Device for processing real-time PCR measurements with a homogeneous reaction medium in which the PCR reagents can diffuse

Abstract

The invention relates to a device for processing real-time PCR measurements with a homogeneous reaction medium in which the PCR reagents can diffuse, which comprises a memory arranged to receive a plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, an analyser arranged to produce a PCR activity image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber with a homogeneous reaction medium in which PCR reagents can diffuse, the PCR activity image having a plurality of activity regions in each of which PCR activity is determined as likely on the basis of the fluorescence value of the pixels of these activity regions in the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, the pixels of the activity regions each receiving a weighting value associated with the likelihood of a PCR reaction such that the sum of all the weighting values is one, and the pixels outside the activity regions receiving a zero value, and an evaluator arranged to analyse the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber by weighting in each image the fluorescence value of each pixel by the weighting value of that pixel in the activity image, and to return one or more of a value indicating whether the filmed real-time PCR reaction exhibits PCR amplification, a threshold cycle value, and a concentration curve during the PCR reaction.

Description

Description Title of the invention: Device for processing real-time PCR measurements with a homogeneous reaction medium in which the PCR reagents can diffuse

[0001] The invention relates to the field of real-time PCR ("Polymerase Chain Reaction").

[0002] PCR can very sensitively detect the presence of nucleic acid sequences in a sample. However, it is rarely used for field applications because the reaction time – approximately 1.5 hours – is relatively slow.

[0003] Indeed, PCR amplification is carried out with cycles whose temperature profile includes one or more steps during the temperature rise towards the melting temperature of around 95 °C. The temperature steps are chosen between the melting temperatures of the probes, and the fluorescence signal is recorded at each of these steps. The speed of the reaction is therefore limited by the duration of each cycle which is directly linked to the speed of temperature changes in the reaction chamber.

[0004] To overcome this problem, very fast "micro heaters" (about 20 minutes for 40 cycles) have been developed. However, the volume treated is very small (about 10 pl), which limits the LOD (for "Limit of Detection"), and therefore the ability to detect small amounts of nucleic acid.

[0005] Droplet PCR systems (see for example the article by Hindson et al "High-Throughput Droplet Digital PCR System for Absolute Quantitation of DNA Copy Number", Analytical Chemistry 2011 83 (22), 8604-8610) have been developed, but they have the disadvantage that the temperature is not homogeneous in the reaction chamber and they are complicated to implement.

[0006] In these digital PCR systems, the PCR is compartmentalized into defined areas (droplets) so that the PCR signal is digitized to allow quantification of the target DNAs individually by counting the drops producing a positive PCR. This quantification is carried out by segmenting the image of the drops at the end of the reaction to determine their position and from this position individually measuring their fluorescence level to determine whether the content of the droplet is PCR positive or not. The algorithms used to analyze digital PCR images such as droplet PCR measure the activity of the PCR reaction in a predefined area.

[0007] These algorithms are not suitable for analyzing the activity of a PCR performed in a homogeneous medium such as a liquid or gel where PCR activity can take place. in an area whose size and shape are not defined a priori and can potentially change over time due to the diffusion of PCR reaction products in the medium. Because of these specificities, the field of digital PCR remains an isolated field, and its lessons are not used in other areas of PCR.

[0008] The Applicant has developed a solution in response to these problems, a solution which was the subject of patent applications FR1601823 and FR1762058. These patent applications teach the use of chambers which have the characteristic of using flat chips allowing a homogeneous heat exchange over the entire surface and of low thickness so that the temperature variations are rapid over the entire volume.

[0009] However, these solutions tend to limit convection movements within the reagent volume. Local amplifications do not propagate, especially in the case of low concentrations of the desired F DNA. The local signal is then attenuated by that of the entire chip, which limits the sensitivity of PCR tests carried out on these chips.

[0010] These solutions present a homogeneous reaction medium in which the PCR reagents can diffuse, as opposed to the digital PCR mentioned above. A homogeneous medium is defined as a liquid reagent or gel in which the PCR reaction can diffuse freely and thus form zones of activity of indeterminate shape. This means that the zones in which amplification takes place are of indeterminate shape, unlike drops whose shape is known.

[0011] No state-of-the-art solution therefore allows real-time PCR to be performed both very quickly and with single-molecule sensitivity.

[0012] The invention improves the situation. To this end, it proposes a device for processing real-time PCR measurements with a homogeneous reaction medium in which the PCR reagents can diffuse, comprising a memory arranged to receive a plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, an analyzer arranged to produce a PCR activity image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber with a homogeneous reaction medium in which the PCR reagents can diffuse, which PCR activity image has a plurality of activity zones in each of which a PCR activity is determined as probable on the basis of the fluorescence value of the pixels of these activity zones in the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber,the pixels of the activity zones each receiving a weighting value associated with the probability of a PCR reaction such that the sum of all the weighting values ​​is one, and the pixels outside the activity zones receiving one, zero value, and an evaluator arranged to analyze the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber by weighting in each image the fluorescence value of each pixel by the weighting value of that pixel in the activity image, and to return one or more of a value indicating whether the filmed real-time PCR reaction exhibits PCR amplification, a threshold cycle value, and a concentration curve during the PCR reaction

[0013] This device is particularly advantageous because it allows the detection of areas of PCR activity and therefore the reprocessing of the fluorescence signal without being penalized by the absence of propagation of local amplifications due to limitations of convection movements inside the reagent volume.

[0014] According to various embodiments, the invention may have one or more of the following characteristics: - the analyzer is arranged to determine a difference image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, in which each pixel corresponds to the difference between the highest value and the lowest value for this pixel among the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, and to determine the activity image from the difference image, - the analyzer is arranged to determine activated areas in the difference image by thresholding the difference image, - the analyzer is arranged to determine a unique weighting value for each of the pixels in the PCR activity image, - the analyzer is arranged to determine the weighting value by applying a fixed threshold binarization function, Otsu, Bemsen, or Niblack, - the analyzer is arranged to apply a binarization function including the regularization of the activity zones by application of dilation, erosion, or convex envelopes, - the analyzer is arranged to apply a binarization function in which the expansion of the activity zones is obtained by adding a pixel around each activity zone until the sum of the areas of the activity zones exceeds a chosen threshold, or to apply a binarization function in which the erosion of the activity zones includes the deletion of the activity zones whose number of pixels is less than a chosen threshold, - the analyzer is arranged to determine the weighting value by dividing the value of each pixel belonging to an activity zone by the sum of all the pixel values ​​of all the activity zones, - the analyzer is a neural network of the mask R-CNN type or of the U-Net type performing semantic segmentation of pixels from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, and - the evaluator is arranged to determine a threshold cycle value for each activity zone, and to calculate a threshold cycle value according to the formula , where St is the total area of ​​the activity zones, Si is = iog2 the area of ​​each activity zone, and Cti is the threshold cycle value of the activity zone with index i.

[0015] The invention also relates to a method for processing real-time PCR measurements with a homogeneous reaction medium in which the PCR reagents can diffuse comprising: a) receiving a plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber with a homogeneous reaction medium in which the PCR reagents can diffuse, b) producing a PCR activity image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, which PCR activity image has a plurality of activity zones in each of which PCR activity is determined to be probable on the basis of the fluorescence value of the pixels of these activity zones in the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber,the pixels of the activity zones each receiving a weighting value associated with the probability of a PCR reaction such that the sum of all the weighting values ​​is one, and the pixels outside the activity zones receiving a zero value, c) analyzing the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber by weighting in each image the fluorescence value of each pixel by the weighting value of this pixel in the activity image, and to return one or more of a value indicating whether the filmed real-time PCR reaction exhibits PCR amplification, a threshold cycle value, and a concentration curve during the PCR reaction.,

[0016] According to various embodiments, the method may have one or more of the following characteristics: - operation b) comprises b 1) determining a difference image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, in which each pixel corresponds to the difference between the highest value and the lowest value for this pixel among the plurality of fluorescence images obtained by filming a real-time PCR reaction in a chamber plate, and b2) determine the activity image from the difference image, - operation b) comprises determining a single weighting value for each of the pixels of the PCR activity image, or determining the weighting value by dividing the value of each pixel belonging to an activity area by the sum of all the pixel values ​​of all the activity areas, - operation b) is obtained by applying a neural network of the R-CNN mask type or of the U-Net type carrying out a semantic segmentation of the pixels from at least some of the images of operation a), and - operation c) includes determining a threshold cycle value for each activity zone, and calculating a threshold cycle value according to the formula t is the total area of ​​the activity zones, Si is Cf is the threshold cycle value of the activity zone of index i.

[0017] The invention also relates to a computer program comprising instructions for executing the method according to the invention, a data storage medium on which such a computer program is recorded and a computer system comprising a processor coupled to a memory, the memory having recorded such a computer program.

[0018] Other characteristics and advantages of the invention will appear more clearly on reading the following description, taken from examples given for illustrative and non-limiting purposes, taken from the drawings in which: - [Fig.l] represents a generic diagram of a device according to the invention, - [Fig.2] represents an example of the realization of a function implemented by the analyzer of [Fig.l], - [Fig.3] represents an example of the realization of a function implemented by the evaluator of [Fig.l], - [Fig.4] represents a fluorescence image obtained by filming a real-time PCR reaction in a flat chamber, - [Fig.5] represents an intermediate image of cycle 10 determined by the analyzer of [Fig.l], - [Fig.6] represents an intermediate image of cycle 41 determined by the analyzer of [Fig.l], - [Fig.7] represents a difference image obtained from intermediate images of the type of those in figures 5 and 6, - [Fig.8] represents a result image obtained by the evaluator of [Fig.l] from the image of [Fig.7], - [Fig.9] represents an amplification signal obtained by combining the image of [Fig.8] with the images of the corresponding film, and - [Fig.10] represents a U-Net neural network architecture according to one of the variants of the invention.

[0019] The drawings and the description below contain, for the most part, elements of a certain character. They may therefore not only serve to better understand the present invention, but also contribute to its definition, if necessary.

[0020] [Fig.1] represents a generic diagram of a device according to the invention.

[0021] The device 2 comprises a memory 4, an analyzer 6 and an evaluator 8.

[0022] Memory 4 receives several types of data. It receives fluorescence images obtained by filming a real-time PCR reaction in a flat chamber with a homogeneous reaction medium in which the PCR reagents can diffuse. Taken together, these images represent the film of the fluorescence emitted following the PCR cycles carried out. A homogeneous medium is understood to mean a liquid reagent or a gel where the PCR reaction can diffuse freely and thus form zones of activity of indeterminate shape. This means that the zones in which amplification takes place are of indeterminate shape, which is why the fluorescence is conventionally averaged over the entire chamber to measure the reaction.

[0023] In the example described here, there are approximately 45 cycles in each real-time PCR reaction, and 45 corresponding images, each associated with a PCR cycle of the reaction. Alternatively, there could be a different number of cycles in each reaction, and multiple images could be associated with a single cycle within a given real-time PCR reaction, allowing for the storage of richer data about the PCR reaction in each cycle. In the following, a movie will be referred to as the set of images of the PCR cycles of a given real-time PCR reaction, and when comparing two images with each other, it is implied that they will be images relating to the same real-time PCR reaction.

[0024] The analyzer 6 processes the films to produce for each film a PCR activity image in which each pixel has a color associated with a value (between zero and one) which indicates a probability that a PCR reaction has been detected for that pixel. Alternatively, the value and the color could be unrelated. The set of pixel values ​​has a sum of one, so that the pixel values ​​together define a probability density. This point will be explained later. This PCR activity image can be obtained in various ways which will be described later, among which the evaluator 6 produces a difference image, which corresponds to the difference between an average of images associated with high PCR activity and an average of images associated with low PCR activity. Again, this will be explained later.

[0025] All these images are suitable for storage in memory 4.

[0026] Finally, memory 4 also receives real-time PCR reaction data, which may indicate whether a nucleic acid sequence has been detected and / or specify a cycle from which the real-time PCR reaction is considered to have revealed this detection. This latter value is also known as "Ct" in the field of real-time PCR. Generally speaking, the Ct value is an indicator of the quantity of the desired nucleic acid sequence that has been detected. Indeed, the more material there is to be detected in a sample, the more the PCR reaction will produce a large quantity of detected material from a cycle whose index is low (i.e., close to the start of the reaction). Thus, a Ct of 15 indicates that there was much more material of the desired nucleic acid sequence in a given sample than a Ct of 30.

[0027] The memory 4 can be any type of data storage suitable for receiving digital data: hard disk, flash memory hard disk, flash memory in any form, random access memory, magnetic disk, locally or cloud distributed storage, etc. The data calculated by the device can be stored on any type of memory similar to the memory 4, or on it. This data can be erased after the device has performed its tasks or retained.

[0028] The invention is based on the principle described below. Due to limitations in convection movements within the reagent volume in the flat chamber, some portions of the chamber do not show a reaction when they should. As a result, the fluorescence signal, which is averaged over the entire surface of the flat chamber, is artificially lowered, which decreases the LOD.

[0029] Device 2 circumvents this problem by analyzing all the frames of a given film, and extracting an image on which the detection of a PCR reaction is encoded via a probability density. This then allows the film to be reprocessed taking into account this probability density to carry out the detection. Thus, only the portions of the flat chamber that are relevant are taken into account, and the probability density allows the relative contribution of each portion to the final result to be normalized.

[0030] For this, the analyzer 6 processes a given film to derive a PCR activity image in which the pixels of a plurality of activity zones in each of which PCR activity is determined to be probable each receive a weighting value associated with the probability of a PCR reaction such that the sum of all the weighting values ​​is equal to one, and the pixels outside the activity zones receive a zero value.

[0031] Then, the evaluator 8 uses the PCR activity image and combines the weighting values ​​with the given movie in order to analyze the relevant areas of the images to derive detection information from a nucleic acid sequence and / or a Ct type value.

[0032] The analyzer 6 and the evaluator 8 directly or indirectly access the memory 4. They can be implemented in the form of appropriate computer code executed on one or more processors. By processors, it is meant any processor suitable for the calculations described below. Such a processor can be implemented in any known manner, in the form of a microprocessor for a personal computer, laptop, tablet or smartphone, a dedicated chip of the FPGA or SoC type, a computing resource on a grid or in the cloud, a cluster of graphics processing units (GPUs), a microcontroller, or any other form suitable for providing the computing power necessary for the implementation described below. One or more of these elements can also be implemented in the form of specialized electronic circuits such as an ASIC. A combination of processor and electronic circuits can also be envisaged.In the case of the gradient boosting machine learning unit, dedicated machine learning processors could also be considered.

[0033] The purpose of the analyzer 6 is therefore to determine areas of interest in the images of a given film, and to qualify these areas of interest in order to restore them in the form of a probability density.

[0034] To implement analyzer 6, several solutions are available: - use of image of differences between fluorescences according to the different cycles, - use of deep neural networks dedicated to object segmentation in mask R-CNN type images (described below), and - use of semantic segmentation neural networks directly giving an image of the classes sought. In the example described here, these classes can include active PCR area, inactive PCR area, exterior of the flat chamber, feeding hole, bubbles, strong light points.

[0035] Indeed, apart from the areas of active or inactive PCR due to convection limits, certain artifacts can disturb the fluorescence images. For example, the camera used to film the reaction has a larger field than the flat chamber, and therefore there is an area that is not relevant by definition. In addition, anomalies are regularly observed at the hole used to introduce the reaction elements into the flat chamber. Similarly, certain imperfections in the flat chamber induce points of strong brightness. Finally, bubbles can form during the reaction. Bubbles are the most troublesome artifact, because they are not stable during the PCR reaction, and their border is often brighter than the rest of the sample.

[0036] The first solution requires separate treatment for each class other than the active PCR zone in order to exclude the corresponding zones.

[0037] The second and third solutions have the advantage of being direct, but neural networks are still difficult to make efficient without overfitting.

[0038] [Fig.2] shows an example of a function implemented by parser 6 that uses the first solution.

[0039] In the first solution, the analyzer 6 operates in four main stages - comparison of images from a given film to create a maximum fluorescence deviation image, - filtering this image to determine areas of interest, - filtered image processing to compensate for edge effects and to remove artifacts, and - modification of the resulting image to produce a probability density with the remaining areas of interest.

[0040] Thus, in a first operation 200, the images of a given film are processed to create the maximum fluorescence deviation image. [Fig. 4] represents an example of the film images received as input. When several images are available for the same cycle of a given film, an average of images taken during the cold phase of each cycle is used to produce an image representative of each cycle. This makes it possible to limit the acquisition noise of the camera, and is preferentially done with the images corresponding to the cold phase of the cycle, which is known to have more stable fluorescence. The exact number of images to calculate the average as well as the exact period of the cycle where they are taken can be parameters or be fixed. Carrying out several experiments can make it possible to calibrate the device 2 for this purpose. Figures 5 and 6 represent examples of these images for cycles 10 and 41 respectively.

[0041] Then, in an operation 210, a maximum fluorescence image and a minimum fluorescence image are generated. For the maximum fluorescence image, the images from operation 200 associated with index cycles 25 to 45 are combined so that, for each pixel, it is the fluorescence value among the 20 images that is retained. Similarly, the minimum fluorescence image is obtained by retaining for each pixel the minimum fluorescence in the images from operation 200 corresponding to cycles 5 to 15. Of course, the index ranges for determining the maximum fluorescence image and the minimum fluorescence image may vary, both in their ends and in their width. Once these two images have been calculated, their difference is realized, and represents for each pixel the maximum fluorescence difference during the most relevant cycles. [Fig.7] represents a difference image obtained at the output of operation 220.

[0042] This difference image is processed in a 220 operation to determine the contours of the areas of interest. This can be achieved, for example, by thresholding at two fluorescence measurement units. Alternatively, the threshold can be changed. In addition, alternative methods could be used to select the areas of interest, such as a Niblack-type or Bemsen-type method.

[0043] The thresholded image from operation 220 is then processed in operation 230 to remove artifacts. As explained above, this processing may include canceling areas that are outside the flat chamber, detecting high-brightness spots and the flat chamber inlet area, and removing bubbles. For the flat chamber inlet area, this may be determined empirically by performing multiple calibration cycles or may be subject to ad hoc detection.

[0044] Bubbles are determined by detection in the film frames before processing by the analyzer 6. For this, the frames are processed by increasing cycle index, and each pixel is qualified in each frame as belonging to a bubble or not. When a pixel is qualified in any frame as belonging to a bubble, it is added to a list of bubble pixels.

[0045] Bubble membership detection can be achieved by applying an algorithm in which, for a given image: - a Canny filter is applied to determine the object borders, - connected components induced by the found borders are created - the presence of bubbles is detected by borders surrounding them, and a closed border (forming a loop) determines two different connected components,

[0046] - each connected component is considered a bubble if its average gray level is below a threshold determined dynamically with respect to the fluorescence unit values ​​of the flat chamber, and - the connected component is expanded so that the bubble area contains the edge.

[0047] The pixels in the bubble pixel list are thus canceled, as are other artifacts. Alternatively, the bubble pixel list could be determined at operation 200 or 210.

[0048] At the end of this operation, the image of the differences therefore includes pixels whose value is not zero, which represent the areas of interest, and all the other pixels have a zero value.

[0049] Finally, the processed image from operation 230 is modified in an operation 240 in order to construct the probability density. For this, the values ​​of the difference image can be processed in two ways: - creation of a variable probability density, by dividing for each pixel of a area of ​​interest the value of this pixel by the sum of all the values ​​of the pixels of the areas of interest, - application of a binarization function, so that all pixels have the same weight. Among the binarization functions, we can use a fixed threshold method, Otsu, Bemsen, or Niblack.

[0050] In the following, the term "area of ​​interest" and "area of ​​activity" may be used interchangeably, since areas of interest correspond to areas in which PCR activity is likely.

[0051] In case a binarization function is used, the areas of interest can be smoothed by applying one or more dilation, erosion or convex hull operations. The idea of ​​this smoothing is that the areas of interest correspond to a location where the PCR reaction took place, and are therefore normally closed.

[0052] For example, dilating areas of interest can involve adding a pixel around the border of each area until the total area of ​​the areas exceeds a chosen threshold.

[0053] Still as an example, erosion can provide for the deletion of areas whose number of pixels is lower than a given threshold (for example 2).

[0054] Once operation 240 is complete, the resulting PCR activity image is transmitted to evaluator 8 which will combine it with the film images to return a detection result and / or a Ct. [Fig.8] represents an example of an image obtained at the output of operation 240. As can be seen in this figure, the edges of the flat chamber delimit a corner.

[0055] As described above, Analyzer 6 can also be implemented using a Mask R-CNN (for "Region Based Convolutional Neural Networks"). This type of neural network allows to detect regions in an image and to perform classification of the detected regions. The article by He et al "Mask R-CNN", arXiv: 1703.06870, accessible at https: / / doi.org / 10.48550 / arXiv.1703.06870 describes how to implement them.

[0056] In the example described here, the training database is built from PCR data performed on CHRONOS Dx machines from BforCure. To ensure variability in the training data, there are 20 machines. Training and testing data are separated by machine, so that data from a machine used for training is not used for testing.

[0057] These data are annotated using dedicated software according to a semantic mask that uses a value per class considered for each pixel of the image. These data are pre-processed to be put into a format compatible with the algorithm training of each type of network.

[0058] Given the nature of the data, the mask RCNN network includes a backbone corresponding to a Resnet-50 (see for example https: / / web.archive.Org / web / 20211227064145 / https: / / iq.opengenus.org / resnet50-archit ecture / for a description of Restnet-50), whose inputs are the same as those of the U-Net network described below, i.e. a stack of 20 images from different cycles. The only modification compared to a classic mask RCNN is that the input does not have 3 channels but 20. The network cannot therefore be re-learned from a pre-trained network. The training is carried out as in the article by He cited above.

[0059] Post-processing may be required at the network output, for example, when some pixels are covered by multiple object instances or none at all. For pixels that are not covered by any object instances, if the pixel is within the flat chamber area, it is assigned the class “inactive PCR area,” and if the pixel is outside the flat chamber area, it is assigned the class “outside the flat chamber.” For pixels covered by multiple object instances, they are assigned the class with the highest score.

[0060] Similar to the reference article, we apply data augmentation methods: rotations, pixel noise, and elastic deformations. Moreover, since our input includes 20 PCR cycles out of about forty (value depending on the reagent kits used), we can augment our data by shifting the cycles proposed as input during training.

[0061] In this case the images of each film are provided to the analyzer 6 which returns the semantically classified areas.

[0062] Still alternatively, analyzer 6 could be implemented using a simplified U-Net type neural network. The article by Ronneberger et al. "U-Net: Convolutional Networks for Biomedical Image Segmentation", Medical Image Computing and Computer-Assisted Intervention (MICCAI) LNCS 9351, pages 234—241 (2015), Springer available at https: / / arxiv.org / pdf / 1505.04597.pdf describes the implementation of this type of neural network. In the present case, the Applicant has identified that it is favorable to use a 3-layer U-Net instead of the 5 of the original model, with 20 input channels each corresponding to an image of a cycle distributed during the real-time PCR reaction from which the film is taken. The U-Net network was trained using images and data from 1000 real-time PCR reactions, 10% of which were retained for testing.As in the reference article cited above, it is advantageous to use data augmentation methods: rotations, noise on pixels and elastic deformations. The rebalancing of the different classes considered is ensured by a weighting map which also takes into account the distance of the pixels to the border zones. [Fig.10] represents an example of implementation of the U-Net network architecture described here.

[0063] The output of the mask R-CNN and the U-NET network is each time a list of semantically classified areas, with the probability density for the class "active PCR area".

[0064] [Fig.3] represents an example of the realization of a function implemented by evaluator 8.

[0065] This function begins with a mixing operation 300 in which the film images and the PCR activity image determined by the evaluator 6 are combined. In the example described here, the value of each pixel in each film image is weighted by the value of the pixel that corresponds to it in the PCR activity image.

[0066] Then, in an operation 320, the evaluator 8 applies an algorithm on the resulting images to each area of ​​interest or to each pixel, in the same way as this is done for the entire image in the state of the art.

[0067] The output of this algorithm may be a value indicating whether or not an amplification has been detected and / or a cycle index value from which the amplification is significant. This algorithm is known elsewhere. The output of this algorithm may also be the concentration curve shown in [Fig. 9]. In this figure, the continuous curve represents the concentration curve measured using the device of the invention, while the dotted curve represents the concentration measured using the state-of-the-art method.

[0068] It is therefore possible to obtain a Ct value for each area of ​​interest, or even for each pixel whose associated probability in the probability density of the PCR activity image is non-zero.

[0069] Optionally, evaluator 8 can return a Ct value for the reaction Real-time PCR from these local Ct values.

[0070] Thus, the Ct value can be calculated as , where St is the total area of ​​the areas of interest (e.g. in pixels), Si is the area of ​​each area of ​​interest (e.g. in pixels), and Cf is the Ct value of the area of ​​interest with index i. This approach can be taken to the extreme by reducing each area of ​​interest to a single pixel.

[0071] As the concentration of the desired nucleic acid sequence doubles at each cycle, the concentration of this sequence is therefore proportional to 2 Ct .

[0072] Figures 4 to 6 illustrate the images obtained by implementing the device 2 with an analyzer 6 conforming to that described with [Eig.2],

[0073] Thus, the PCR activity image is obtained as follows: - for each cycle, determination of a fluorescence image corresponding to the cold phase of the cycle. This image is obtained by averaging the images taken during this phase, taken from the second half of the cycle, - construction of the difference image from an image of the maximum fluorescences of the last cycles, from which an image of the minimum fluorescences of the first cycles is subtracted. The cycles used for the calculation of this image are cycles 5 to 15 for the minimum image and cycles 26 to 45 for the maximum image, - transformation of the difference image into a binary image by applying a fixed threshold of 2 fluorescence values ​​with application of erosion, a convex hull to "fill the holes", and dilation until obtaining a total surface of at least 1000 pixels, - removal of areas of the image that do not correspond to the flat chamber with which the PCR is performed, and - removal of areas that correspond to bubbles.

[0074] Thus, starting from [Fig.6], which represents the averaged image for cycle 41, we obtain an image of the differences represented in [Fig.7]. Finally, the thresholding operation gives the image in [Fig.8]. This image is combined with the film images to produce [Fig.9], which shows a Cti at 35.6849, which gives a Ct of 38.363 by applying the formula explained previously.

Claims

Demands

1. A real-time PCR measurement processing device with a homogeneous reaction medium in which PCR reagents can diffuse, comprising: - a memory (4) arranged to receive a plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, - an analyzer (6) arranged to produce a PCR activity image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber with a homogeneous reaction medium in which the PCR reagents can diffuse, which PCR activity image has a plurality of activity zones in each of which PCR activity is determined to be probable on the basis of the fluorescence value of the pixels of these activity zones in the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, the pixels of the activity zones each receiving a weighting value associated with the probability of a PCR reaction such that the sum of all the weighting values ​​is one, and the pixels outside the activity zones receiving a value of zero, - an evaluator (8) arranged to analyze the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber by weighting in each image the fluorescence value of each pixel by the weighting value of that pixel in the activity image, and to return one or more of a value indicating whether the filmed real-time PCR reaction exhibits PCR amplification, a threshold cycle value, and a concentration curve during the PCR reaction.

2. Device according to claim 1, wherein the analyzer (6) is arranged to determine a difference image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, wherein each pixel corresponds to the difference between the highest and lowest values ​​for that pixel among the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, and to determine the activity image from the difference image.

3. Device according to claim 2, wherein the analyzer (6) is arranged to determine areas of activity in the difference image by thresholding the difference image.

4. Device according to claim 2 or 3, wherein the analyzer (6) is arranged to determine a unique weighting value for each of the pixels of the PCR activity image.

5. Device according to claim 4, wherein the analyzer (6) is arranged to determine the weighting value by applying a fixed threshold binarization function, of Otsu, Bemsen, or Niblack.

6. Device according to claim 4 or 5, wherein the analyzer (6) is arranged to apply a binarization function comprising the regularization of activity zones by application of dilation, erosion, or convex envelopes.

7. Device according to claim 6, wherein the analyzer (6) is arranged to apply a binarization function in which the expansion of the activity zones is achieved by adding a pixel around each activity zone until the sum of the areas of the activity zones exceeds a chosen threshold, or to apply a binarization function in which the erosion of the activity zones includes the removal of activity zones whose number of pixels is less than a chosen threshold.

8. Device according to claim 2 or 3, wherein the analyzer (6) is arranged to determine the weighting value by dividing the value of each pixel belonging to an activity zone by the sum of all the pixel values ​​of all the activity zones.

9. Device according to claim 1, wherein the analyzer (6) is an R-CNN type mask neural network or a U-Net type neural network performing semantic segmentation of pixels from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber.

10. A device according to any one of the preceding claims, wherein the evaluator (8) is arranged to determine a threshold cycle value for each zone and calculate a threshold cycle value according to the formula where St is the surface total of activity zones, Si is the area of ​​each activity zone, and Cf is the threshold cycle value of the activity zone with index i.

11. A method for processing real-time PCR measurements in a homogeneous reaction medium in which the PCR reagents can diffuse comprising: a) receiving a plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber with a homogeneous reaction medium in which the PCR reagents can diffuse, b) producing a PCR activity image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, which PCR activity image exhibits a plurality of activity zones in each of which PCR activity is determined to be probable based on the fluorescence value of the pixels in those activity zones in the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, the pixels in the activity zones each receiving a weighting value associated with the probability of a PCR reaction such that the sum of all the weighting values ​​is one, and the pixels outside the activity zones receiving a value of zero,c) analyze the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber by weighting in each image the fluorescence value of each pixel by the weighting value of that pixel in the activity image, and to return one or more of a value indicating whether the filmed real-time PCR reaction exhibits PCR amplification, a threshold cycle value, and a concentration curve during the PCR reaction.

12. A method according to claim 11, wherein operation b) comprises: b1) determining a difference image from the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, in which each pixel corresponds to the difference between the highest and lowest values ​​for that pixel among the plurality of fluorescence images obtained by filming a real-time PCR reaction in a flat chamber, and b2) determining the activity image from the difference image.

13. A method according to claim 12, wherein operation b) comprises determining a unique weighting value for each of the pixels in the PCR activity image, or determining the weighting value by dividing the value of each pixel belonging to an activity area by the sum of all the pixel values ​​of all the activity areas.

14. A method according to claim 11, wherein operation b) is achieved by applying an R-CNN mask-type neural network or a U-Net-type neural network performing semantic segmentation of pixels from at least some of the images of operation a).

15. A method according to any one of claims 10 to 14, wherein operation c) comprises determining a threshold cycle value for each activity zone, and calculating a threshold cycle value according to the formula, where St is the total area of ​​the zones of activity, Si is the area of ​​each activity zone, and Cf is the threshold cycle value of the activity zone with index i.

16. Computer program product, characterized in that it is arranged to implement the method according to any one of claims 11 to 15 when executed by a computer.

Images