Method for correcting image distortion caused by a sensor in a hybrid pixel detector for x-rays or electrons, and hybrid pixel detector

Abstract

A method for correcting image distortion caused by a sensor (10) in a hybrid pixel detector (1) for x-rays or electrons comprises the steps of - in response to radiating the sensor (10) in an operational environment receiving a response per sensor pixel (101), - applying a correction to the received responses resulting in an image corrected for distortion effects, the correction being determined based on an evaluation of a trial image represented by trial responses from the sensor pixels (101) in response to radiating the sensor (10) in a trial mode.

Description

[0001] Method for correcting image distortion caused by a sensor in a hybrid pixel detector for x-rays or electrons , and hybrid pixel detector

[0002] Technical Field

[0003] The present invention refers to a method for correcting image distortion caused by a sensor in a hybrid pixel detector for x-rays and electrons , and to a hybrid pixel detector .

[0004] Background Art

[0005] Direct-converting pixel-based radiation detectors measuring incident radiation become more and more popular for many reasons . A direct-converting radiation detector is understood as a detector for directly converting incident radiation into electrical charges without intermediary conversion . Suitable sensor material for a sensor of such direct-converting radiation detector may be Si , for example . A pixel-based radiation detector is understood as a device that col lects signals from many adj acent pixels of a sensor o f such pixel-based radiation detector . Accordingly, a spatial or position resolution can be achieved by such pixel-based radiation detector .

[0006] Typically, such direct-converting pixel-based radiation detector , in short hybrid pixel detector , compri ses a sensor ti le of the sensor material in a planar dimension . Electrical contacts in form of e . g . metal li zations or e . g . ion implants are formed on a back side of the sensor tile and are in contact with the sensor material . These contacts are electrically isolated from each other and define pixels o f the sensor ti le given that each contact collects and leads of f electrical charges generated in the portion / volume of the contacted sensor material after its interaction with the incoming radiation / ionizing particle. On a front side of the sensor tile, a front electrode is arranged for applying a bias voltage. The applied bias voltage forms a static electric field between the front electrode and the contacts at the back side of the sensor tile.

[0007] An ionizing particle hitting the sensor tile generates a charge signal induced in the sensor material. The multitude of sensor pixels finally produce a 2D image of incident radiation, i.e. a distribution of radiation across the area of the sensor tile. A totality of responses from the sensor pixels for a given exposure time defines an intensity image.

[0008] A drift of the created charges inside the sensor tile depends on several parameters. Among these are mobility and lifetime of the created charges, the applied bias voltage, and a structure of the pixelated contacts. It is assumed that additional static or dynamic electrical fields may occur inside the sensor tile due to imperfection of the crystal structure and thus associated crystal defects and dislocation networks. These fields may lead to a non-uniform charge collection among the sensor pixels, which effect is also referred to as distortion .

[0009] Hence, it is desired to introduce a hybrid pixel detector producing images absent or with reduced effects evoked by such non-uniform charge collection among the sensor pixels .

[0010] Disclosure of the Invention

[0011] The problem is solved by a method for, preferably computerized, correcting image distortion caused by a sensor of a hybrid pixel detector for x-rays or electrons . In response to radiating the hybrid pixel detector, and in particular a sensor o f the hybrid pixel detector, in a trial mode , a trial image is produced which trial image is represented by a collection of the corresponding responses stemming from the sensor pixels . Based on an evaluation of these trial responses a correction is determined for correcting for distortion ef fects . The correction is applied to sensor pixel responses received in an operational mode of the hybrid pixel detector .

[0012] Accordingly, in response to radiating the hybrid pixel detector in an operational environment an image is generated represented by the collection of the responses from the sensor pixels . The correction is applied to the sensor pixel responses and the collection o f corrected responses produce an image corrected for distortion ef fects .

[0013] Whi le a pixel-based radiation detector provides position-resolved images represented by the responses of the individual sensor pixel s in response to radiation, the above mentioned distortion e ffects primarily lead to a lateral dri ft and hence a distortion in the allocation of charges between sensor pixels . In order to determine such mis-allocation, in one embodiment , and pre ferably prior to an operational mode of the radiation detector, the sensor of the hybrid pixel detector is radiated resulting in a trial image .

[0014] Whi le it would be desired to generate, in such trail mode , a uni form radiation at the sensor in order to determine uni form sensor pixel responses in an ideal sensor, and, hence, detect distortion in a real sensor by evaluating sensor pixel deviations directly from the trial responses , it is pre ferred to first model a f latfield image from the trial responses . This is preferred given that radiation from a radiation source , even of designed to uni formly radiate at its best, typically leads to a non-uni form radiation impact at the plane of the sensor tile , e . g . a bowl shaped radiation pro file . Accordingly, it is preferred to model , from the trail responses , a f latfield image , which, compensates for the anticipated inhomogeneity of radiation expected at the plane of the sensor tile . Corresponding parameters may be applied per sensor pixel response in the trial mode to determine or model a flat field image .

[0015] Accordingly, in cases where a uni form trial radiation cannot be achieved at the sensor, it is preferred to first model a uni form radiation in form of one or more flatfield images from the trial responses . Given that such flatfield only compensates for a inhomogeneity in radiation, distortion ef fects originating in the sensor still are present in and detectable from the flatfield image . Accordingly, in the flatf ield, the pixel intensity values deviate from each other and / or deviate from an average of responses across all sensor pixel s and / or deviate from a uni form response expected for an ideal sensor .

[0016] Accordingly, a flatf ield image is understood as a trial image corrected for irradiation inhomogeneity . The trial image is the image acquired in trial mode , in which the sensor of the hybrid pixel detector preferably is irradiated with a homogeneous irradiation prof ile . From this trial image, the flatfield image is calculated .

[0017] It is noted that the terms " response", "sensor pixel response" or "pixel response" may refer to the electrical signal suppl ied by the sensor pixel , and more pre ferably to any signal or representation derived from the bare electrical signal produced by the sensor in its sensor pixels . Given that the bare electrical signal pre ferably is pre-processed or processed in a readout circuit assigned to the sensor pixel , a response shal l also encompass the processed signal either at the output of the readout circuit or at any stage in the readout circuitry . Hence, a response may be represented by one or more of the ampli fied, digitized, pul se-counted, etc . , bare electrical signal produced by the respective sensor pixel as long as it represents the ingress of charge into the sensor pixel in some way, and in particular the magnitude of such ingress, which is also referred to in the following as amount or level or counter value, for example. Accordingly, when evaluating the deviations in the responses, it is preferred that such deviations are represented by quantifiable measures / magnitudes such as an amount or level or a counter value.

[0018] It is further noted that sensor pixels are defined by electrical contacts preferably arranged on a back side of a sensor tile and electrically isolated from each other. Given that such contact provides an electrical signal in response to radiating the sensor preferably on its front side, sensor pixels of the sensor tile may be defined by way of these contacts. However, a sensor pixel may also be represented by the corresponding charge collection volume in the sensor tile, from which charges are collected and are led off by the assigned contact.

[0019] The distortion induced deviations are preferably determined with respect to the amounts in the responses of different sensor pixels, preferably after flatfield modelling, and preferably also with respect to a direction in the plane of the sensor. The latter allows for defining the distortion effects by mis-allocations of charges in at least neighboring sensor pixels. Hence, any parameter / s describing such mis-allocation and hence the underlying distortion effects, in which way ever, preferably per sensor pixel pairs, are also referred to as distortion parameters .

[0020] In an operational mode, i.e. in an operational environment, the correction is performed based on the determined deviations. In the operational mode, measurements resulting in images are taken / recorded. The responses of the sensor pixels in the operational mode are corrected by means of the correction determined in the trial mode. Corrected responses are provided, per sensor pixel, such that the resulting image is an image corrected for distortions .

[0021] By these means, the overall image quality is improved. In particular, the spatial resolution of the image allows for an improved localization of structures or events in the image. The image distortion caused by a mis-allocation of charges in the plane of the sensor between sensor pixels is reduced if not avoided.

[0022] Given that the variations, i.e. non-uniformi- ties, in the responses in the trial mode, reflect the distortion effects, a quantified and preferably directed difference in the responses of preferably at least adjacent sensor pixels may represent the non-uni formities in form of deviations and hence distortion parameters which may be determined and stored per sensor pixel, preferably in relation to another sensor pixel. In this regard, a response in the trial mode shall also encompass the corresponding value of the flatfield image in case of applying a flatfield calculation to the trial image.

[0023] Accordingly, in one embodiment, the distortion parameters preferably represent a directed excess or a shortage of charges detected in the trial responses of the sensor pixels in response to the trial radiation.

[0024] In a preferred embodiment, the determination of the distortion includes an algebraic approach. In this embodiment drift values are determined based on the variations, per sensor pixel, indicating a portion in the amount of the trial response drifting to or from other sensor pixels, in particular to or from neighboring sensor pixels. Preferably, the drift values are determined at least for each sensor pixel - neighboring sensor pixel pair. In case of a grid of rectangular sensor pixel areas, a sensor pixel has eight neighboring sensor pixels, and, hence, eight drift values are determined. In another embodiment, drift values are determined for any sensor pixel pair within a defined range, preferably within 500 pm, from the respective sensor pixel, e.g. measured from its center. In a different embodiment, the drift values are determined for any sensor pixel pairs in the sensor irrespective of adjacency, and preferably for all sensor pixel pairs .

[0025] Each drift value may represent a measure of the amount in a response allocated to or from another sensor pixel. This way of considering the mis-allocations of charges in the sensor is also referred to as stealing, given that one sensor pixel may steal charge entries from one or more other, and in particular neighboring sensor pixels, while at the same time may be stolen by one or more other, and in particular neighboring sensor pixels. Preferably, the drift values are stored in a tensor.

[0026] In such and other embodiments, the correction preferably is derived from the distortion represented by the drift values, preferably by way of inversion of the drift values. Preferably, for correcting a response the mathematically inverted drift values are applied to the response. E.g., in case a drift value shows a positive sign in the tensor representing stealing charges from a neighboring pixel, the drift value is applied by a negative sign in the correction routine applied to the response of this neighboring pixel thereby returning the stolen charges to this neighboring pixel.

[0027] In a different embodiment, the determination of the distortion parameters includes a geometrical approach. In this embodiment, the trial responses are mapped into geometrical effective areas, each geometrical effective area being assigned to a sensor pixel. In this embodiment, the geometrical effective area represents an adapted area of a sensor pixel, preferably around its contact, which geometrical effective area denotes a virtual area assigned to a sensor pixel in which charges in the corresponding charge collection volume are collected. Preferably, the geometrical effective area of a sensor pixel is determined by relocating vertices defining a virtual geometrical effective area for an ideal sensor pixel according to the variations in the trial responses in a plane of the sensor pixels. Preferably, a vertex of the virtual geometrical effective area is iteratively relocated by means of a relocation vector comprising a deviation entry for each vertex. The vertex is relocated in each iteration step into the direction of the relocation vector.

[0028] The relocation of the vertices preferably is done in an iterative process. In each iterative step, the following calculations are performed, per sensor pixel: Calculating a normalized amount of the sensor response. The normalized amount is the value (=amount) of the sensor pixel response divided by the pixel area defined by the vertices position;

[0029] Calculating a relocation vector. The x and y coordinates of the relocation vector is the differences of the normalized pixel amounts in x and y direction for the summed upper / lower and left / right pixel pairs surrounding the vertex;

[0030] - Relocating the vertex, by adding step size times relocation vector to the vertex' coordinates.

[0031] This is done until the differences of the normalized amounts and unity is smaller than a predefined threshold, or until a number of iteration steps is reached.

[0032] The step of applying the correction parame- ter / s to a response in the operational mode of the radiation detector is preferably performed by a processing unit. The processing unit may, in one embodiment be part of the hybrid pixel detector, e.g. embodied as a dedicated part of the readout circuitry, in particular a part which receives the signals from the individual readout circuits assigned to the sensor pixels. In a different embodiment, the processing unit is arranged outside the hybrid pixel detector, such as an FPGA, a computer located on the same site as the hybrid pixel detector, or remote in a server , or in the cloud .

[0033] The steps of identifying the correction pa- rameter / s in the trial mode may either be performed by a processing unit at a factory o f the manufacturer of the hybrid pixel detector, e . g . by a computer or server , or by a processing unit of the hybrid pixel detector, such as an FPGA, or in the cloud .

[0034] Pre ferably, the correction and / or the underlying distortion parameters , e . g . in form of dri ft values and / or a map of the geometrical e f fective areas are stored in an electronic storage of the radiation detector such as an FPGA, alternatively in an electronic storage of a computer or server connected to the hybrid pixel detector and preferably located on-site, or in the cloud .

[0035] The trial mode may, in particular in case o f static distortions only, be conducted exclusively prior to initially operating the hybrid pixel detector in the operational environment . Here, the trial radiation may be controlled by a control unit external to the hybrid pixel detector, or by an internal control unit in case the hybrid pixel detector is part of a system including a radiation source .

[0036] However, in case of dynamic di stortions , the trial mode may only or in addition to a trial mode prior to the initial operation be performed during operation . In particular, the trial mode may be performed one or more of prior to, in between and after radiating the sensor o f the hybrid pixel detector in the operational environment for taking an image or a series of images .

[0037] According to another aspect of the present invention, a computer program product is provided compri sing computer program code means implementing a method according to any o f the preceding embodiments when executed on a process ing unit , in particular a processing unit of the hybrid pixel detector .

[0038] According to a further aspect of the present invention, a hybrid pixel detector is provided for pos ition-resolved detection of x-rays or electrons . The hybrid pixel detector compri ses at least one sensor tile with a front side facing incident radiation, and a back side oppos ite the front s ide . The sensor tile comprises sensor material sensitive to the radiation . A front electrode is arranged on the front s ide of the sensor tile . A set o f contacts of electrical ly conducting material is arranged on the back side of the sensor tile and is in contact with the sensor material , thereby def ining sensor pixel s . Readout circuitry is provided comprising a set of readout circuits in electrical connection with the contacts , wherein each readout circuit is configured to provide a response representative of the radiation incident in the corresponding sensor pixel in response to radiating the sensor tile o f the hybrid pixel detector in an operational environment . A processing unit is provided and configured to apply a correction to the responses resulting in an image corrected for the di stortion e f fects , which correction is determined based on an evaluation of non-uni formities in trial responses received from the sensor pixels in response to radiating the sensor o f the hybrid pixel detector in a trial mode .

[0039] Pre ferably, the process ing unit is configured to calculate a flatfield image from the trial image represented by the collection of sensor pixel responses in the trial mode . Preferably, the proces sing unit is configured to determine the correction based on an evaluation of non-uni formities in the flatf ield image .

[0040] In a pre ferred embodiment , each readout circuit of the set comprises a counter for counting pulses generated in the corresponding sensor pixel in response to the radiation incident thereto , and i s configured to provide the output s ignal subj ect to the counted pulses . The correction is determined based on the evaluation of the counted pulses per readout circuit .

[0041] The hybrid pixel detector may be appl ied in an electron microscope . The hybrid pixel detector may also be applied in an X-ray detection system .

[0042] Embodiments disclosed in connection with the method shall also be cons idered as disclosed in connection with the other claim categories .

[0043] Brief Description of the Drawings

[0044] The invention will be better understood and obj ects other than those set forth above wil l become apparent when consideration is given to the fol lowing detai led description thereo f . Such description makes reference to the annexed drawings , wherein the Figures illustrate :

[0045] Figure 1 a schematic cut view o f a hybrid pixel detector implementing hybrid pixel detection (HDD) , according to an embodiment of the present invention;

[0046] Figure 2 the sensor of Figure 1 in an enlarged view;

[0047] Figure 3 cut views o f a sensor of a hybrid pixel detector illustrating distortion e ffects ;

[0048] Figure 4 a view on a back side of a sensor o f a hybrid pixel detector illustrating distortion ef fects ;

[0049] Figure 5 a view on a back side of a sensor o f a hybrid pixel detector illustrating a method according to an embodiment o f the present invention;

[0050] Figure 6 a view on a back side o f a sensor of a hybrid pixel detector illustrating a method according to another embodiment of the present invention .

[0051] Modes for Carrying Out the Invention Hybrid pixel detector, generic embodiments:

[0052] A hybrid pixel detector (HPD) configured to detect radiation and specifically x-rays or electron radiation comprises a pixelated sensor, which is connected pixel by pixel to readout circuitry. The readout circuitry preferably is embodied in a readout chip, also referred to as ROC. Ionizing radiation, e.g. electrons above an energy of a few keV up to several 100s of keV in an electron microscope, or x-rays with an energy of a few 100s of eVs up to several 100s of keV, or gamma-rays, or alpha-rays, or other rays interact / s with the sensor material and deposits energy inside the sensor's material. Preferably, the sensor is made from a semiconductor material, such as from Silicon (Si) , Gallium Arsenide (GaAs) , Cadmium Telluride (CdTe) , Cadmium Zinc Telluride (CdZnTe) or Germanium (Ge) . Also more exotic materials like perovskites may be envisaged.

[0053] The deposited energy converts to electronhole pairs in the sensor material. A top side of the sensor preferably comprises a continuous front electrode, while a bottom side of the sensor comprises electrically separated back electrodes also referred to as contacts. A contact in turn at least co-def ines a sensor pixel. An electric field is formed inside the sensor by applying a voltage to the front electrode of the sensor and setting a virtual ground e.g. to an input of a first amplification stage of a ROC's pixel. A contact may be built up from a stack of layers, such as n++ or p++ implants, metal layers, insulation layers, e.g. SiO, SiN, ..., under bump metallization, etc. Charges created of one polarity in the sensor material - either electrons or holes, depending on the polarity of the applied voltage - drift along the field lines towards the front electrode, whereas charges created of opposite polarity drift towards one or several of the pixelated electrodes on the bottom side of the sensor material.

[0054] The readout circuitry preferably is embodied as ASIC and comprises a readout circuit per sensor pixel, also referred to as ROC pixel. A bottom layer of the sensor pixel contact is connected with means of an electrical connection such as a bump bond to a top layer of an electrode of a ROC pixel. Charges collected in a sensor pixel and supplied as electrical signal are thus processed in the corresponding ROC pixel. Signal processing in each ROC pixel may include one or more of (pre-) amplification, shaping, filtering, discriminating, integrating, storing, and / or counting of the electrical signals created in the sensor. In case the ROC pixels are of a counting type, then it is preferred that a threshold is applied subsequent to the preamplifier. In this case, when the amplified signal is larger than the threshold, a logical '1' signal is generated. In a counter logic, all logical '1' signals may be counted. Furthermore, it is preferred that each ROC pixel comprises a retrigger architecture. In this case, the length of a retrigger pulse may be configured.

[0055] In each ROC pixel, an electrical signal provided by the corresponding sensor pixel may achieve a final state, where it may be stored for a certain time, e.g. until it can be read out. The final state may be one or multiple counter values per ROC pixel, one or multiple analog signals per pixel, or a combination thereof. A ROC pixel may also store previous final states, e.g. in an analog or digital first-in-first-out (FIFO) circuit, in storage cells or similar.

[0056] Preferably, the electrical signal processed by the readout circuit is referred to as response or as response from the sensor pixel . A final state in a ROC pixel may also be referred to as "amount" of the response. This amount maybe an analog or a digitized charge, a counter value, a combination thereof, etc. It signifies the raw image content for a pixel in an image. Additionally to the amount, the information of the response may contain an x and y index of the pixel, a time stamp, etc. Besides particle counting, other readout schemes can be applied.

[0057] A hybrid pixel detector can be used in different applications, e.g. for imaging or measuring diffraction peaks of x-ray radiation, for measuring electron radiation e.g. in an electron microscope.

[0058] Radiation detector, specific embodiment:

[0059] Figure 1 illustrates a radiation detector implementing hybrid pixel detector (HPD) technology according to an embodiment of the present invention. The radiation detector 1 comprises a sensor 10 that is supported by and is electrically connected to a CMOS readout circuitry 2. The readout circuitry 2 in turn is supported by and electrically connected to a circuit board 3, e.g. by means of bond wires 31.

[0060] The sensor 10 includes a sensor tile 100 comprising or consisting of a sensor material suitable to convert incident X-rays / electrons - indicated by arrow e - into an electric charge. As sensor material, silicon, or a high-Z material like gallium arsenide (GaAs) , cadmium telluride (CdTe) , cadmium zinc telluride (CZT) , mercury iodide (Hgl) , perovskites, or others can be used. The sensor tile 100 has a planar extension in x- and y-direction.

[0061] A front electrode 102 is arranged on a front side 111 of the sensor tile 100. The front electrode 102 may comprise one or multiple thin layers of metal. In the latter scenario, the front electrode 102 preferably consists of a stack of materials, formed e.g. from aluminium, gold, platinum or other metals. In order to be able to collect the large amount of created charges due to the high incoming radiation current and its high energy, a bias voltage creating an electric field of at least 300 V / mm sensor thickness preferably is applied to the front electrode .

[0062] On a back or bottom side 112 of the sensor tile 100, pixelated electrical electrodes 101 are provided, which transmit the electrical signal to the readout circuitry 2. The readout circuitry 2 comprises a readout circuit per sensor pixel, also referred to as ROC pixel, and a corresponding contact 21. Each one of the electrodes 101 is connected to the contact 21 of the corresponding readout circuit by means of bump bonds 22. Assignments between electrodes 101 and contacts 21 other than 1:1 are possible.

[0063] In a preferred embodiment, the ROC pixel first collects the charge from the corresponding sensor pixel at its contact 21, amplifies this electrical signal in an amplifier stage, applies an adjustable threshold in a comparator to the amplified signal and counts the thresholded signals - preferably digital pulses at the comparator output - within a defined interval. Once the interval is terminated, the counter values are read out and send serially or in parallel to the readout board 3.

[0064] Figure 2 illustrates the sensor 1 of Figure 1 in an enlarged view.

[0065] Effect of distortion on a pixelated radiation detector :

[0066] A drift of the charges created inside the sensor depends on several parameters. Among these are the mobility and lifetime of the corresponding charge, a bias voltage applied to the front electrode, a structure of the pixelated electrodes, as well as potential static or dynamic fields inside the sensor. The applied bias voltage forms a static electric field between the front side of the sensor and the sensor pixels on the back side of the sensor.

[0067] According to the schematic diagram of Fig 3a) , an ionizing particle 102 enters the sensor tile 100, e.g. the sensor tile of the radiation detector of Fig. 1, e.g. typically being between 30 pm and 3000 pm thick. The ionizing particle 102 interacts at one or several points 103 with the sensor material and deposits part or all of its energy as electron-hole pairs. In an ideal sensor without static or dynamic fields inside the sensor, the charge cloud will mainly follow the illustrated field lines 104 implied by the external bias voltage. Additionally, in the lateral direction, the cloud size will expand due to charge diffusion and / or Coulomb repulsion or other effects leading to an increase of the charge cloud. Contacts 101 are embodied as metallizations or implants at the back side of the sensor tile 100, each defining a sensor pixel 202. Typical sensor pixel pitches may be between 10 pm and 10000 pm .In case the particle 102 interacts in the x-y direction in the middle between two contacts 101i and 101i as is illustrated in Fig. 3a) , then approximately the same amount of charge will drift to either contact 101i and 101i, left and right hand side of the x-location of the interaction point 103. Fig. 3a) , thus, represents and ideal sensor with sensor pixels 201i and 201i corresponding to the contacts 101i and 101i and indicated by the corresponding charge collection volumes separated by the dashed lines. In this case, a direct relation between the x-y coordinates of the interaction points and the x-y coordinates of the sensor pixels 202 collecting the corresponding charges is given.

[0068] In a real world sensor, however, additional static or dynamic fields may occur inside the sensor tile 100 due to imperfection of the crystal structure and thus associated crystal defects and dislocation networks. These fields may lead to non-uniform charge collection. In Fig. 3b) , a situation is illustrated in which a sensor tile 100 exhibits internal static or dynamic fields 105 indicated by arrows. In this case, charges do follow the superimposed field lines 104, with the field line components built up from a vector sum of the external applied field originating in the bias voltage and the internal static or dynamic fields 105. Besides the charge cloud expanding laterally due to charge diffusion and / or Coulomb repulsion, the charge may be laterally displaced. A direct relation between the x-y coordinate of the interaction point and the x-y coordinates of the responding pixel is thus not given anymore. As indicated by the dashed lines, the voxel of sensor pixel 201i is enlarged with respect to the charge collection volume of sensor pixel 2012. An image collected with this sensor will thus show distortion, which affects image quality. Depending on a strength of the lateral deviation, i.e. the static or dynamic field 105, the image distortion may be significant, e.g. up to 500 pm in x and / or y direction.

[0069] Si sensors typically show rather sufficient uniformity and little image distortion. However, subject to the application and subject to the crystal structure of the Si used for the sensor tile, distortion still may be regarded as detrimental. In compound materials, instead, the image distortion effect is very common, e.g. in materials such as GaAs, CdTe, CZT, perovskites, etc. Most materials show both static and dynamic fields, but in some materials one or the other may be more prevalent.

[0070] Figure 4 illustrates a schematic top view on a back side of a sensor tile 100 of a sensor of a hybrid pixel detector, according to an embodiment of the present invention. Contacts 101 are again defined by implants or metallizations deposited on the sensor tile 100, in this case in shape of a square. The dash-dotted box labelled 201 around each contact 101 indicates a virtual effective area assigned to the corresponding sensor pixel within each box 201, in an ideal sensor. In this ideal sensor, a charge deposited by a particle at location 203 will be captured by electrode 101i given that the virtual effective dash-dotted area covers the location 203. Instead, a charge deposited by a particle at location 204 will be captured by electrode IOI2 given that the virtual effective dash-dotted area of electrode IOI2 covers the location 204.

[0071] However, in a real world sensor charges or parts of charges may be laterally displaced to a corresponding neighbor pixel. In Fig. 4, the solid lines 202 denote this effect and illustrate effective areas of sensor pixels 202 in a real sensor tile 100 rather than in an ideal sensor tile 100. Thus, given the straight line borders of the effective areas, a charge deposited by a particle at location 203 will still be captured by electrode 101i given that the effective straight line area of electrode 101i covers the location 203. However, a charge deposited by a particle at location 204 will now be captured by electrode 101i instead of electrode IOI2 given that the effective straight line area of electrode 101i now covers the location 204, and no longer the effective area of electrode IOI2. Accordingly, corresponding sensor pixel 202i has expanded at the expense of sensor pixel 202i. As a result, charges deposited in particular in border regions of effective areas of sensor pixels 201 of an ideal sensor may in a real sensor be collected by neighboring sensor pixels, as is illustrated in Fig. 4.

[0072] Trial mode for measuring distortion:

[0073] In order to determine the effects of distortion on the pixel responses as illustrated in Fig. 4, it is preferred that a particular image is taken by the hybrid pixel detector which image is also referred to as a trial image.

[0074] Prior to the radiation, measurement conditions are preferably determined. A measurement condition may include one or more of the following: an energy of the ionizing particle, in case the ionizing particle's energy is monochromatic; a shape of the energy spectrum of the ionizing particles; a type of ionizing radiation (e.g. a electron, or an x-ray or a gamma-ray, . ..) ; a measured flux of ionizing particles per unit area (typical fluxes are between a few particles per mm2 and second up to several billion ionizing particles per mm2 and second) ; settings of the ROC pixels; in case the ROC pixels implement particle counting, an applied energy threshold; the preamplifiers open or closed gain, its transimpedance, or other typical preamplifier settings; a set length of a retrigger pulse if any; in case of integrating ROC pixels corresponding settings; a temperature of the sensor and / or the ROC; relative humidity; atmosphere pressure. Preferably, the measurement conditions for the trial mode are set as close as possible to the measurement conditions expected for the operational mode.

[0075] The number of measurement conditions may be quite large. As certain measurement conditions can be parameterized by a continuous variable, it may be advantageous to record trial images for a sub-set of the measurement conditions, and derive the un-distortion parameters from this subset. In this case, the un-distortion parameters for values of the corresponding measurement condition variable in between may be interpolated. Using interpolation thus reduces the number of trial images that need to be recorded in order to correct image distortions. Hence, in an embodiment trial images are recorded for a subset of the measurement conditions. Un-distortion parameters are preferably interpolated between the corresponding measurement conditions.

[0076] The trial image is taken in response to a uniform illumination / radiation of the sensor. Exposing the sensor to a uniform radiation is preferably achieved by placing the sensor / radiation detector at a place such that the illuminating field is sufficiently uniform. This can e.g. be achieved by setting the radiation detector at a sufficiently far distance from a quasi-point- like radiation source. In case of charged radiation such as electrons, lenses can be used to achieve a similar sufficiently uniform field. X-ray illumination is typically not uniform enough and may show, for example, a bowl-like shape, e.g. due to an insufficient distance from the radiation source. One may convert the trial image into a flatfield image as it would be obatained from a truly homogeneous illumination. For this, it is preferred to determine e.g. an average bowl-shape model, or a two dimensional spline by fitting to a local median or mean filtered response from the sensor pixels in the trial image. The flatfield image is then obtained by normalizing the trial image by dividing the responses in each pixel with a corresponding value from the fitted model function.

[0077] In case the trial image is sufficiently uniform it can be directly used as a flatfield image for the following processing. A uniform trial image could e.g. directly be achieved by scanning an inhomogeneous irradiation field over the detector.

[0078] Any other method capable of determining (estimating) the flatfield image could be employed, too.

[0079] In a subsequent step, the flatfield image is divided by the median or mean of the pixel's responses. The result is a percentage measure the response of a sensor pixel is above or below the median or mean of pixel responses across all sensor pixels.

[0080] Preferably one or several trial images are recorded with the uniform trial radiation, the number preferably is subject to sufficient statistics derivable from the recorded trial images .

[0081] Accordingly, a simple, practicable and pragmatic way is described to determine sensor distortion. For a given measurement condition, as minimum one flat- field image is required to be taken and evaluated in order to determine the image distortion, and to subsequently extract image un-distortion parameters aka correction parameters , which later on are applied to the responses received in operational mode, i . e . in an operational environment , e . g . at a customer .

[0082] Determining correction / un-distortion parameters :

[0083] In order to determine the correction parameters , several embodiments are described in the following . What is common to al l correction mechani sms i s that the flatf ield image first i s evaluated as to the non-uni form- ities in the flatf ield image . As laid out above, in one embodiment, the non-uni formities may be represented by the amount a pixel response i s above or below the average (mean or median) pixel response amount , which deviation amount may represent one embodiment o f a distortion parameter per sensor pixel .

[0084] It is preferred, that the correction parameters aka un-distortion parameters are derived / calculated based on the distortion parameters , again per sensor pixel . Preferably, the correction parameters are derived from the distortion parameters by inversion, again per sensor pixel .

[0085] Operational mode versus trial mode :

[0086] In contrast to the trial mode, in the operational mode the hybrid pixel detector is operated in an operational environment . In the operational mode, the correction determined in the trial mode are applied . In the operational mode, users wil l record data with the hybrid pixel detector needed for the respective application . Once an image is recorded in the operational mode , it is preferably corrected in place by applying the correction determined during the trial mode . This proces s may happen on the ROC, or near the ROC or on the readout FPGA or near the readout FPGA or on the detector server or on any other place o f the readout chain or remote from the hybrid pixel detector .

[0087] Accordingly, preferably the method includes the following : Image correction parameters for un-distorting image distortion originating from sensor e f fects are calculated for a given measurement condition from the corresponding trial or flatfield image alone . Trial or flat fields for di fferent measurement conditions are recorded in the trial or cal ibration mode as they are typical ly acquired for HPD during the production phase . Once the image distortion parameters are known from the flatfield, an algorithm may be applied that undistorts any distorted image taken with the same sensor . The un-dis- tortion of recorded images is then applied in the operation mode . The un-distortion may be done while images are being recorded . It may al so happen after the images are recorded and stored in a post-proces sing step .

[0088] Algorithmic distortion correction :

[0089] In an embodiment of the present invention, it is as sumed that the distortion of the image originates in an inhomogeneous sensor such that a charge generated by the incident particle ins ide the sensor tile - after considering charge di ffusion - is not collected in the sensor pixel corresponding to the x-y pos ition of the interaction point , but in a di fferent pixel . This e ffect is referred to as " stealing" in the fol lowing .

[0090] In the present embodiment , it is calculated from the flatfield image , pre ferably per sensor pixel , how much signal the sensor pixel steals from another sensor pixel , or, respectively how much signal is stolen by another sensor pixel from the sensor pixel. By doing so, a tensor is generated, also referred to as stealing tensor. The stealing tensor is a data structure which describes for each sensor pixel how much signal the sensor pixel steals from a different pixel, or, respectively, how much the different sensor pixel steals from the sensor pixel. The stealing tensor represents an embodiment of the distortion parameters. As indicated above, the stealing tensor preferably is stored, on one or several locations of for example on the ROC, on the readout FPGA, on a readout server, on a memory or storage device inside the readout chain, or on a different storage device.

[0091] Preferably, for un-distorting an image the stealing is inverted, i.e. in case it was found that a pixel steals a part from a different pixel's amount, then this part is returned to the original pixel. The determining of the correction parameters from the distortion parameters, i.e. presently from the stealing tensor, may be performed iteratively. In one embodiment, preferably applied as first step, the correction is determined by inverting the stealing tensor / matrix. However, alternatively or in addition, the correction is applied approximatively in multiple iterative steps, in which the correction values are improved step by step.

[0092] The concept of stealing is illustrated Fig. 5, again in a top view on a back side of a sensor tile 100. The contacts are no longer depicted for a better illustration, however, are present, of course, such as in Fig. 4. The straight lines 202 again, as in Fig. 4, represent effective areas of the sensor pixels. Presently, effective areas of four sample sensor pixels are denoted by 301, 302, 303 and 304. The dashed arrows denote the "stealing" of charges by a sensor pixel from another sensor pixel. The direction of an arrow denotes the sensor pixel that steals from its neighbor sensor pixel. In this illustration, sensor pixel 301 steals from pixel 303 and from sensor pixel 304, but is stolen from sensor pixel 302. In this respect, an algorithm is applied that finds the portion of the amount of charges that a sensor pixel steals from a neighbor sensor pixel by analyzing the flatfield image / s. Preferably, during operation mode, the stealing is inversely applied to the received responses from the sensor pixels.

[0093] A specific implementation of this embodiment makes use of an iterative algorithm to calculate the stealing tensor. In each iterative step, for each sensor pixel and each tensor element a difference of the amount between two sensor pixels is calculated by subtracting from each pixel the amount of the other pixel contributing to the tensor element. The difference is multiplied with a weight w, and a step width s, and then is subtracted from the original sensor pixel's amount at the end of the iterative step for all pixels and is added to the other pixel / s. While the step width s is a constant, usually between 10-10and 1, the weight w may depend on the two sensor pixel's amounts. For pixel amounts al and a2, one of the following functions for w can be applied, for example :

[0094] (1) w = 1

[0095] (2) w = abs (al-1) * (a2-l) )

[0096] (3) w = max (abs (al-1 ) , abs(a2-l) )

[0097] The iteration is either terminated after a predefined number of iteration steps, or using a resultbased abortion criterion, e.g., when the absolute deviation from unity of the pixel amounts, summed over all pixels, is smaller than a predefined threshold.

[0098] The stealing tensor preferably is represented by the sum of the products of step size s times weight w times the differences of the pixel amounts, over all iteration steps, and for each two pixels. Accordingly, for each sensor pixel pair, an identical value v is stored in each of related stealing tensor elements that is calculated accordingly. In a preferred embodiment, the number of tensor elements is reduced by half by keeping only one of the tensor elements with identical values. For an image with dimensions n x m, i.e. for a sensor with n x m sensor pixels, the stealing tensor will be very large of order of magnitude 0 (n2x m2) . It is preferred to assume that charge stealing only occurs between neighboring pixels, e.g. between a pixel and its 8 neighbors, or between a pixel and its 8 direct neighbors and its 16 next to direct neighbors etc. This assumption reduces the size of the stealing tensor to the order of magnitude of 0(n x m x 8 ) or O(n x m x 24) . In such scenario the iterative calculations are only applied for either any sensor pixel pairs out of the reduced set of 9 or 25 neighboring sensor pixels. Within the iteration, the above product of difference, weight and step width is then only subtracted from the sensor pixels of the reduced set.

[0099] In order to undistort an image, e.g. later in the operational mode, preferably the following algorithm is applied: For each sensor pixel pair (no 1, no 2) , the corresponding value v = vl = -v2 is looked up in the stealing tensor. The value v times the value of the pixel that is stolen from is added to the pixel's value, and is subtracted from the other pixel's value. Hence, for pixel values al and a2, new pixel values alnew and a2new are calculated by: alnew = al + al*v a2new = a2 - al*v

[0100] The idea is that al gets the same amount as a2 is losing.

[0101] This approach represents a first order approximation of the un-dis tortion correction, which, in another embodiment, may be applied iteratively in order to improve the corrected values. In this way, the stolen amount is given back to the original owner. This will give a first order corrected image, or, can be repeated multiple times to improve the un-correction. When repeating, the corrected image of the previous step is used for the calculation of the amounts to be transferred between pixels, which will then be applied to the recorded image to obtain the next iteration step of the corrected image.

[0102] Geometric distortion correction

[0103] In an embodiment of the present invention, the apparent deformation of the pixels in view of distortion effects is derived from the flatfield image in such a way that the pixels' amounts measured in the flatfield image corresponds to the geometrical effective area of the pixel in the x-y plane. The larger a pixel's amount, the larger its geometrical effective area. Accordingly, it is preferred to map the deviations determined from the flatfield image into a geometrical distortion map for the entire sensor tile.

[0104] In a preferred embodiment of this approach, the geometrical distortion of the pixel's shapes is calculated as follows. The calculation is illustrated in Fig. 6 by way of an exemplary view of a back side of a sensor tile 100:

[0105] The dash-dotted boxes referred to by 201 around each electrode (not shown) again indicates a virtual effective area assigned to the corresponding sensor pixel in an ideal sensor. The vertices of the effective areas of the ideal sensor, i.e. the vertices with coordinates (x,y) of the ideal sensor pixels are shifted in the x-y plane, forming new vertices with coordinates (x' ,y' ) . For example, vertex 401 of the ideal sensor pixel grid shifts into vertex 402, offset in both x and y coordinates with respect to vertex 401. The vertices of the effective areas of the ideal sensor are moved such, that the area now enclosed by the new vertices (x',y' ) corresponds to the amount recorded for the corresponding real sensor pixel. The corrected effective areas of sensor pixels 202 correspond to the quadrangles defined by these new vertices, e.g. for sensor pixels 202i, 2022, 202a, 2024. They are illustrated in Fig. 6 by straight lines. In general, the shape of these sensor pixels will rather not be regular .

[0106] In an embodiment, the new vertices coordinates (x',y' ) are found as follows: A set of new vertices (x' , y' ) coordinates is computed in such a way that the new effective areas assigned to the sensor pixels - the areas bordered by straight lines in Fig. 6 - minimize a difference to the corresponding amount in the flatfield image. The computation includes solving a system of nonlinear equations that can be performed by standard numerical solvers, e.g. Python Scipy Minimize. A dimensionality of the system scales with the number of pixels. For an image size of n x m pixels, a total amount of 2 x (n+1) x (m+1) unknowns have to be found.

[0107] In another embodiment, the vertices of the effective areas of the ideal sensor are shifted iteratively in the x-y plane for determining the vertices of the effective areas of the real sensor. A shifting vector v = (v_x, v_y) is defined by the amount of the responses in the four pixels adjacent to the vertex. In each iteration step, a part or all of the vertices are shifted by a vector b*v proportional to the shifting vector v. The proportionality factor b may be chosen as a constant for each iteration step, or adopted in each step to optimize the convergence of the algorithm. In each step, the amount in a given pixel is divided by the effective area assigned to the sensor pixel in the real sensor, given by the new vertices. The algorithm is stopped either after a predefined number of iteration steps, or until the new amounts meet a predefined condition, e.g. the new trial image is satisfyingly uniform. In another refined embodiment of the above approach, the shifting vector v = (v_x, v_y) is calculated from more sensor pixels then the ones adjacent to the vertex. For the calculation of the shifting vector v, the pixels' amounts may be weighted according to their distance to the vertex, e.g. by a 2D Gaussian function, or another distribution decaying over distance. This may be advantageous in case the distortion extend over a larger range than one pixel.

[0108] In a further embodiment, further vertices may be added on the edges between two vertices of the original rectangle or square. In the iteration steps, these vertices are moved according to the shifting vector for these vertices.

[0109] Accordingly, in the embodiment applying geometric distortion correction, the geometrical map of effective areas assigned to the sensor pixels for the specific (real) sensor represents the distortion parameters, and preferably is stored in one of the storage locations described further below.

[0110] The geometrical map may, in one eombodiment, be used as is, i.e. during or after recording images in an operational environment, these images are corrected according to the derived and stored geometrical distortion map. In one embodiment, this can be achieved by by directly using the irregular pixel map (geometrical distortion map) described by the new vertices

[0111] In other embodiments, the correction is based on one of:

[0112] 1) Re-distributing the sensor pixel response relative to the ideal pixel grid based on fractions proportional to the overlap of the geometric effective area with each ideal pixel .

[0113] Note that this allows to calculate the distortion tensor above directly. 2) Integrating signals over a certain detector area, the true overlap of the geometric active areas with the to be integrated detector area can be taken into account.

[0114] 3) In order to provide measured signal intensities at the true (undistorted) average pixel positions: Determine corrected pixel coordinates as the center of gravity of each pixel's geometric effective area. This is especially useful in the further processing of operational images e.g. of diffraction data in which the intensities of a diffraction ring are azimuthally integrated, or, a diffraction peak is integrated to obtain integrated intensity as a function of diffraction angle. The integrated signal will have better positional accuracy when using the corrected pixel coordinates in the integration.

[0115] In another embodiment, the measured amounts are assigned to an image of different, typically higher, pixel resolution than the measured image, which allows for oversampling, and resolving events in a subpixel resolution .

[0116] Points in time for trial mode:

[0117] In case the distortion fields are mainly of static nature, it may suffice to record a trial image for a given measurement condition only once, and derive the distortion and the corresponding un-dis tortion parameters accordingly once. In a first scenario, such trial mode is performed prior to an initial operation of the hybrid pixel detector, preferably at a factory of the manufacturer .

[0118] In another scenario, and in case of mainly static distortion fields, a pragmatic approach may be to include the recording of the trial image and the corresponding calculations in calibration routines that are usual ly performed at the factory anyway . Again, the trial image preferably i s recorded at a factory o f the manufacturer . Calibration routines are performed anyway for many sensors / radiation detectors . Calibration routines are col lectively referred to as calibration mode in which operational parameters are detected and / or determined as pre ferred for dif ferent measurement conditions prior to an initial operation of the sensor / detector . Cal ibration is typically performed on individual sensors / hybrid pixel detectors in order to account for the individual variations of material , manufacturing, as sembly, etc .

[0119] The calibration routines may already include the recording and calculation o f flatf ield images for other purposes . In a dif ferent embodiment , the recording of a flatfield image is newly introduced in existing calibration routines for the above purpose of identifying distortion ef fects and determining corresponding quantified distortion parameters , e . g . representing the magnitude and direction o f distortion per sensor pixel , and corresponding correction parameters .

[0120] In case the distortion fields have a strong dynamic component , recording trial images for the calculation of the distortion parameters may be needed to be taken closer to when images are taken in an operational environment . In thi s case, it may be advantageous to record the corresponding trial image / s at the beginning and / or end of a user ' s measurement series , or even interleaved with the measurement series in the operational environment . Image distortion and / or un-distortion parameters may be derived directly during the taking the series of images and appl ied on the images taken in the series . In another embodiment, the image distortion and / or undistortion parameters are calculated after the image series is taken and applied in a post-processing step . In case the distortion strongly changes during data taking, then the distortion and / or un-distor- tion parameters may be interpolated between the distortion and / or un-distortion parameters derived from the trial images either taken at the beginning and the end of the image series, or even interpolated from the distortion and / or un-distortion parameters derived from interleaved trial images.

[0121] Interpolation preferably is embodied by either interpolating the un-distortion parameters from the various trial images, or by interpolating the trial images, i.e. the corresponding distortion parameters and deriving the image un-distortion parameters from the interpolated trial images.

[0122] The interpolation may be linear, or quadratic, or cubic or any other polynomial function, or following a spline, or any other adequate interpolation function known from literature.

[0123] Location of correction processing and storage :

[0124] In a preferred embodiment, in particular in case of performing the trial mode in the factory, the processing of the correction is performed in a processing unit remote from the radiation detector, e.g. a server computer in the factory or a server computer in the cloud assigned to the manufacturer.

[0125] In a different embodiment, the processing of the correction is performed in a processing entity of the hybrid pixel detector, i.e. arranged at the location and preferably in the housing of the hybrid pixel detector. Specifically, the processing unit may be a processor implemented in the ROC, or a processor separate from the ROC. In a different embodiment, the hybrid pixel detector may comprise an interface to a server of the manufacturer , wherein the proces sing of the correction parameters may be performed on this server .

[0126] In a pre ferred embodiment , one or more o f the distortion parameters and the image un-distortion parameters are stored in a memory of the readout circuitry . In an alternative , the distortion and / or un-distortion parameters are stored in a memory of the radiation detector , such as a readout FPGA. In a further alternative , the distortion and / or un-distortion parameters are stored on a detector ass igned server or on any other place of the readout chain .

Claims

Claims1. Method for correcting image distortion caused by a sensor (10) in a hybrid pixel detector (1) for x-rays or electrons, comprising the steps of- in response to radiating the sensor (10) in an operational environment receiving a response per sensor pixel (101) ,- applying a correction to the received responses resulting in an image corrected for distortion effects, the correction being determined based on an evaluation of a trial image represented by trial responses from the sensor pixels (101) in response to radiating the sensor (10) in a trial mode.

2. Method of claim 1, wherein the correction is determined based on an evaluation of non-uni formities in a flatfield image derived from the trial responses from the sensor pixels (101) in response to radiating the sensor (10) in the trial mode.

3. Method of claim 2, comprising the additional steps of prior to initially radiating the sensor (10) in the operational environment:- in response to radiating the sensor (10) in the trial mode receiving the trial responses from the sensor pixels (101) ,- evaluating the non-uni form! ties in the flatfield image derived from the trial responses,- based on the evaluation of the non-uniform- ities in the flatfield image determining distortion parameters,- defining the correction based on the distortion parameters .

4. Method of claim 2 or claim 3, comprising the additional steps of one or more of prior to, in between and after radiating the radiation detector (1) in the operational environment for taking an image or a series of images represented by the collective responses of the sensor pixels :- in response to radiating the radiation detector (1) in the trial mode receiving the trial responses from the sensor pixels (101) ,- evaluating the non-uni form! ties in the flatfield image derived from the trial responses,- based on the evaluation of the non-uniform- ities in the flatfield images derived from the trial responses determining distortion parameters- defining the correction based on the distortion parameters .

5. Method of claim 3 or claim 4, comprising controlling a radiation source to acquire the trial image by the radiation detector (1) .

6. Method of any of the preceding claims 3 to5, wherein the evaluation of the non-uni formi- ties in the flatfield image includes determining, per sensor pixel (101) , a deviation from one or more of a normalized, a mean and a median value of the flatfield.

7. Method of claim 6, comprising determining, based on the determined deviations, per sensor pixel (101) , drift values indicating a portion in the amount of the trial response drifting to or from other sensor pixels.

8. Method of claim 7, wherein the drift values are determined at least for each sensor pixel - neighboring sensor pixel pair, preferably wherein the drift values are determined for any sensor pixel pair within a defined range, preferably within 500 pm.

9. Method of claim 7 or claim 8, wherein the drift values are inverted, preferably mathematically inverted, wherein the correction is determined dependent on the inverted drift values.

10. Method of any of the preceding claims 7 to 9, wherein the correction is determined per sensor pixel (101) , wherein the correction per sensor pixel (101) is determined dependent on the drift values determined for the sensor pixel with respect to other sensor pixels, preferably wherein the correction is determined iteratively.

11. Method of claim 6, comprising mapping, per sensor pixel, the sensor pixel area into a geometrical effective area (202) with a size proportional to the sensor pixel response in the trial mode .

12. Method of claim 11, wherein the geometrical effective area (202) of a sensor pixel is determined by relocating vertices defining a virtual geometrical effective area (201) for an ideal sensor according to the deviations in the trial responses in a plane of the sensor pixels (101) .

13. Method of claim 12, wherein a vertex of the virtual geometrical effective area (201) is iteratively relocated by means of a relocation vector comprising a deviation entry for each vertex, and wherein the vertex is relocated in each iteration step into the direction of the relocation vector .

14. Method of any of the preceding claims, which steps are performed by a processing unit, and which correction is stored in an electronic storage, preferably wherein the processing unit and the electronic storage are components of the hybrid pixel detector ( 1 ) .

15. Computer program product, comprising computer program code means implementing a method according to any of the preceding claims when executed on a processing unit.

16. Hybrid pixel detector configured to detect x-rays or electrons, comprising- at least one sensor tile (100) with a front side (111) facing incident radiation, and a back side (112) opposite the front side (111) , the sensor tile (100) comprising sensor material sensitive to x-rays or electrons ,- a front electrode (102) arranged on the front side (111) of the sensor tile (100) ,- a set of contacts (101) of electrically conducting material arranged on the back side (112) of the sensor tile (11) and in contact with the sensor material, thereby defining sensor pixels (101) ,- readout circuitry ( 2 ) comprising a set of readout circuits in electrical connection with the contacts , wherein each readout circuit of the set is configured to provide a response representative of the radiation incident in the corresponding sensor pixel ( 101 ) in response to radiating the hybrid pixel detector ( 1 ) in an operational environment, and a processing unit configured to apply a correction to the responses resulting in an image corrected for the distortion ef fects , which correction is determined based on an evaluation of a trial image represented by trial responses received from the sensor pixels ( 101 ) in response to radiating sensor ( 10 ) in a trial mode .17 . Hybrid pixel detector of claim 16 , wherein the processing unit is arranged remote from the sensor tile ( 100 ) .18 . Hybrid pixel detector of claim 16 or claim 17 , wherein each readout circuit of the set comprises a counter for counting pulses generated in the corresponding sensor pixel in response to the radiation incident thereto , and is configured to provide the counted pulses as response, and wherein the correction parameters are determined based on the evaluation of the counted pulses of the readout circuits .RECTIFIED SHEET (RULE 91) ISA / EP

Images