Method for temperature compensation of an image sensor sensitivity

CA 02395571 2002-06-25 WO 01/49044 PCT / FROO / 03717 1 TEMPERATURE COMPENSATION PROCESS FOR THE SENSITIVITY OF AN IMAGE DETECTOR The present invention relates to solid state image detectors and is proposed to eliminate variations in their sensitivity, in particular those due to temperature variations.

In these image detectors, the acquisition of an image is carried out using several photosensitive points each formed of a photodiode and a switch.

The photosensitive dots are produced using thin film deposition techniques of semiconductor materials such as hydrogenated amorphous silicon (aSiH).

These photosensitive points arranged in a matrix or bar make it possible to detect images contained in visible or near visible radiation.

The signals they produce are then digitized so that they can be easily stored and processed.

These arrangements of photosensitive points find a particularly interesting application in the medical field or that of industrial control where they detect radiological images.

It suffices to cover them with a scintillator and expose the latter to X-rays carrying a radiological image.

The scintillator converts the incident X-radiation into radiation in the wavelength band to which the photosensitive dots are sensitive.

Large photosensitive matrices are now found which can have several million photosensitive dots.

Reference is made to FIG. 1 which shows an image detector of known matrix type.

It has only nine photosensitive points so as not to overload the figure unnecessarily.

Each photosensitive point P1 to P9 is formed by a photodiode Dp and a switch function element Dc represented in the form of a switching diode.

We could have chosen a transistor as a switch-function element.

Photodiode Dp and switching diode Dc are interconnected in a head-to-tail assembly.

Each photosensitive point P1 to P9 is connected between a row conductor Y1 to Y3 and a column conductor X1 to X3.

Line conductors Y1 to Y3 are connected to an addressing device 3 CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 2 known by the Anglo-Saxon name of driver.

There can be several drivers 3 if the matrix is ​​large.

The addressing device 3 generally comprises shift registers, switching circuits, clock circuits.

The addressing device 3 brings the line conductors Y1 to Y3 to voltages which either isolate the photosensitive points P1 to P3 connected to the same line conductor Y1 from the rest of the matrix, or put them into conduction.

The addressing device 3 makes it possible to carry out sequential addressing of the line conductors Y1 to Y3.

The column conductors X1 to X3 are connected to a reading device CL.

During an image taking phase during which the photosensitive points P1 to P9 are exposed to information to be captured and that they are in a receptive state, that is to say that their photosensitive Dp and switching diodes Reverse-biased Dc each constitute a capacitor, an accumulation of charges occurs at the junction point A between the two diodes Dp, Dc.

The quantity of charges is substantially proportional to the intensity of the signal received, whether it is a very intense illumination, insofar as one remains in the linear detection range 2o of the photosensitive diodes, or the darkness.

Then follows a reading phase, during which is applied to the row conductors Y1 to Y3 sequentially a reading pulse which puts the photodiodes Dp in conduction and allows the evacuation of the charges accumulated in the column conductors X1 to X3 towards the device of CL reading and their integration.

The CL reading device will be seen in more detail.

It has as many read circuits 5 as there are column conductors X1 to X3 and these read circuits are of the charge integrator circuit type.

Each photosensitive point is connected to a read circuit 5.

Each charge integrator circuit is produced by an operational amplifier G1 to G3 mounted as an integrator using a read capacitor C1 to C3.

Each capacitor C1 to C3 is mounted between the negative input of the operational amplifier G1 to G3 and its output S1 to S3.

Each column conductor X1 to X3 is connected to the negative input of an operational amplifier G1 to G3. The positive input of each of the operational amplifiers G1 to G3 is brought to a voltage of AC 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 3 constant VR input reference which imposes on each column conductor X1 to X3 this reference voltage.

Each operational amplifier G1 to G3 has a reset switch II to 13 connected in parallel with the capacitor C1 to C 3.

The outputs S1 to S3 of the integrator circuits are connected to a multiplexing device 6 which delivers in series signals corresponding to the loads which have been integrated by the load integrator circuits.

In the reading phase, these signals correspond to the charges accumulated during an integration time by all the 1o photosensitive points of the same line.

The signals delivered by the multiplexing device 6 are then digitized in at least one analog-digital converter 7, the digitized signals at the output of the analog-digital converter 7 translate the content of the image to be detected.

These digitized signals are transmitted to a management system 8 which can store them, process them and display them.

It can be seen that the sensitivity of such detectors varies, which leads to both local and global variations in the brightness of the detected image.

The variation in sensitivity has several origins.

There is on the one hand a spatial variation and on the other hand a thermal variation.

This means on the one hand that two photosensitive points of the detector may not give the same response when they are exposed to exactly the same luminous flux, and on the other hand that a photosensitive point exposed to the same luminous flux does not give the same response at 25 C as at 35 C.

These defects are partly due to the semiconductor components constituting the photosensitive dots which do not all come from the same production batch and partly to the scintillator material used in radiology.

We obtain images with non-uniform areas which should not be there and which are increasingly pale the more the temperature increases.

If we know how to overcome the spatial variation in sensitivity by performing an image correction with a so-called gain image, it is not possible to use the gain image to overcome the variations. sensitivity to thermal causes.

The gain image is an image made with uniform illumination calibrated in the absence of a subject or object to be examined.

It makes it possible to correct the variations in spatial sensitivity since with a CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 4 uniform illumination the image should be uniform.

This gain image is produced with a very low frequency, of the order of one year.

The signals delivered by the photosensitive dots during the reading of the gain image are stored in the management device 8, and are then used to correct any useful image, for the spatial inhomogeneity of sensitivity.

This method cannot be used to overcome the variations in sensitivity, the cause of which is thermal: it would be necessary to produce images of gain at the rate of the variations in temperature, which would significantly increase the frequency of taking the images of the temperature. gain.

This is not compatible with the way operators use such image detectors.

The present invention proposes to use a gain image or a quasi gain image adapted to the ambient temperature in order to overcome variations in the sensitivity of the image detector, in particular due to heat, but this gain image is not simply taken just before making the correction to be adapted to the ambient temperature, it is produced from a calculation leading to the determination of the ambient temperature.

To achieve this, the present invention proposes a method for temperature compensation of the sensitivity of an image detector comprising photosensitive points each with a photodiode, connected to read circuits, characterized in that the photosensitive points are divided into photosensitive points detectors capable of detecting an image when they are exposed to information carrying the image and when they are sensitive to this information, and blind spots protected from information, and that it consists of:

- when the photosensitive dots are brought to a reference temperature, calculating an average leakage current in the photodiodes of the blind photosensitive dots and calculating an average from the signals delivered by the blind photosensitive dots during a first reading operation , - when the photosensitive points are brought to an ambient temperature to be determined, to calculate an average from the signals delivered by the blind photosensitive dots during another reading operation, CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 - to calculate the ambient temperature from the current of average leakage and the difference between the two averages, - to develop a gain image or a quasi-gain image adapted to the ambient temperature, 5 - to correct an image taken at ambient temperature with the gain image or the quasi image of gain.

Preferably, the signals delivered for the preparation of the first average and for the preparation of the average at the ambient temperature to be determined correspond to integrated loads with a first integration time lo substantially equal to the nominal integration time of the detector. image.

For the calculation of the average leakage current at the reference temperature, it is possible, at the reference temperature, to develop a pair of averages from the signals delivered by the blind photosensitive points with two different integration times and to carry out a calibration with the couple of averages.

One of the torque averages is advantageously the first average. The other average of the couple corresponds to integrated loads with an integration time greater than the nominal 2o integration time of the image detector.

The development of the gain image adapted to the determined ambient temperature can be done from a series of gain images previously stored in a memory device, each of them having been taken at a particular temperature. different, all of these particular temperatures forming a range of temperatures at which the image detector is capable of operating.

The development of the quasi gain image adapted to the determined ambient temperature can be done, more advantageously, from a base gain image taken at a base temperature, corrected using a coefficient variation of the base gain image as a function of the temperature and taking into account the difference between the determined ambient temperature and the base temperature.

The base temperature can be the reference temperature.

The development of the averages can advantageously be done from black images.

CA 02395571 2002-06-25 WO 01/49044 PCT / FROO / 03717 6 The present invention also relates to an image detector for implementing the compensation method, comprising photosensitive points each with a photodiode, these photosensitive points being connected to reading circuits.

These photosensitive points are divided into photosensitive points detectors capable of detecting an image when they are exposed to information carrying the image and blind points protected from the information.

The detector comprises means which deliver averages from the signals delivered by the blind photosensitive points, means 1o for calculating the average leakage current in the photodiodes of the blind photosensitive points, means for calculating the ambient temperature from the current of average leak and a difference between averages, means for developing the gain image or the quasi gain image from the calculated ambient temperature and means for correcting an image taken at ambient temperature with the gain image or the quasi gain image.

The means for calculating the average leakage current receive the averages of the signals delivered by the blind photosensitive dots in digital form.

The means for developing the gain image may include a memory device containing one or more gain images, each corresponding to a temperature.

The means for developing the quasi-gain image may include a memory device containing a base gain image taken at a base temperature and a coefficient of variation of the base image with temperature.

It is preferable that the blind photosensitive dots are connected to end portions of conductors to which the photosensitive detector dots are connected.

The blind photosensitive dots are covered with a material opaque to the information received by the photosensitive detector dots, this material being in particular black paint.

The detector photosensitive dots are covered with a scintillator material which transforms X-rays into radiation to which they are sensitive, the blind photosensitive dots being covered with a material opaque to X-rays, such as lead.

CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 7 The material opaque to information is located between the material opaque to X-rays and the blind photosensitive dots.

Other characteristics and advantages of the invention will become apparent on reading the following description, illustrated by the figures which represent:

- Figure 1, already described, an example of a known image detector;

FIG. 2, an example of an image detector according to the invention capable of operating with the method according to the invention.

As in Fig. 1, the photosensitive dots 01 to 06 and R1 to R9 are shown with a photodiode Dp and a switch function element Dc shown as a switching diode.

This switching diode could have been replaced by a transistor.

Photodiode Dp and switching diode Dc are interconnected in a head-to-tail assembly.

Each photosensitive point is connected between a row conductor Y1 to Y3 and a column conductor W1, W2 and Z1 to Z3.

The photosensitive points 01 to 06 and R1 to R9 are arranged in a matrix according to rows and columns but they could be arranged in a linear bar.

Compared to the example of FIG. 1, the image detector shown has more photosensitive points and more column conductors but the same number of row conductors.

The line conductors are connected to an addressing device 3 comparable to that described in FIG. 1.

According to one characteristic of the invention, the photosensitive points are divided into two categories, photosensitive points detectors R1 to R9 which, when they are exposed to information carrying an image and when they are sensitive to this information, are capable of detecting the image and blind photosensitive dots 01 to 06 serving for the compensation These blind photosensitive dots 01 to 06 are masked from the information carrying an image to be detected.

When detecting an image, be it an object or patient image or even a black (no illumination) or gain image, the blind photosensitive dots receive nothing.

These blind photosensitive points 01 to 06 will be read in the same way as the photosensitive points detectors R1 to R9.

CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 8 The blind photosensitive dots 01 to 06 are connected to end portions of the conductors of lines Y1 to Y3.

In the example described, they are located at the start of the line, they could be located at the end of the line.

The number of blind photosensitive dots is not critical, around ten per row seems reasonable if a row has about 2000 detector photosensitive dots.

These photosensitive points 01 to 06, R1 to R9 are implanted on an insulating support referenced 21.

To mask the blind photosensitive points 01 to 06 vis 1o with respect to the information to which the detector photosensitive points are exposed, they are covered with a PN material opaque to the information received by the detector photosensitive points, black paint for example fits very well.

In the configuration where the image detector according to the invention is used in a radiological application, the photosensitive points detectors R1 to R9 are covered with a scintillator material SC which transforms an X-ray into radiation in the band of lengths of wave to which the photosensitive points detectors R1 to R9 are sensitive.

As for the blind photosensitive dots 01 to 06, they are not covered with the scintillator material SC but with a material PB opaque to X-rays, a layer of lead for example.

In this configuration, the PN material opaque to the information received by the detector photosensitive dots is optional but if it is used, it is placed between the blind photosensitive dots 01 to 06 and the PB material opaque to X radiation.

The entire surface of the image detector on the side from which the X-rays originate is covered with a PP protective material based, for example, on carbon fibers.

In FIG. 2, these materials are only partially seen As in the example of FIG. 1, the reading device CL comprises as many reading circuits 30a, 30b as there are column conductors W1, W2 and Z1 to Z3 and each of these read circuits is of the charge integrator circuit type with an operational amplifier Ga, Gb mounted as an integrator using a read capacitor 31a, 31b and a reset switch la, lb mounted in parallel with the read capacitor 31 a, 31 b.

As in FIG. 1, the outputs 32a, CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 9 32b of the integrator circuits Ga, Gb are connected to a multiplexing device 60 which is connected in series. signals corresponding to the loads which have been integrated by the load integrator circuits.

The signals delivered by the multiplexing device 60 are then digitized in at least one analog-to-digital converter (ADC) 70.

The signals coming from the reading circuits 30b connected to the photosensitive detector points R1 to R9 translate the image to be detected, the others coming from the reading circuits 30a connected to the blind photosensitive points 01 to 06 are used for the compensation.

The digitized signals are then transmitted to a management system 80 which can store them, process them and display them.

According to the method according to the invention, when the photosensitive points are brought to a reference temperature 6ref, an average leakage current 10ref is calculated in the photodiodes of the blind photosensitive points and a first average COD1 is calculated from the signals delivered by the blind photosensitive dots 01 to 06 during a read operation.

The reference temperature can be measured using a thermometer.

When the photosensitive dots are brought to an ambient temperature 0 to be determined, another mean COD1 ′ is produced from the signals delivered by the blind photosensitive dots 01 to 06 during another reading operation.

The two averages COD1, COD1 ′ are used in digital form and it is preferable to produce them from the signals delivered by the blind photosensitive dots already converted by the analog-to-digital converter 70.

It is even conceivable that it is the analog-to-digital converter 70 which delivers the averages.

However, it is possible to take the averages in analog form and then convert them.

In the non-limiting example of FIG. 2, the means 700 for working out the averages are shown diagrammatically by dashes, they include the analog-to-digital converter 70.

The signals taken into account for the preparation of the first average COD1 correspond to the charges stored by the blind photosensitive points with a first integration time t1.

This first integration time is preferably substantially the nominal integration time of the detector, that is to say the integration time CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 corresponding to current use of the detector.

For example, t1 can be given a value between 0.5 and 5 seconds.

Likewise, the signals taken into account for the preparation of the other mean COD1 ′, at ambient temperature to be determined, correspond to the charges stored by the blind photosensitive points with an integration time t1 ′ and this time d. integration t'l is preferably substantially the nominal integration time of the detector and therefore substantially equal to tl.

The ambient temperature 0 is calculated from the difference between the means COD1 ′, COD1 and the mean leakage current IOref at the reference temperature.

Then an image of gain IG or a quasi-image of gain QIG adapted to the determined ambient temperature 0 is produced in order to correct an image detected while the photosensitive dots are always brought to the determined ambient temperature 0.

We will now see an embodiment of these different steps.

The leakage current in a photodiode varies with temperature exponentially.

This leakage current is given by the following formula:

10 = 16ref x 10ffl-erepi0 The current Ip from a photosensitive point is given by:

CODxFSRxCIec Ip 2n xGxt with:

= COD: the code, expressed in binary (LSB), delivered by the analog converter 70 during an operation to read the charges stored by this photosensitive point after an integration time of t.

The integration time t corresponds to the time elapsed between two successive evacuations of the charges accumulated at point A of the photosensitive point.

COD can take values ​​from 0 to 214 if the converter has 14 bits.

= FSR: the coding voltage range of the analog-to-digital converter.

This range can be for example 4 volts.

= n is the resolution of the analog-to-digital converter.

This resolution may for example be 14 bits.

CA 02395571 2002-06-25 WO 01/49044 PCT / FROO / 03717 11 = Clec represents the equivalent reading capacity at the output of the reading circuit CL.

= G the voltage gain separating the output of the read circuit from the input of the analog-to-digital converter.

s During a reading operation of a blind photosensitive point 01 to 06 (which is therefore not exposed to illumination), the charges accumulated by this photosensitive point do not correspond exactly to the leakage current of the photodiode.

These charges also include driving charges created in the photosensitive point 1o from the pulses it receives from the addressing device 3 and charges from the photosensitive points connected to the same column conductor as that which is read.

To overcome these additional charges, a calibration operation is performed to calculate the leakage current.

When the 15 photosensitive points are brought to the reference temperature Oref, a pair of means COD1, COD2 is produced from the signals delivered by the blind photosensitive points with two different integration times.

It was assumed subsequently, for the sake of non-limiting simplification, that one of the averages of the pair COD1 corresponds to that produced at the reference temperature Oref but with a view to calculating the temperature 0 to be determined.

The other average COD2 of the pair is worked out from signals delivered by the blind photosensitive dots 01 to 06 during a second reading operation at the reference temperature Oref.

The integration time t1 relating to the first average COD1 of the torque is approximately the nominal integration time of the detector.

The integration time t2 relating to the second average COD2 of the couple is greater than the first time t1.

It can for example be chosen 2 to 10 30 times longer than the time t1.

This time t2 is for example 1 to 20 seconds.

It is preferable that the read operations for the calculation of the average leakage current and for the elaboration of the averages are carried out from black images obtained when the photosensitive dots are not exposed to any illumination.

In a CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 12 radiological application, this eliminates the need for X-radiation.

But one could consider using useful images.

With these two averages COD1, COD2, it is then easy to calculate the average leakage current IAref at the reference temperature Oref according to the formula:

16ref -_ (COD2 - COD1) xFSRxClec 2n xGx (t2 - t1) We can then calculate the ambient temperature to be determined 0 by the formula:

6-8ref 1 + log (1- (COD1'-COD1) xFSRxClec) 2n xGxl6refxtl 'When the ambient temperature 0 has been calculated, it suffices to produce an image of gain IG or a quasi image of gain QIG adapted to this temperature 0 ambient, this image of gain IG or this quasi image of gain QIG will be used to digitally correct, in sensitivity, a useful image at temperature 0.

The development of the gain image IG adapted to the temperature can be done from a series of gain images IG1 ..... IGn previously stored in a memory device 100, each of them being associated with a temperature.

We subjected, beforehand, the image detector 2o to a series of temperatures at which it may have to operate, and for each of them, an image of gain IG1 ..... IGn was taken that the we have stored in the memory device 100 so as to constitute a library of images of gain IG1 ..... IGn.

If a less precise correction can be admitted, it suffices to take a base gain image IGb at a base temperature and determine a coefficient K of variation of the base gain image as a function of the temperature. The base gain image IGb is stored in the memory device 100.

This variant requires less memory capacity which can be appreciable.

The quasi gain image QIG is obtained by applying the coefficient K to the base gain image IGb taking into account the difference between the base temperature and the calculated ambient temperature.

The base temperature can be equal to the reference temperature Oref.

CA 02395571 2002-06-25 WO 01/49044 PCT / FR00 / 03717 13 In FIG. 2, the two variants are illustrated in the same memory device but this is not limiting.

Measurements carried out show that the coefficient K is of the order of -0.5% / C.

All the functions which have just been described can be performed by the management device 80 itself or by any processing unit 800 placed between the analog-to-digital converter 70 and the management device 80 which then allows the use of the corrected images. .

FIG. 2 shows in detail the configuration with processing unit 800 which includes the means 81 for calculating the average leakage current 1o, the means 82 for calculating the ambient temperature 0 to be determined, the means 83 for developing the gain image IG or the quasi gain image QIG which cooperate with the memory device 100 and the correction means 84.

These correction means 84 receive the signals delivered by the photosensitive dots during an operation of reading a useful image to be corrected.

These signals are delivered by the analog-to-digital converter 70.