electrons in response to incident light and outputs an output signal corresponding to the amount of light of the incident light, and deriving an output value for each predetermined measurement period, wherein the measurement period is divided into n (n is an integer of 2 or more) intervals, a signal value in each of the n intervals is derived from the target signal, and when deriving the output value for the measurement period from the signal values ​​in the n intervals, the output value is derived from the signal values ​​in the n intervals in such a way that the magnitude of the largest maximum signal value among the signal values ​​in the n intervals does not affect the magnitude of the output value." According to this signal processing method, for the reasons described above, high-speed processing of events showing abnormal values ​​becomes possible. [Effects of the Invention]

[0019] According to the present invention, it is possible to provide a signal processing circuit, an optical detection device, and a signal processing method that can process events exhibiting abnormal values ​​at high speed. [Brief explanation of the drawing]

[0020] [Figure 1] This is a diagram showing the configuration of a laser scanning microscope. [Figure 2]This is a diagram illustrating the scanning of excitation light across a sample. [Figure 3] This is a cross-sectional view of a vacuum tube sensor. [Figure 4] This is a diagram illustrating the flow of signal processing. [Figure 5] This is a diagram to explain X-ray noise. [Figure 6] This figure shows an example of observed X-ray noise. [Figure 7] This diagram illustrates the method for calculating pixel values ​​in the example. [Figure 8] This is a diagram illustrating the method for calculating pixel values ​​in the embodiment. [Figure 9] This diagram illustrates the case of extremely low light levels. [Figure 10] This is a flowchart to explain the processing flow. [Figure 11] This is a diagram to explain the judgment process. [Figure 12] (a) and (b) are diagrams illustrating the decision-making process. [Figure 13] Figures (a) and (b) show examples of images with noise, with (a) being an example under extremely low illumination and (b) being an example under medium illumination. [Figure 14] Figures (a) and (b) show examples of images obtained when noise reduction processing is performed, with (a) being an example for extremely low illumination and (b) being an example for medium illumination. [Figure 15] Figures (a) and (b) show examples of images obtained when the judgment process is performed, with (a) being an example for extremely low illumination and (b) being an example for medium illumination. [Figure 16] This is a diagram illustrating the configuration of a modified laser scanning microscope. [Modes for carrying out the invention]

[0021] Embodiments of the present invention will be described in detail below with reference to the drawings. In the following description, the same or equivalent elements will be denoted by the same reference numerals, and redundant explanations will be omitted.

[0022] As shown in Figure 1, the laser scanning microscope 1 comprises a light source 2, a dichroic mirror 3, scanning mirrors 4 and 5, an objective lens 6, and a photodetector A (photodetector) including a vacuum tube sensor 7. The laser scanning microscope 1 is, for example, a confocal laser scanning microscope, which irradiates a sample S with excitation light L1 (irradiation light) output from the light source 2, and detects measurement light L2 generated in the sample S in response to the irradiation of excitation light L1 using the vacuum tube sensor 7. The sample S is the object to be observed and is, for example, placed on a stage. For example, the sample S is a cell, and the measurement light L2 is fluorescence (fluorescence signal).

[0023] Light source 2 is a point light source composed of, for example, a CW (Continuous Wave) laser, and outputs laser light as excitation light L1 (irradiation light). Dichroic mirror 3 is an optical element that reflects excitation light L1 while transmitting measurement light L2. Scanning mirrors 4 and 5 are mirrors for scanning the irradiation position of excitation light L1 on the sample S. Scanning mirrors 4 and 5 are, for example, rotatable around different axes of rotation, and scan the irradiation position of excitation light L1 along the XY plane (a plane perpendicular to the direction in which excitation light L1 is irradiated onto the sample S).

[0024] The objective lens 6 is positioned facing the sample S. The excitation light L1 output from the light source 2 is reflected sequentially by the dichroic mirror 3 and scanning mirrors 4 and 5, then focused by the objective lens 6 and irradiated onto the sample S. The measurement light L2 generated in the sample S passes through the objective lens 6, is reflected by the scanning mirrors 5 and 4, passes through the dichroic mirror 3, and enters the vacuum tube sensor 7 of the photodetector A. Between the vacuum tube sensor 7 and the dichroic mirror 3, a pinhole member 8 is positioned with a pinhole 8a formed therein to block the measurement light L2 that is generated and arrives outside the focal point. The vacuum tube sensor 7 is a photodetector that detects the measurement light L2. Details of the vacuum tube sensor 7 will be described later.

[0025] As shown in Figure 2, in the laser scanning microscope 1 of this embodiment, the observation surface of the sample S is virtually divided into multiple regions R arranged, for example, in a grid pattern, and measurements are performed while scanning the irradiation position of the excitation light L1 on the sample S so that these regions R are irradiated sequentially. The laser scanning microscope 1 outputs an image as an observation result, in which one pixel P corresponds to each region R, and the image is composed of multiple pixels P. Hereafter, "pixel" will refer to the pixel that makes up the observation image.

[0026] As shown in Figure 3, the vacuum tube sensor 7 is an electron tube comprising a housing portion 11 that defines the internal vacuum space, and a photocathode 12 and an electron detection unit 13 arranged within the housing portion 11. The electron detection unit 13 includes a semiconductor electron detection element, such as an avalanche diode. In this example, the vacuum tube sensor 7 is configured as a hybrid photodetector (HPD), an electron tube that combines the photocathode 12 and the electron detection unit 13, and increases the electron multiplication efficiency by accelerating electrons (photoelectrons) emitted from the photocathode 12 with a desired acceleration voltage and injecting them into the electron detection unit 13.

[0027] The housing 11 includes a stem 14, a side tube 15, a converging electrode 16, and an input panel 17. The stem 14 is formed, for example, in the shape of a disc and constitutes the bottom of the housing 11. The side tube 15 is formed in the shape of a cylinder, for example, made of ceramic, and is connected to the stem 14. The converging electrode 16 is formed in the shape of a cylinder, for example, made of a conductive material, and is connected to the side tube 15, forming a cylindrical tube together with the side tube 15. If the converging electrode 16 is not provided, the side tube 15 and the input panel 17 are directly connected.

[0028] The input panel 17 is formed, for example, in the shape of a disc and is positioned opposite the stem 14 to form the top of the housing 11. The input panel 17 is provided to cover the opening formed by the focusing electrode 16. The input panel 17 is made of a material that is transparent to the measurement light L2 (photons).

[0029] A photocathode 12 is formed on the inner surface 17a of the input panel 17. The inner surface 17a is the inner surface (stem 14 side) of the input panel 17. Measurement light L2 transmitted through the input panel 17 is incident on the photocathode 12. The photocathode 12 emits electrons (photoelectrons) E in response to the incident light (photons) of measurement light L2. Note that the incident light may also contain various X-rays as noise components, as described later. In this example, the photocathode 12 is a semiconductor photocathode formed from a material containing GaAsP, but it may also be a semiconductor photocathode containing other semiconductor materials, or an alkali photocathode containing alkali metal materials. During operation, an acceleration voltage of, for example, about -8kV is applied between the photocathode 12 and the electron detection unit 13. Electrons E emitted from the photocathode 12 are accelerated by the acceleration voltage between the photocathode 12 and the electron detection unit 13, and are focused by the focusing electrode 16 before being incident on the electron detection unit 13.

[0030] The electron detection unit 13 (electron multiplication unit) includes an avalanche diode, which is a semiconductor electron detection element. In this example, the avalanche diode is made of silicon. Electrons E emitted from the photocathode 12 are incident on the electron detection unit 13. When electrons E accelerated according to the acceleration voltage are incident on the electron detection unit 13, electron implantation multiplication occurs in accordance with the acceleration voltage, and the electrons E are multiplied (electron irradiation gain G1). Furthermore, the electrons E are further multiplied by avalanche multiplication that occurs inside the electron detection unit 13 according to the voltage applied to the avalanche diode (avalanche gain G2). In this way, the electron detection unit 13 multiplies the incident electrons E. During operation, a reverse bias voltage of, for example, about 500V is applied to the avalanche diode.

[0031] A connector section 18, for example, an SMA (Sub Miniature Type A) connector, is connected to the stem 14, and a current signal corresponding to the number of electrons E after multiplication by the electron detection section 13 is output from the connector section 18 to the outside of the vacuum tube sensor 7. The number of electrons E is proportional to the amount of light (number of photons) of the measurement light L2 incident on the photocathode 12. In other words, the vacuum tube sensor 7 outputs an output signal corresponding to the amount of light of the measurement light L2 incident on the photocathode 12.

[0032] As shown in Figure 4, the light detection unit A comprises a vacuum tube sensor 7, a signal processing circuit C, and a control unit 24. The signal processing circuit C comprises a current-voltage converter (current-voltage conversion unit) 21, an AD (analog-to-digital) converter 22, and a circuit unit 23. The current-voltage converter 21, the AD converter 22, and the circuit unit 23, as the signal processing circuit C, perform predetermined processing on the output signal from the vacuum tube sensor 7 and output it to the control unit 24. The control unit 24 performs image construction processing and other operations based on the input signal.

[0033] The current-voltage converter 21 is composed of, for example, a transimpedance amplifier (TIA amplifier). The current-voltage converter 21 receives an analog current signal output from the vacuum tube sensor 7. The current-voltage converter 21 converts this current signal into an analog voltage signal while increasing its signal strength.

[0034] The current-voltage converter 21 converts a current signal into a voltage signal at a predetermined cut-off frequency f c . Usually, there is a limit to the frequency that can be multiplied in a TIA amplifier circuit, and the gain decreases in the high-frequency range. The cut-off frequency is defined, for example, as a value at which the gain decreases by 3 dB from the normal value when exceeding the frequency. The cut-off frequency f c and the rise time τ r of the output waveform signal are known to have the relationship of the following formula (1). Although the value of the constant 0.35 in formula (1) may change slightly, there is an inverse proportional relationship between the cut-off frequency f c and the rise time τ r .

Equation

[0035] The AD converter 22 samples the analog voltage signal output from the current-voltage converter 21 at a predetermined sampling frequency S rThis converts it into a digital voltage signal. Circuit 23 is composed of an integrated circuit such as an FPGA (Field Programmable Gate Array). The target signal ST, which is a digital voltage signal output from the AD converter 22, is input to circuit 23. The target signal ST is a signal corresponding to the analog current signal (output signal) output from the vacuum tube sensor 7. In this example, circuit 23 functions as a processing unit 25 that processes the target signal ST. Hereafter, circuit 23 will also be referred to as processing unit 25. For example, processing unit 25 calculates (derives) the value of each pixel P in the observed image based on the target signal ST. Details of the processing by processing unit 25 will be described later.

[0036] The control unit 24 is composed of, for example, a computer equipped with a processor such as a CPU and a storage medium such as RAM or ROM. The control unit 24 functions as an image processing unit that constructs an observation image based on the values ​​of each pixel P output from the processing unit 25. The control unit 24 may also be given the function of a control unit that controls the operation of each part of the laser scanning microscope 1.

[0037] Figure 5 is a diagram illustrating X-ray noise, one of the various types of noise that exhibit abnormal values. X-ray noise is observed in the laser scanning microscope 1. As shown in Figure 5, when electrons E emitted from the photocathode 12 are incident on the electron detection unit 13, characteristic X-rays or bremsstrahlung X-rays may be emitted from the electron detection unit 13. Both are X-ray noise and can be addressed by the present invention, but for the sake of simplicity, only characteristic X-rays will be described below.

[0038] When characteristic X-rays are incident on the photocathode 12, many electrons are generated in the photocathode 12. Some of these electrons are annihilated within the photocathode 12, but most of the remaining electrons are emitted from the photocathode 12 toward the electron detection unit 13. In this case, when one characteristic X-ray is incident on the photocathode 12, not one electron is emitted as when a photon is incident, but rather, for example, dozens of electrons are emitted. When these electrons are incident on the electron detection unit 13, the output signal from the electron detection unit 13 increases, causing the value of pixel P to become excessively large, and in some cases, the pixel P may appear as a bright spot in the observed image. In the laser scanning microscope 1 according to the following embodiments, X-ray noise caused by such characteristic X-rays will be described as an event showing an abnormal value.

[0039] Figure 6 shows an example of X-ray noise observation. Figure 6 shows the pulse signal (single pulse) of the target to be measured (measurement signal) and the pulse signal (noise signal) of the X-ray noise to be removed, as measured by an oscilloscope. In Figure 6, the horizontal axis is time and the vertical axis is signal intensity (voltage). As shown in Figure 6, the intensity (pulse height) of the noise signal is much larger than the intensity (pulse height) of the measurement signal. The pulse width (full width at half maximum) of the noise signal is about twice that of the pulse width (full width at half maximum) of the measurement signal. Furthermore, the noise signal is measured almost simultaneously with the measurement signal. This is because the noise signal caused by characteristic X-rays emitted from the electronic detection unit 13 is measured almost simultaneously with the measurement signal, which is measured in response to the incidence of the measurement light L2. The inventors actually measured the noise signal while changing the amount of incident light L2 and obtained from the measurement results that the signal amount and generation probability of the noise signal are constant.

[0040] Figure 7 is a diagram illustrating the method for calculating pixel values ​​in a reference example. Figure 7 shows an example of measurement results when acquiring data corresponding to one pixel P, that is, one pixel that makes up an observed image. The measurement period shown in Figure 7 corresponds to the time that excitation light L1 is irradiated onto one pixel P. In the laser scanning microscope 1, this time is, for example, about 1 μs. Excitation light L1, which is CW light, is output from the light source 2. These points are the same in Figures 8, 9, and 11, which will be described later.

[0041] As shown in Figure 7, the sensor output from the vacuum tube sensor 7 includes a measurement signal. This "sensor output" corresponds to the "analog voltage signal" output from the current-voltage converter 21 in Figure 4. This measurement signal may be, for example, a single pulse or a signal consisting of a DC component. As shown in Figure 7, if X-ray noise occurs during the measurement period, a noise signal will appear in the sensor output from the vacuum tube sensor 7. In reality, the sensor output from the vacuum tube sensor 7 is output as a signal that combines the measurement signal and the noise signal, but for the sake of simplicity, the measurement signal and the noise signal are described separately. This is also true in Figures 8, 9, and 11, which will be discussed later.

[0042] In the example calculation method, a measurement window with a time width corresponding to the entire measurement period is set, and the output value during the measurement period (the value of one pixel P in the example of Figure 7) is calculated from the averaged measurement values ​​in the measurement window. In the example of Figure 7, for comparison with the embodiment described later, the measurement window is virtually divided into four sections of the same time width, and the signal values ​​in these sections are assumed to be 108, 8, 5, and 2 in order of earliest measurement time, similar to the embodiment described later. In this case, the section with a signal value of 108 is assumed to have a larger signal value compared to the other sections due to the influence of noise signals. These signal values ​​are calculated from the target signal ST, which is a digital voltage signal output from the AD converter 22 shown in Figure 4. In this case, if the average value per section is used in the four sections described above, the averaged measurement value in the measurement window becomes (108 + 8 + 5 + 2) / 4 = 30.8, and the value of pixel P becomes 30.8. Note that the value of pixel P may also be the sum of the measurement values ​​within the measurement period instead of the averaged value, in which case the value of pixel P becomes 123. Thus, in the calculation method of the example, if a noise signal occurs within the measurement window range, the noise signal has a significant impact on the value of pixel P. As a result, the value of pixel P may become excessive, and there is a risk that the pixel P will appear as a bright spot in the observed image. Furthermore, since the only discernible event is that the output signal has become larger, it is not possible to determine whether a noise signal has occurred or whether the signal value for each section has become larger overall (i.e., whether the actual amount of incident light itself has increased).

[0043] One way to address such problems is to perform measurements using a photon counting method. In this case, X-ray noise is also measured as a single photon, so the noise signal can be identified. However, because it is a digital judgment based on a set threshold, the dynamic range becomes low. Another way to address this is to set the acceleration voltage low and operate the vacuum tube sensor 7. In this case, the energy of the electrons is reduced to begin with, making it difficult to generate X-rays and thus X-ray noise. However, the internal gain becomes low, reducing the detection efficiency, and it becomes particularly difficult to observe single photons. Therefore, the laser scanning microscope 1 removes X-ray noise using the following method.

[0044] Figure 8 is a diagram illustrating the method for calculating pixel values ​​according to the embodiment. As shown in Figure 8, in the calculation method according to the embodiment, the measurement period (time per pixel P) is divided into n (where n is an integer of 2 or more) (4 in this example) intervals of the same time width, the signal value in each interval is calculated from the target signal ST, and the output value (value of pixel P) for the measurement period is calculated from the signal values ​​in the 4 intervals. This process is performed by the processing unit 25 (circuit unit 23) shown in Figure 4.

[0045] In other words, the processing unit 25 sets up four measurement windows, each having a time width corresponding to one of four sections obtained by dividing the measurement period into four parts, and calculates the value of pixel P from the signal values ​​in these measurement windows. The signal value in each section (measurement window) is, for example, the average value of the target signal ST in that section (the integral value divided by the time width). In the example in Figure 8, the signal values ​​in the four sections are 108, 8, 5, and 2 in order of earliest measurement time. The section with a signal value of 108 has a larger signal value compared to the other sections due to the influence of noise signals. These signal values ​​are calculated by the processing unit 25 from the target signal ST, which is a digital voltage signal output from the AD converter 22 shown in Figure 4.

[0046] The processing unit 25 performs noise reduction processing, which is one of the abnormal value handling processes, and calculates the average value of the remaining signal values ​​after excluding the largest maximum signal value among the signal values ​​in the four intervals as the output value (value of pixel P) for the measurement period (n-division filter processing). In the example in Figure 8, the average value of the remaining signal values ​​8, 5, and 2 after excluding the largest maximum signal value 108 among the four signal values ​​108, 8, 5, and 2 is (8+5+2) / 3=5, and the value of pixel P is 5. If the above noise reduction processing is not performed, the averaged measurement value in the measurement window will be (108+8+5+2) / 4=30.8. In other words, the value of pixel P will be 30.8, which is more than 6 times the value after noise processing. Thus, in the calculation method according to this embodiment, in order to ensure that the magnitude of the maximum signal value (108 in this case) does not affect the magnitude of the value of pixel P (5 in this case), the value of pixel P is calculated from the signal values ​​in the four intervals by specifically calculating the value of pixel P from the remaining signal values ​​after excluding the maximum signal value. This allows for the calculation of the pixel P value while excluding the effects of X-ray noise. Furthermore, if the same processing is performed for all measurement periods, the pixel P value may be the average value of the intervals, which is 5, as described above, or it may be the sum of the values ​​for the entire measurement period, which is 20 (5 × 4 intervals = 20). Note that in this specification, "removing noise" means removing at least a portion of the noise (reducing noise), and does not necessarily mean that all noise is completely removed.

[0047] Furthermore, in the laser scanning microscope 1, the cutoff frequency f of the current-voltage converter 21 is c For example, the bandwidth is widened to 5MHz or more. In this case, the pulse width W1 of the noise signal to be removed that appears in the target signal ST (sensor output) becomes narrower than the width W2 of each of the four sections. This makes it possible to suppress the noise signal to be removed from appearing across multiple sections, and X-ray noise can be effectively removed. As mentioned above, the cutoff frequency f c As it increases, the rise time τ r The frequency becomes shorter, and the response waveform becomes sharper. Cutoff frequency f cWhen the bandwidth is increased, not only does the pulse width of the measurement signal narrow, but the pulse width of the noise signal also narrows. Therefore, it becomes easier to keep noise pulses within a single interval. As a result, the difference between the interval containing noise pulses and the interval without noise pulses becomes larger, making it possible to determine whether a noise signal has been generated or whether the signal value for each interval has increased overall (i.e., whether the actual amount of incident light itself has increased). Therefore, especially in the case of observations where a predetermined amount of light is continuously incident, it becomes easier to determine whether or not noise reduction processing should be performed on the measurement period. In addition, the larger difference between the interval containing noise pulses and the interval without noise pulses makes it easier to identify the intervals that should be removed. Note that although the noise pulse appears within a single interval in Figure 8, in reality, noise pulses may appear across two intervals. In the noise reduction method according to the embodiment, at least a portion of the noise can be removed even in such cases.

[0048] On the other hand, when the intensity of the measurement light L2 is extremely low (in the case of extremely low illuminance), applying the calculation method according to the embodiment may result in the following problems. Figure 9 is a diagram illustrating the case of extremely low illuminance. As shown in Figure 9, in the case of extremely low illuminance, the intensity of the measurement signal is extremely small. In the example in Figure 9, no noise is generated, and the signal values ​​in the four intervals are 1, 0, 0, and 0 in order of earliest measurement time. In this example, even in the interval where the measurement signal was measured, the signal value is small, at 1. If the calculation method according to the embodiment described above is applied in this case, the average value of the remaining signal values, excluding the largest maximum signal value of 1 among the four signal values, becomes 0, and the value of pixel P becomes 0. Thus, when the calculation method according to the embodiment is applied in the case of extremely low illuminance, there is a risk that the measurement signal with extremely low intensity will not be reflected in the measurement result. However, since signals with extremely low intensity can be removed, it can function as a dark pulse filter to remove dark pulses.

[0049] Therefore, the following determination process may be used to determine whether or not noise occurred during the measurement period. If noise occurred during the measurement period, a noise reduction process may be performed. If no noise occurred, a normal calculation process (derivation process) may be performed to calculate the average value of the signal values ​​over n intervals as the output value. Figure 10 is a flowchart illustrating the processing flow. Figures 11 and 12 are diagrams illustrating the determination process. The processing in the laser scanning microscope 1 will be explained below in accordance with the flowchart in Figure 10.

[0050] First, the processing unit 25 receives a trigger from the control unit 24 to start the measurement period (pixel P) (step S1). Next, the processing unit 25 receives the target signal ST (digital voltage signal) for the measurement period from the AD converter 22 (step S2). In step S2, for example, the processing unit 25 receives digital voltage signals for a number of measurement points corresponding to the time width of the measurement period. Next, the processing unit 25 divides the measurement period into n intervals and calculates the signal value for each interval from the target signal ST (step S3).

[0051] Next, the processing unit 25 determines whether the determination formula is satisfied (step S4). The details of the determination formula will be described later. If the determination formula is satisfied (YES in step S4), the processing unit 25 performs noise reduction processing and calculates the average value of the remaining (n-1) signal values ​​after removing the largest maximum signal value from the n intervals as the output value (value of pixel P) (step S5). On the other hand, if the determination formula is not satisfied (NO in step S4), the processing unit 25 performs normal calculation processing and calculates the average value of the signal values ​​in the n intervals as the output value (value of pixel P) (step S6). After the execution of step S5 or S6, processing for the next measurement period (pixel P) is performed.

[0052] The determination formula is expressed by formula (2).

number

[0053] As shown in Figure 12(a), the detection efficiency of the vacuum tube sensor 7 is PDE, and the number of electrons incident on the photocathode 12 due to noise is P. ephotocathod Therefore, the number of electrons P emitted from the photocathode 12 due to noise is... enoise This is expressed by equation (3). When the noise is X-ray noise, the energy of the characteristic X-ray is E. Xray Let the electron-hole pair production constant of the photocathode 12 (the energy required for ionization at the photocathode 12) be κ. photocathode Therefore, the number of electrons P ephotocathod This is expressed by equation (4).

number

number

number

[0054] Figure 12(b) schematically shows the single-pulse signal that appears in the target signal ST (sensor output) when a single photoelectron is emitted from the photocathode 12, and the noise signal that appears in the target signal ST when the noise to be removed is incident on the photocathode 12. The signal quantity (number of charges) Q of the single-pulse signal. single This is expressed by equation (6).

number

number

[0055] Based on the above, the determination formula in equation (8) is set.

number

[0056] The right-hand side (second value) of equations (2) and (8) will be explained. The signal quantity Q of a single-pulse signal. single The number of electrons P emitted from the photocathode 12 due to noise. enoise The value obtained by multiplying by (N) corresponds to the signal strength of the noise signal. pixel dwellBy dividing by ( / n), a value corresponding to the signal value per section divided into n parts is calculated. By comparing this value with the first value, it is possible to determine whether or not noise signals are present in the target signal ST during the measurement period. In calculating the second value, the number of electrons P enoise From the number of electrons P enoise The value used is obtained by subtracting 6σ, which is six times the fluctuation σ. This allows the judgment to be made while taking into account the fluctuation of the noise signal equivalent to 6σ. In this example, the second value is Q. single (P enoise It is based on -6σ, but the second value is Q single (P enoise It may be based on -mσ) (where m is 2 or greater). That is, m is not limited to 6; it can be an integer such as 2, 3, 4, 5, 7, 8, or a value with a fractional part after the decimal point, such as 2.5 or 3.5. In other words, m can be any real number greater than or equal to 2. m can be a real number greater than or equal to 2, 3, 4, 5, 6, etc.

[0057] In step S4 of Figure 10, by performing a determination using the determination formula (2), it is possible to determine whether or not noise occurred during the measurement period. If noise occurred during the measurement period, noise reduction processing can be performed (step S5), and if no noise occurred, normal calculation processing can be performed (step S6).

[0058] Figure 13 shows an example of an image with noise, Figure 14 shows an example of an image obtained after noise reduction processing is performed, and Figure 15 shows an example of an image obtained after judgment processing is performed. In Figures 13 to 15, (a) is an example for extremely low illumination, and (b) is an example for medium illumination (when the intensity of the measurement light L2 is normal). Figure 13 corresponds to the image obtained when the calculation method of the reference example (Figure 7) is applied, Figure 14 corresponds to the image obtained when the calculation method of the above embodiment (without judgment processing) is applied, and Figure 15 corresponds to the image obtained when the calculation method of the above embodiment (with judgment processing) is applied. The images in Figures 13 to 15 are images obtained by observing a sample having multiple areas with different brightness levels (the letters "H", "P", and "K" are written on a black background) using a laser scanning microscope, and the images were obtained while generating noise through simulation.

[0059] In the image of Figure 13(b), bright spots are generated in the black areas due to noise. Comparing Figure 13(b) and Figure 14(b), the bright spots are reduced in the image of Figure 14(b), indicating that X-ray noise can be removed by noise reduction processing. Comparing Figure 13(a) and Figure 14(a), the signal amount (the signal representing the letter "K") is reduced in the image of Figure 14(a), indicating that the signal amount is reduced by noise reduction processing at extremely low illumination. Comparing Figure 13(a) and Figure 14(a), the dark noise (bright spots generated in the black areas) is reduced in the image of Figure 14(a), indicating that dark noise can be reduced by performing noise reduction processing. Comparing Figure 15(b) and Figure 14(b), the bright spots are reduced in the image of Figure 15(b), similar to the image of Figure 14(b), indicating that X-ray noise can be removed by noise reduction processing. Comparing Figure 15(a) with Figures 13(a) and 14(a), the signal intensity (the signal representing the letter "K") in Figure 15(a) is higher than in Figure 14 and is at the same level as in Figure 13. This shows that the decrease in signal intensity can be suppressed even in extremely low light conditions by performing the judgment process. [Mechanism of Action and Effects]

[0060] In the signal processing circuit C used in the photodetector A of the laser scanning microscope 1, the processing unit 25 divides the measurement period into n intervals, calculates the signal value in each of the n intervals from the target signal ST, and calculates the output value for the measurement period from the signal values ​​in the n intervals. Furthermore, when the processing unit 25 calculates the output value for the measurement period, it is configured to perform noise reduction processing so that the magnitude of the largest maximum signal value Wmax among the signal values ​​in the n intervals does not affect the magnitude of the output value. This makes it possible to calculate the output value for the measurement period after removing high-power noise (X-ray noise in the above example) as needed. In other words, because the noise reduction processing is performed at the level of numerical data (output value) rather than image data, high-speed processing is possible. This enables high-speed processing of high-power noise (X-ray noise in the above example).

[0061] More specifically, for example, if the measurement period corresponds to one pixel P, the output value can be used as the value of pixel P. In this way, noise can be removed before calculating the value of pixel P, rather than after. This makes it possible to remove high-output noise while improving real-time performance and data accuracy. In other words, the method described in Patent Document 1 above identifies pixels where noise has occurred after image acquisition (after calculating the pixel value) and replaces the value of those pixels with alternative values, resulting in low real-time performance. Therefore, for example, when displaying the acquired image as live video, there is a risk that noise will appear in the live video. In contrast, the signal processing circuit C used in the photodetector A of the laser scanning microscope 1 removes high-output noise before calculating the value of pixel P, thus improving real-time performance. Therefore, for example, when displaying the acquired image as live video, it is possible to suppress the appearance of noise in the live video. Also, since it is not necessary to perform noise processing after image acquisition, data analysis can be carried out immediately after image acquisition. Furthermore, the method described in Patent Document 1 above replaces the value of pixels where noise has occurred with alternative values, resulting in low data accuracy. In contrast, the signal processing circuit C used in the photodetector A of the laser scanning microscope 1 calculates the value of pixel P using signals other than high-power noise acquired during the measurement period. Therefore, the value of pixel P is calculated using only measured data, thereby improving data accuracy. Furthermore, while the method described in Patent Document 1 may lead to increased processing complexity and a greater amount of data processing, the signal processing circuit C used in the photodetector A of the laser scanning microscope 1 can suppress this increase in data processing volume.

[0062] In the noise reduction process, the processing unit 25 calculates the average value of the remaining signal values ​​after excluding the maximum signal value Wmax from the signal values ​​in n intervals as the output value for the measurement period. This allows the output value for the measurement period to be calculated from the signal values ​​in n intervals in a way that the magnitude of the maximum signal value Wmax does not affect the magnitude of the output value.

[0063] Cutoff frequency f of current-voltage converter 21 c However, the pulse width of the noise signal to be removed that appears in the sensor output is set to be narrower than the width of each of the n intervals. This suppresses the appearance of the noise signal to be removed across multiple intervals, thereby preventing the high-output noise removal process from becoming overly complex.

[0064] When the processing unit 25 calculates the output value during the measurement period, the first value based on the maximum signal value Wmax is the number of electrons P emitted from the photocathode 12 when incident light (e.g., characteristic X-rays) that causes the noise to be removed is incident on the photocathode 12. enoise The electron number fluctuation σ and the signal quantity Q that appears in the target signal ST when one photoelectron is emitted from the photocathode 12 are considered. single If the value is greater than or equal to the second value based on the above, a noise reduction process is performed. If the first value is less than the second value, a calculation process is performed to calculate the average value of the signal values ​​over n intervals as the output value (steps S4-S6). This allows for noise reduction to be performed if noise occurs during the measurement period, and a calculation process to calculate the average value of the signal values ​​over n intervals as the output value if no noise occurs.

[0065] The first value of the determination formula is the value obtained by subtracting the smallest minimum signal value Wmin among the signal values ​​in n intervals from the maximum signal value Wmax. This allows for a suitable determination of whether or not noise occurred during the measurement period.

[0066] The second value of the decision formula is Q single (P enoise It is based on -mσ). This allows for a suitable determination of whether or not noise occurred during the measurement period. In the above embodiment, m is 6. In this case, high-output noise can be suitably removed even when the number of pixels P included in the observed image is extremely large.

[0067] The vacuum tube sensor 7 has an avalanche diode (electron detection unit 13) that multiplies electrons from the photocathode 12. Because such a vacuum tube sensor 7 uses a solid-state element like an avalanche diode, it is easy to perform noise reduction processing appropriately. Furthermore, in such a vacuum tube sensor 7, since electrons are accelerated and incident on the avalanche diode, high-energy electrons are incident on the internal structure of the vacuum tube sensor 7, making it prone to generating high-power noise based on characteristic X-rays. Therefore, applying the present invention to such a vacuum tube sensor 7 is particularly preferable because noise reduction processing is important and easy to perform appropriately.

[0068] The noise reduction method (signal processing method) according to the embodiment includes a processing step of processing a target signal ST corresponding to the output signal from a vacuum tube sensor 7 which has a photocathode 12 that emits electrons in response to incident light and outputs an output signal corresponding to the amount of light of the incident light. In the processing step, the measurement period is divided into n (n is an integer of 2 or more) intervals, the signal value in each of the n intervals is calculated from the target signal ST, and the output value for the measurement period is calculated from the signal values ​​in the n intervals. In the processing step, the output value is calculated from the signal values ​​in the n intervals such that the magnitude of the largest maximum signal value Wmax among the signal values ​​in the n intervals does not affect the magnitude of the output value.

[0069] The present invention is not limited to the above embodiments. The high-power noise to be removed is not limited to X-ray noise. Figure 16 is a configuration diagram of a modified laser scanning microscope 1. The laser scanning microscope 1 further includes a light source 9 for applying light stimulation to a sample S. The light source 9 outputs stimulation light L3 toward the sample S. In this case, the response of the sample S (e.g., living organism) to the light stimulation can be observed. Noise may also be generated when the stimulation light L3 enters the optical system of the laser scanning microscope 1 as stray light. This noise caused by the stimulation light L3 is also one of the high-power noises to be removed in the laser scanning microscope 1. Like X-ray noise, the noise caused by the stimulation light L3 is measured almost simultaneously with the measurement signal. The laser scanning microscope 1 can also remove noise caused by the stimulation light L3. Furthermore, it can remove not only noise caused by some incident light, but also noise signals caused by some event occurring in the vacuum tube sensor 7.

[0070] In the above embodiment, the processing unit 25 calculated the average value of the remaining signal values ​​after excluding the maximum signal value Wmax from the signal values ​​in n intervals as the output value for the measurement period. However, the processing unit 25 may also calculate the median value of the signal values ​​in the n intervals as the output value for the measurement period during the noise reduction process. For example, as in the example in Figure 8, if the signal values ​​in the four intervals are 108, 8, 5, and 2 in order of earliest measurement time, the median value for the four intervals is 6.5, and the value of pixel P is 6.5. Thus, although the maximum signal value Wmax has an influence on the selection of the numerical value to be used for the median, the maximum signal value Wmax is the most significant value and is therefore not used when actually deriving the magnitude of the median. In other words, no matter what the magnitude of the maximum signal value Wmax is, it does not affect the magnitude of the median. Therefore, even in this case, the output value can be calculated from the signal values ​​in n intervals in such a way that the magnitude of the maximum signal value Wmax does not affect the magnitude of the output value.

[0071] In the above embodiment, the first value of the determination formula was the value obtained by subtracting the smallest minimum signal value Wmin among the signal values ​​in n intervals from the maximum signal value Wmax. However, the first value may be the value obtained by subtracting the average value of the remaining signal values ​​excluding the maximum signal value Wmax from the maximum signal value Wmax, the median value of the remaining signal values ​​excluding the maximum signal value Wmax from the maximum signal value Wmax, or any one of the remaining signal values. In this case as well, it is possible to suitably determine whether or not noise occurred during the measurement period. Alternatively, the first value may be the maximum signal value Wmax itself. For example, if the measured signal is relatively small, it is possible to determine whether or not noise occurred during the measurement period without subtracting the minimum signal value Wmin, which corresponds to the magnitude of the measured signal, from the maximum signal value Wmax.

[0072] The vacuum tube sensor 7 does not necessarily have an avalanche diode. For example, the vacuum tube sensor 7 may be a photomultiplier tube equipped with a dynode as an electron multiplier unit that multiplies electrons from the photocathode 12, or it may be any other vacuum tube sensor as long as it has a photocathode. In the above embodiment, the measurement period corresponded to the time that excitation light L1 was irradiated onto one pixel P, but it is not limited to this, and an appropriate measurement period can be set according to the application. Note that it is necessary to divide the measurement period into n intervals and identify the interval that contains events that show abnormal values ​​such as high-power noise, so the measurement period needs to be of a certain length or longer, and it is preferable that it is at least twice the temporal element of the data shown by the event showing the abnormal value (for example, the pulse width when the electrical output shown by the event showing the abnormal value is converted into a voltage signal). In the above embodiment, the number of divisions n of the measurement period was 4, but the number of divisions n can be an integer of 2 or more, for example it may be 2.

[0073] In the above embodiment, the processing unit 25 (circuit unit 23) calculated the value of each pixel P in the observed image based on the target signal ST, but the control unit 24 may perform these processes. In other words, the signal processing circuit C may include the control unit 24. The control unit 24 may function as the processing unit 25 that processes the target signal ST instead of the circuit unit 23. The signal processing circuit C may have any configuration as long as it is a circuit capable of processing signals originating from electrons from the photocathode 12 of the vacuum tube sensor 7 in accordance with the present invention. Similarly, the photodetector A does not have to include the control unit 24, and may have any configuration as long as it is a photodetector capable of processing signals originating from electrons from the photocathode 12 of the vacuum tube sensor 7 in accordance with the present invention.

[0074] In the above embodiment, an example of the present invention being applied to a laser scanning microscope was described, but the present invention is applicable to other configurations, and may be applied to, for example, a flow cytometer (flow cytometry) or specimen testing. [Explanation of Symbols]

[0075] 7...Vacuum tube sensor, 12...Photocathode, 21...Current-voltage converter (current-voltage conversion unit), 25...Processing unit, C...Signal processing circuit, ST...Target signal, Wmax...Maximum signal value, Wmin...Minimum signal value.

Claims

1. In a signal processing circuit comprising a processing unit that processes a target signal corresponding to the output signal from a vacuum tube sensor having a photocathode that emits electrons in response to incident light and outputs an output signal corresponding to the amount of light of the incident light, and derives an output value at predetermined measurement intervals, The processing unit is configured to perform an abnormal value handling process in which, when dividing the measurement period into n (where n is an integer of 2 or more) intervals, deriving the signal value in each of the n intervals from the target signal, and deriving the output value for the measurement period from the signal values ​​in the n intervals, the output value is derived from the signal values ​​in the n intervals in such a way that the magnitude of the largest maximum signal value among the signal values ​​does not affect the magnitude of the output value.

2. The signal processing circuit according to claim 1, wherein the processing unit, in the abnormal value handling process, derives the average value of the remaining signal values ​​in the n intervals, excluding the maximum signal value, as the output value.

3. The signal processing circuit according to claim 1, wherein the processing unit derives the median value of the signal values ​​in the n intervals as the output value in the abnormal value handling process.

4. The system further includes a current-voltage conversion unit that converts a current signal corresponding to the amount of incident light into a voltage signal at a predetermined cutoff frequency. The signal processing circuit according to any one of claims 1 to 3, wherein the cutoff frequency is set such that the pulse width of the voltage signal corresponding to the maximum signal value is narrower than the width of each of the n intervals.

5. When the processing unit derives the output value during the measurement period, the first value based on the maximum signal value is the number of electrons P emitted from the photocathode when the incident light corresponding to the maximum signal value is incident on the photocathode. enoise The fluctuations in the number of electrons σ and the signal quantity Q that appears in the target signal when one photoelectron is emitted from the photocathode are given. single A signal processing circuit according to any one of claims 1 to 3, wherein if the value is greater than or equal to the second value based on the above, the abnormal value handling process is executed, and if the first value is smaller than the second value, a derivation process is executed to derive the average value of the signal values ​​in the n intervals as the output value.

6. The signal processing circuit according to claim 5, wherein the first value is the value obtained by subtracting from the maximum signal value the smallest minimum signal value among the signal values ​​in the n intervals, the average value of the remaining signal values ​​in the n intervals excluding the maximum signal value, the median value of the remaining signal values ​​in the n intervals excluding the maximum signal value, and one of the remaining signal values, or the maximum signal value.

7. The second value is Q single (P enoise A signal processing circuit according to claim 5, based on -mσ (where m is 2 or more).

8. The signal processing circuit according to any one of claims 1 to 3, wherein the vacuum tube sensor further comprises an avalanche diode for multiplying electrons from the photocathode.

9. A light detection device comprising the vacuum tube sensor and the signal processing circuit according to any one of claims 1 to 3.

10. The photodetector according to claim 9, wherein the vacuum tube sensor further comprises an avalanche diode that multiplies electrons from the photocathode.

11. The process includes a step of processing a target signal corresponding to the output signal from a vacuum tube sensor that has a photocathode that emits electrons in response to incident light and outputs an output signal corresponding to the amount of light of the incident light, and deriving an output value at predetermined measurement intervals, A signal processing method comprising the above-mentioned processing step, in which the measurement period is divided into n (where n is an integer of 2 or more) intervals, the signal value in each of the n intervals is derived from the target signal, and when deriving the output value for the measurement period from the signal values ​​in the n intervals, the output value is derived from the signal values ​​in the n intervals in such a way that the magnitude of the largest maximum signal value among the signal values ​​in the n intervals does not affect the magnitude of the output value.

Images