Tumor imaging device

Abstract

The imaging device comprises a memory 10 designed to receive imaging data 12, biopsy data 16 including needle data 160 defining the dimensions of a puncture made using a biopsy needle, and procedure data 162, an oversampler 20 designed to oversample the matrix image 14 into a processed image 22, a cutter 30 designed to divide the processed image 22 from the oversampler 20 into a set of regions 32, a selector 40 designed to identify a subset 42 of diffusion parameters within a set 36 of diffusion parameters and return a subset 44 of regions extracted from the set 32, and a guide 50 designed to identify from the subset 44 of regions data relating to a set of puncture parameters 52, puncture orientation and puncture entry point that together with the needle data 160 define a puncture zone.

Description

[Technical Field]

[0001] The present invention relates to the field of medical imaging, and more particularly to tumor imaging. Tumor imaging encompasses non-invasive methods for analyzing tumors. The analysis can be aimed at identifying the extent of cancer cells within the tumor. Tumor imaging is performed by magnetic resonance imaging (MRI), particularly diffusion-weighted magnetic resonance imaging. However, these two imaging techniques currently do not provide accurate information about tumors within a patient's body, such as the absolute distribution of cancer cells or tumor burden. Here, tumor burden refers to the total number of tumor cells of all types or of each individual tumor cell type, particularly of a cancer cell type. [Background technology]

[0002] Biopsy is known as an invasive method used to supplement or replace tumor imaging. A biopsy is a localized sampling of tissue with a biopsy needle. Biopsies provide localized information about tumors that is not accessible by non-invasive techniques. However, biopsies are localized samplings, and the sampled tissue is usually not representative of the tumor being studied. Furthermore, biopsies are invasive surgical procedures and present medical risks such as tumor expansion, spread of cancer cells to normal tissues, bleeding, infection, or nerve damage.

[0003] There is no method to identify representative locations for biopsy, and as a result, it is currently necessary to perform multiple biopsies, sometimes as many as 10, to increase the likelihood of obtaining sufficient information to identify the desired absolute distribution or tumor burden information. Summary of the Invention

[0004] The present invention improves the above situation. To this end, the present invention provides: a memory, imaging data including at least one raster image derived from a cross-sectional diffusion-weighted magnetic resonance image of a tumor, each pixel of the raster image being associated with a diffusion parameter whose value represents the mobility of water molecules within the tumor; biopsy data including needle data defining the dimensions of a puncture made by the biopsy needle; procedure data including two or more numbers of punctures to be performed in a biopsy procedure and a set of constraints related to intervention constraints that must be satisfied to achieve the number of punctures performed; a memory configured to receive the an upsampler configured to upsample the raster image into a processed image, the spatial resolution of which is such that each pixel of the processed image corresponds to a substantially square area of ​​the cross-sectional image, the side of which has a length that is less than a needle diameter dimension of the needle data; and a slicer configured to divide the processed image from the upsampler into a set of regions, each region comprising: at least two pixels of the processed image immediately adjacent to each other; associated with a diffusion parameter whose value is substantially equal to the average of the values ​​of the diffusion parameters associated with its constituent pixels; the variance of the values ​​of the diffusion parameter associated with the pixels of the region is less than a given variance value; a portion of the cross-sectional diffusion-weighted magnetic resonance image of the tumor, the portion having a dimension greater than or substantially equal to the dimension of a puncture made by a biopsy needle of the biopsy data; a slicer, the diffusion parameters associated with each region of the set of regions forming a set of diffusion parameters; a selector, identifying a subset of diffusion parameters from the set of diffusion parameters, the subset including the same number of diffusion parameters as the number of punctures performed in the procedure data, wherein the subset of diffusion parameters includes at least a first diffusion parameter selected from a first decile of diffusion parameter values ​​in the set of diffusion parameters, a second diffusion parameter selected from a last decile of diffusion parameter values ​​in the set of diffusion parameters, and a diffusion parameter whose value is between the value of the first diffusion parameter and the value of the second diffusion parameter when the number of punctures performed in the procedure data is more than two; returning a subset of regions derived from the set of regions, each region associated with a diffusion parameter substantially equal to one of the parameters of the subset of diffusion parameters; a selector configured to: a guide configured to identify, from the subset of regions, a number of puncture parameter set data equal to the number of punctures performed in the procedure data, each set of puncture parameters including a puncture depth, a puncture direction and a puncture entry point, the guide defining, together with the needle data, a puncture zone that is substantially included in one of the regions of the subset of regions such that together they satisfy a set of constraints of the procedure data; An imaging device is provided, comprising:

[0005] This imaging device is highly advantageous since it allows automatically identifying a reduced number of biopsy locations that guarantee relevant information: indeed, the punctures identified by the imaging device are far more representative of the entire tumor than randomly performed punctures, while at the same time being fewer in number.

[0006] In various alternative forms, the device may have one or more of the following features: The upsampler is configured to upsample the raster image to a processed image by performing interpolation. The upsampler is configured to perform bicubic interpolation to upsample the raster image to a processed image. The slicer is configured to split the processed image from the upsampler by a superpixel method. The number of punctures performed in the biopsy procedure of the procedure data stored in the memory is four or less. The selector selects a first diffusion parameter from the 2nd percentile of diffusion parameter values ​​of the set of diffusion parameters and a second diffusion parameter from the 98th percentile of diffusion parameter values ​​of the set of diffusion parameters. The first diffusion parameter and the second diffusion parameter are each equal to the median of their respective quartiles. When the number of punctures performed in the procedure data is greater than two, the selector selects diffusion parameters of the subset of diffusion parameters such that the diffusion parameters are equidistant from each other in pairs. When the number of punctures performed in the procedure data is greater than two, the selector identifies a median value of the diffusion parameter of the set of diffusion parameters, and identifies a diffusion parameter j, where j is a natural number of the subset of diffusion parameters that is between 1 and the number of punctures performed in the procedure data, according to the following formula: Dj=Dini+Delta(j-1) Here, Dini satisfies the following equation: Dini = M-min(|M-Dmin|,|M-Dmax|) Delta satisfies the following formula: Delta=2(M-Dini) / (N-1) Dmin is equal to the value of a first diffusion parameter specified by the selector, and Dmax is equal to the value of a second diffusion parameter of the subset of diffusion parameters specified by the selector. The set of puncture parameters identified by the guide, each including a puncture depth, a puncture direction, and a puncture entry point, together with the biopsy data, defines a puncture trajectory of the biopsy needle, and the set of procedure data constraints stored in the memory includes: - the punctures in the biopsy procedure must have the same entry point into the tumor; -The puncture trajectory must be in a straight line and within the maximum puncture angle. - The puncture depth is less than the maximum puncture depth; - the puncture zone is spaced from the edge of the tumor by a predetermined distance; - if the tumor contains a necrotic zone, the puncture trajectory does not penetrate this necrotic zone; - if the tumor contains an adipose tissue zone, none of the puncture zones intersect with this adipose tissue zone; at least one of the entry points of the puncture parameter set data is fixed in advance; and one or more constraints selected from the group including: the memory is further configured to receive puncture data including a set of cell densities, each cell density being associated with one of the puncture zones of the puncture parameter set data, each cell density having a value representative of the cell density of said puncture zone; The imaging device a correlator configured to associate each cell density value associated with one of the puncture zones with a diffusion parameter value of a set of regions in which said puncture zone is substantially included, and to derive correlation data associating the cell density values ​​with the diffusion parameter values ​​of the set of diffusion parameters; a constructor configured to determine density data from the processed image and the correlation data, the density data comprising a density image in which each pixel corresponding to an area of ​​the cross-sectional image is associated with a diffusion parameter value of a pixel of the processed image corresponding to said area of ​​the cross-sectional image of said pixel and the cell density to which that value is associated in the correlation data; Further provided with: the correlator performs a strictly monotonic linear or non-linear interpolation on pairs formed from diffusion parameter values ​​and cell density values ​​associated with the puncture zone, and derives correlation data from the results of said interpolation. The cell density of the puncture data and density data corresponds to the total cell density or the cancer cell density. The apparatus further comprises an estimator configured to identify a tumor burden of the tumor based on the density data.

[0007] Further features and advantages of the present invention are set forth in detail in the following description, taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic diagram of an imaging device according to the present invention. [Figure 2] 2 is a diffusion-weighted magnetic resonance raster image of a tumor received by the imaging device of FIG. 1. [Figure 3] 2 is a schematic diagram of a set of constraints received by the imaging device of FIG. 1; [Figure 4] 2 is a processed image obtained by the imaging device of FIG. 1. [Figure 5] FIG. 2 illustrates a set of regions identified by the imaging device of FIG. [Figure 6] FIG. 2 illustrates a set of diffusion parameters determined by the imaging device of FIG. [Figure 7] FIG. 2 is a diagram showing a set of puncture parameters identified by the imaging device of FIG. 1. [Figure 8] FIG. 2 illustrates a subset of the areas identified by the imaging device of FIG. [Figure 9] FIG. 8 shows a region that is a subset of the region of FIG. 7. [Figure 10] FIG. 2 illustrates another subset of the area identified by the imaging device of FIG. [Figure 11] FIG. 11 is a diagram showing details of FIG. [Figure 12] FIG. 2 shows an alternative to the imaging device of FIG. [Figure 13] 13 is a diagram illustrating a first linear interpolation performed by the imaging device of FIG. 12. FIG. [Figure 14] 13 is a diagram illustrating a second linear interpolation performed by the imaging device of FIG. 12. FIG. [Figure 15]13 is a diagram illustrating a third linear interpolation performed by the imaging device of FIG. 12. FIG. [Figure 16] FIG. 13 illustrates tumor burden estimation performed by the imaging device of FIG. [Figure 17] FIG. 17 shows the tumor volume from FIG. 16 as determined by the imaging device of FIG. 12. [Figure 18] FIG. 1 shows measured tumor volumes for randomly performed punctures that satisfy a set of constraints. DETAILED DESCRIPTION OF THE INVENTION

[0009] The accompanying drawings contain mostly definitive elements, and therefore not only serve to better understand the invention, but can also contribute, where appropriate, to its definition.

[0010] This specification is followed by Appendix A containing mathematical formulas, which is made an integral part of this specification.

[0011] Please refer to Figures 1 to 11.

[0012] The imaging device 1 according to the present invention includes a memory 10, an upsampler 20, a slicer 30, a selector 40, and a guide 50.

[0013] The memory 10 can be any type of data storage suitable for receiving digital data, i.e., a hard disk, a flash memory hard disk (SSD), any form of flash memory, random access memory, a magnetic disk, local or cloud distributed storage, etc. Data calculated by the device 1 can be stored in any type of memory similar to the memory 10 or in the memory 10. This data can be erased or retained after the device has performed its task.

[0014] The upsampler 20, the slicer 30, the selector 40 and the guide 50 are here programs executed by a computer processor. Alternatively, one or more of these elements may be implemented in a different way by a dedicated processor. By processor, any processor adapted for data processing as described below is to be understood. Such a processor may be realized in any known manner as a microprocessor of a personal computer, a dedicated chip of FPGA or SoC type, a computing resource on a grid, a microcontroller, or any other form suitable for providing the computing power required for the embodiments described below. One or more of these elements may also be realized in the form of specialized electronic circuits, such as ASICs. A combination of a processor and an electronic circuit may also be envisaged.

[0015] Memory 10 receives imaging data 12, including at least one raster image 14 from a cross-sectional diffusion-weighted magnetic resonance imaging (diffusion MRI or DWI) image of a tumor 200. Raster image 14 is also known as a "D-map." Diffusion-weighted magnetic resonance is a tissue imaging technique that measures the diffusion of water molecules within tissue. As such, each pixel of raster image 14 is associated with a diffusion parameter whose value indicates the mobility of water molecules within the tissue of tumor 200. The higher the value of the diffusion parameter, the greater the diffusion of water molecules, or in other words, the more freedom of water molecules to move within the tissue.

[0016] Diffusion parameter values ​​are highly dependent on the diffusion-weighted magnetic resonance imaging machine that generates them, making it generally impossible to analyze the parameter values ​​without additional information. In the present example, the raster image 14 shown in Figure 2 has pixels with sides corresponding to a substantially square zone with sides of 2.1 mm.

[0017] Our research has shown that diffusion values ​​correlate with cell density. Therefore, starting from raster images 14 derived from cross-sectional diffusion-weighted magnetic resonance images, it is theoretically possible to observe, at least qualitatively, the cellular heterogeneity within a tumor. However, the resolution of raster images provided by current diffusion-weighted magnetic resonance imaging devices is too low to show such heterogeneity in a useful way, and these images cannot be used in practice unless a total tumor removal is performed to see the correspondence. Total tumor removal is, of course, out of the question (and often impossible), especially since imaging and biopsy would be useless, even if possible.

[0018] In one embodiment, the imaging data 12 includes multiple raster images 14 representing successive cross sections of the tumor 200, thereby providing a three-dimensional view of the tumor 200. In the embodiment described herein, the imaging data 12 includes nine raster images 14. In another embodiment, the imaging data 12 includes forty raster images 14, from which nine raster images 14 are derived that represent sections of the tumor 200 closest to the median plane. One of these raster images 14 is shown in FIG. 2, where the scale on the right side of the figure indicates the correspondence between pixel intensity and diffusion parameter values.

[0019] Memory 10 also receives biopsy data 16 , including needle data 160 and procedure data 162 .

[0020] The needle data 160 defines the dimensions of the puncture performed with the biopsy needle. In the embodiment described herein, the puncture dimensions include a puncture diameter and a puncture length, and the sampling with the biopsy needle is a substantially cylindrical puncture zone defined by the diameter and length. In the example described herein, the puncture diameter is substantially 0.84 mm, and the puncture length is 4.55 mm ± 1.45 mm.

[0021] The procedure data 162 includes the number of punctures performed in the biopsy procedure. The number of punctures performed is an integer greater than or equal to 2. In practice, at least two punctures are required to obtain sufficient information about the tumor 200. In the embodiment described herein, the number of punctures is between 2 and 4, inclusive.

[0022] The procedure data 162 also includes a set of constraints 300. The set of constraints defines intervention constraints that must be met in a biopsy procedure that involves performing one or more punctures. The constraints in the set of constraints may arise from the type of organ in which the tumor 200 is located, the size of the tumor 200, the presence of necrotic tissue 500 within the tumor 200, or any other parameter that places constraints on the performance of the biopsy.

[0023] In the example described here, tumor 200 is a lung tumor, and the biopsy constraints associated with this tumor, shown in FIG. 3, are as follows: - the different biopsies must all have the same entry point 302 into the tumor 200; the different needle trajectories 304, 306, 308 must be in a straight line and within a maximum penetration angle 310; -The depth to which the biopsy needle is inserted into the tumor is less than the maximum puncture depth of 312.

[0024] In the example described here, the maximum penetration angle 310 is 20 degrees and the maximum penetration depth 312 is 2.2 cm.

[0025] The constraints also include the following: - for example, in the case of a puncture of at least 2.5 mm from the edge of the tumor, the proximity of the zone punctured by the biopsy to this edge, - the presence of necrotic tissue within the tumor, which should not be passed through the needle during biopsy; - exclusion of adipose tissue from the zone where puncture can be performed, - Fixing a specific entry point or set of entry points to the margin of the tumor; - Avoiding the most vascularized zones to reduce the risk of cancer cell spread, the presence of bones that prevent puncture at certain entry points and angles, for example ribs in the case of lung tumors, and - The characteristics of the organ and / or biopsy procedure envisaged. For example, for a transrectal ultrasound biopsy of the prostate, the patient is in the left lateral position, limiting accessibility.

[0026] As mentioned above, each pixel corresponds to an area of ​​the cross-sectional image that is substantially square and 2.1 mm on a side, and the needle penetration zone has a diameter of substantially 0.84 mm and a length of 4.55 mm ± 1.45 mm. Therefore, there are two problems in analyzing the raster image 14. 1) On the one hand, the pixels are too large to show tumor heterogeneity. 2) On the other hand, the pixel is too small for the biopsy needle to sample tissue within the puncture zone with certain diffusion parameters and homogeneity characteristics that are substantially contained in the area corresponding to a single pixel.

[0027] The upsampler 20 refines the raster image to overcome the first problem.

[0028] To that end, upsampler 20 receives raster image 14 and upsamples it to processed image 22. Processed image 22 has a much higher spatial resolution than raster image 14, as can be seen by comparing Figures 2 and 4. Each pixel in processed image 22 corresponds to a substantially square cross-sectional area of ​​the image. The resolution of processed image 22 is selected by upsampler 20 so that each of its pixels has a side length that is shorter than the puncture diameter of needle data 160.

[0029] In the embodiment described herein, the upsampler 20 applies an interpolation technique to upsample the raster image 14. The interpolation technique used may be bicubic interpolation, bilinear interpolation, a Bell filter, a Hermite filter, a Mitchell filter, a Lanczos filter, or any other conventional interpolation technique adapted to the image. In the example described herein, the upsampler 20 applies the bicubic interpolation technique described, for example, in Keys, R. (1981). "Cubic convolution interpolation for digital image processing." IEEE Transactions on Signal Processing. In Acoustics, Speech, and Signal Processing (Vol. 29, p. 1153).

[0030] Alternatively, the upsampler 20 can upsample the raster image 14 by applying a discrete cosine transform (DCT) technique, such as those described in Dugad, R., & Ahuja, N. (2001). "A fast scheme for image size change in the compressed domain," IEEE Transactions on Circuits and Systems for Video Technology, 11(4), 461-474 and Park, H., Park, Y., & Oh, SK (2003). "L / M-fold image resizing in block-DCT domain using symmetric convolution," IEEE Transactions on Image Processing, 12(9), 1016-1034. Alternatively, machine learning-based techniques, particularly deep learning, can be used, such as those described in Wang, Z., Chen, J., & Hoi, SC (2019). "Deep learning for image super-resolution: A survey," arXiv preprint arXiv:1902.06068.

[0031] The processed image 22 has pixels corresponding to substantially square zones whose sides are at least five times smaller than the sides of the zones of pixels of the raster image 14, and preferably 20 times smaller as in the example described here.

[0032] In another embodiment, imaging device 1 functions in three dimensions, and imaging data 12 includes nine consecutive cross-sectional raster images 14, each pair separated by a distance of 6 mm. Raster images 14 thus define voxels measuring 2.1 mm by 2.1 mm by 6 mm. Processed image 22 defines voxels that are at least five times smaller in each dimension, and preferably 20 times smaller. In the example described here, the voxels of processed image 22 each define a zone measuring 0.105 mm by 0.105 mm by 0.3 mm.

[0033] The processed image 22 has a much higher resolution than the raster image 14. The areas corresponding to the pixels in the processed image are much smaller than the heterogeneity within the tumor. Now, the pixels of the processed image 22 must be grouped to slice the cross-sectional image into regions that are homogeneous enough to overcome the first problem and large enough to overcome the second problem.

[0034] To that end, slicer 30 receives processed image 22 and divides it into a set of regions 32. Each region 34 in set of regions 32 contains at least two immediately adjacent pixels of processed image 22.

[0035] Region 34 is associated with a diffusion parameter value that is substantially equal to the average of the diffusion parameter values ​​of its constituent pixels. This diffusion parameter value for region 34 may be selected, for example, as the average or median of the diffusion parameter values ​​associated with the pixels of region 34.

[0036] For each region 34, slicer 30 identifies a set of regions 32 such that the variance of the diffusion parameter values ​​associated with the pixels in that region 34 is below a threshold value. Thus, a region 34 in a set of regions 32 is a set of contiguous pixels in processed image 22 whose associated diffusion parameter values ​​are substantially homogeneous, in other words, have low variance.

[0037] Slicer 30 further identifies set of regions 32 such that regions 34 correspond to portions of the cross-sectional diffusion-weighted magnetic resonance image whose dimensions are greater than or substantially equal to the dimensions of the puncture made by the biopsy needle of needle data 160, where the portions to which regions 34 correspond have substantially similar dimensions, e.g., with substantially equal respective areas. The diffusion parameters associated with each region 34 in set of regions 32 form set of diffusion parameters 36.

[0038] In one embodiment, the slicer 30 divides the processed image 22 into a set of regions 32 using a superpixel method with size constraints and low variance of diffusion parameter values ​​associated with the pixels of each region 34. In the example described herein, the superpixel method used is that described in Ren, X., & Malik, J. (2003, October). "Learning a classification model for segmentation." In Proceedings Ninth IEEE International Conference on Computer Vision. (p. 10). IEEE. In the example described herein, shown in FIG. 5, the superpixels correspond to portions of the cross-sectional diffusion-weighted magnetic resonance image having a size slightly larger than the corresponding portions of the pixels in the raster image 14. Alternatively, one of the superpixel methods described in Stutz, D., Hermans, A., & Leibe, B. (2018). "Super-pixels: An evaluation of the state-of-the-art." Computer Vision and Image Understanding, 166, 1-27, can be used.

[0039] 5, a portion of the cross-sectional diffusion-weighted magnetic resonance image is not considered because it is region 500 that corresponds to necrotic tissue. Identification of data corresponding to region 500 can be performed by upsampler 20, slicer 30, or another part of imaging device 1. Alternatively, data corresponding to region 500 can be identified by an operator, for example, prior to performing a biopsy procedure.

[0040] Alternatively, the slicer 30 can divide the processed image 22 into a set of regions 32 by the convex polygon partitioning method described in Duan, L., & Lafarge, F. (2015). "Image partitioning into convex polygons." In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 3119-3127). This method is based on the construction of a Voronoi diagram, whose cells are convex polygons, by matching lines already drawn in the processed image 22. Here, it is possible to obtain regions 34 whose dimensions can be controlled and which preserve the shape or edges of objects in the processed image 22 at a sub-pixel scale. Here, the set of regions 32 provides a representation of inhomogeneities in the processed image 22.

[0041] In the three-dimensional case where some processed images 22 are produced by the upsampler 20, the slicer 30 segments the processed images 22 using a super-voxel method such as that described in Xu, C., & Corso, JJ (2012, June). "Evaluation of super-voxel methods for early video processing." In 2012 IEEE conference on computer vision and pattern recognition (pp. 1202-1209). IEEE.

[0042] Therefore, the region 34 identified by the slicer 30 is - large enough so that the needle puncture zone can be substantially contained within one of the regions 34; - the sampling within the region 34 is sufficiently homogeneous to be representative of that region 34;

[0043] The selector 40 identifies which of the regions 34 are associated with the sample so that the diffusion parameter values ​​within these regions are representative of all diffusion parameters.

[0044] To this end, the selector 40 identifies within the set 36 of diffusion parameters a subset 42 of diffusion parameters that includes the same number of diffusion parameters as the number of punctures performed in the procedure data 162. Because the number of punctures performed is at least two, the subset 42 of diffusion parameters includes at least two diffusion parameters. A first diffusion parameter 420 of the diffusion parameters is selected from the first decile of the diffusion parameter values ​​in the set 36 of diffusion parameters. A second diffusion parameter 426 of the diffusion parameters is selected from the last decile of the diffusion parameter values ​​in the set 36 of diffusion parameters. The first diffusion parameter 420 and the second diffusion parameter 426 may each be, for example, the median of their respective deciles. This ensures that the subset 42 of diffusion parameters identified by the selector 40 represents the diffusion parameter values ​​of the regions 34 of the set 32 ​​of regions, particularly the extreme values ​​of the diffusion parameters in the set 36 of diffusion parameters.

[0045] In one embodiment, the first diffusion parameter 420 and the second diffusion parameter 426 are selected from the 5th and 95th percentiles, preferably the 2nd and 98th percentiles, respectively, of the set of diffusion parameters, where the first diffusion parameter 420 and the second diffusion parameter 426 are each the median of their respective quartiles.

[0046] When the number of punctures performed in the procedure data 162 is greater than two, the selector 40 identifies additional diffusion parameters 424 in addition to the first diffusion parameter 420 and the second diffusion parameter 426. The values ​​of each of these additional diffusion parameters 424 are between the values ​​of the first diffusion parameter 420 and the second diffusion parameter 426.

[0047] In one embodiment, the selector 40 selects a diffusion parameter 424 that falls between the value of the first diffusion parameter 420 and the value of the second diffusion parameter 426, such that the diffusion parameters of the diffusion parameter subset 42 have values ​​that are equidistant from each other by two. To do so, the selector 40 applies equation (1) in Appendix A, where D is the value of the first diffusion parameter 420, D is the value of the second diffusion parameter 426, N is the number of punctures performed in the biopsy procedure of the procedure data 162, j is an integer between 1 and the number of punctures performed (inclusive), and D is the value of the j-th diffusion parameter of the diffusion parameter subset 42. The diffusion parameter subset 42 thus obtained is shown in FIG. 6, where the x-axis indicates the diffusion parameter value and the y-axis indicates the probability of having the diffusion parameter value. The subset 42 of diffusion parameters includes a small number of diffusion parameters (four in the example described here), which are largely representative of all diffusion parameters in the set 36 of diffusion parameters, including the extreme values.

[0048] Alternatively, if the set of diffusion parameters 36 has a highly skewed distribution of its diffusion parameters around the median value M of the diffusion parameters of the set of diffusion parameters 36, the selector 40 identifies the diffusion parameters 424 according to equation (2) of Appendix A, where D is identified according to equation (3) of Appendix A and Delta is identified according to equation (4) of Appendix A. This alternative allows the longest tail of the diffusion parameter distribution of the set of diffusion parameters 36 to be given less weight, so that the subset of diffusion parameters 42 is more representative of the diffusion parameter distribution of the set of diffusion parameters 36. This skewness can occur, for example, when there is excessive noise in the raster image 12, in which case small diffusion parameter values ​​are disturbed more than high diffusion parameter values. The identification of this skewness, and thus the selection of this skewness method or the equidistance method described above, can be performed manually or automatically by the selector 40.

[0049] In one embodiment, the selector 40 calculates a distortion coefficient for the set of diffusion parameters 36. If the distortion coefficient exceeds a certain threshold, the selector 40 identifies the diffusion parameters 424 according to equation (2). If not, the selector 40 applies equation (1).

[0050] The selector 40 returns a subset of regions 44 that includes regions 46 derived from the set of regions 32. Each region 46 of the subset of regions 44 is associated with a diffusion parameter substantially equal to one of the diffusion parameters of the subset of diffusion parameters 42. Furthermore, in the example described herein, all of the diffusion parameters are associated with at least one of the regions 46. Alternatively, for each diffusion parameter of the subset of diffusion parameters 42, the subset of regions 44 may include all of the regions 34 of the set of regions 32 that are associated with that diffusion parameter of the subset of diffusion parameters 42. In embodiments in which the set of diffusion parameters is sliced ​​into quartiles, the diffusion parameter of each region 46 of the subset of regions 44 falls in the same quartile as the diffusion parameter of the associated subset of diffusion parameters 42.

[0051] The subset of regions 44 includes regions 46 with diffusion parameter values ​​that are well distributed within the set of diffusion parameters 36. In effect, sampling from these multiple regions 46, each differing by two diffusion parameter values, represents all regions 34.

[0052] However, tumor 200 has several characteristics, such as the shape of constraint set 300 or the presence of necrotic regions 500. Based on these characteristics, guide 50 identifies which regions of region subset 44 should be sampled and how the sampling should be performed.

[0053] To that end, the guide 50 identifies, from the subset of regions 44, a number of puncture parameter set data 52 equal to the number of punctures to be performed in the procedure data 162. Each set of puncture parameters in the puncture parameter set data 52 includes a puncture depth 520, a puncture direction 522, and a puncture entry point 524, as shown in the example shown here in FIG. 7. Each set of puncture parameters defines, together with the needle data 160, a puncture zone 526 (seen in FIG. 11) that is substantially contained in one of the regions 46 in the subset of regions 44. The guide 50 identifies the sets of puncture parameters in the puncture parameter set data 52 such that the sets of puncture parameters together satisfy the set of constraints 300 of the procedure data 162.

[0054] In the example shown in FIG. 8 , guide 50 receives from selector 40 a subset of regions 44 including four regions 46. Here, subset 42 of diffusion parameters includes three diffusion parameters P1, P2, and P3. Diffusion parameters P1 and P2 are each associated with a single region of subset 44 of regions, referenced as “1” and “2,” respectively, in FIG. 8 . Parameter P3 is associated with two regions of subset 44 of regions, referenced as “3” in FIG. 8 . Guide 50 then selects from these regions 46 a set 54 of regions, shown in FIG. 9 , adapted for puncture according to set of constraints 300. Because the procedure involves three biopsies, set 54 of regions includes three regions 540, 542, and 544. Regions 540 and 542 correspond to parameters P1 and P2, respectively. Region 544, labeled “3,” is selected from one of the two regions corresponding to diffusion parameter P3 in FIG. 8 .

[0055] In some cases, the guide 50 may determine that the subset of regions 44 is inconsistent with the set of constraints 300, meaning that it is not possible to perform the same number of punctures as originally planned that would satisfy the set of constraints 300. In this case, the selector 40 identifies a new subset of diffusion parameters 42 that includes one less diffusion parameter than before, and then proceeds to identify a subset of regions 44. This is repeated until the guide 50 identifies a subset of regions 44 that is consistent with the set of constraints 300, or until it is determined that the biopsy procedure is not possible.

[0056] 10 and 11, the new set of regions 54 selected by the guide 50 includes three regions 540, 542, and 544. Here, the puncture zone 526 of the puncture parameter set data 52 specified by the guide 50 is substantially contained within the region 544 of the new set of regions 54.

[0057] The puncture parameter set data 52 specified by the guide 50 is provided to a surgeon or radiologist, who can then perform biopsies that are much more relevant than if they were selected manually or randomly. Furthermore, because only a small number of biopsies are performed, less than four in the example described here, medical risks are significantly reduced.

[0058] Here, reference is made to FIGS.

[0059] In the embodiment described herein, the imaging device 1 extracts the results of a biopsy procedure performed in accordance with the puncture parameter set data 52. The results are stored in memory as puncture data 58. From the puncture data 58, the imaging device 1 reconstructs a density map of the tumor 200.

[0060] To that end, the imaging device 1 comprises a correlator 60 and a constructor 70. The correlator 60 derives from the puncture data 58 a correlation between the diffusion parameter values ​​of the set of diffusion parameters 36 and the cell density values ​​within the tumor 200. The cell density may be the total cell density or the cancer cell density. The constructor 70 uses this correlation to reconstruct a cell density map within the tumor 200.

[0061] The puncture data 58 includes a set of cell densities, each of which is associated with one of the puncture zones 526 of the puncture parameter set data 52. Each density has a value indicating the cell density of the associated puncture zone 526.

[0062] The puncture data 58 can be obtained after a biopsy procedure by a surgeon or radiologist and after analysis of a sample taken from the puncture zone 526 of the puncture parameter set data 52. The analysis of the puncture data 58 can be performed by a histological method such as that described in Yin, Y., Sedlaczek, O., Mueller, B., Warth, A., Gonzalez-Vallinas, M., Lahrmann, B., Grabe, N., Kauczor, HU., Breuhahn, K., Vignon-Clementel, IE., Drasdo, D. (2018). "Tumor cell load and heterogeneity estimation from diffusion-weighted MRI calibrated with histological data: an example from lung cancer." IEEE Transactions on Medical Imaging, 37(1), 35-46.

[0063] The correlator 60 receives the puncture data 58 and associates each cell density value associated with one of the puncture zones 526 with a diffusion parameter value for the region 46 that substantially contains the puncture zone 526. The correlator 60 determines correlation data 62 that associates cell density values ​​with diffusion parameter values ​​of the set of diffusion parameters 36.

[0064] In the described embodiment, the correlator 60 identifies pairs each relating a diffusion parameter value to a cell density value associated with one of the puncture zones in the puncture parameter set data 52. Based on these pairs, the correlator 60 performs linear interpolation, thereby relating the diffusion parameter values ​​in the set of diffusion parameters 36 to cell density values. The correlator 60 identifies correlation data 62 based on the results of the linear interpolation. FIGS. 13, 14, and 15 show linear interpolation for the number of punctures in the procedure data 162: two, three, and four, respectively. FIGS. 13, 14, and 15 are fitted to an x-axis representing the diffusion parameter value and a y-axis representing the cell density. A higher diffusion parameter value indicates a higher mobility of water molecules in the tissue, resulting in a lower cell density.

[0065] Alternatively, the interpolation performed by correlator 60 can be a strictly monotonic polynomial interpolation, or other strictly monotonic function.

[0066] The constructor 70 determines density data 72 from the processed image 22 and the correlation data 62. The density data 72 comprises a density image. Each pixel of the density image corresponds to an area of ​​the cross-sectional image to which a pixel of the processed image 22 also corresponds. The constructor 70 associates each pixel of the density image with a cell density value that is related in the correlation data 62 to the diffusion parameter value of that pixel in the processed image 22. Thus, from the raster image 14, the imaging device 1 can now determine a reduced number of samplings to be performed and then reconstruct an accurate map of the cell density of the tumor 200 based on the results of these samplings.

[0067] See Figure 16. Figure 16 is a fit with the axis x representing the number of punctures performed and the axis y representing the tumor burden value.

[0068] In one embodiment (not shown), the imaging device 1 further includes an estimator configured to determine the tumor volume of the tumor 200. Here, the estimator performs a numerical area integration over the entire area of ​​the density image to derive the tumor volume. FIG. 16 shows tumor volumes 602, 604, and 606 determined by the estimator depending on whether the number of punctures is two, three, or four, respectively, and compares the tumor volume with a reference tumor volume 608. The more punctures are performed, the closer the tumor volume determined by the estimator is to the reference tumor volume 608. The imaging device 1 can therefore measure the tumor volume with only two biopsies, whereas conventional methods involving randomly taken biopsies with eight or even ten biopsies are less accurate. With three or four biopsies, the measured tumor volume is even more accurate, even though the number of punctures is kept low.

[0069] Here, the imaging device 1 can therefore identify the tumor volume of the tumor 200 from the results of sampling according to the puncture parameter set data 52, and visualize and / or quantify the heterogeneity of the tumor.

[0070] See Figures 17 and 18. Each figure shows three histograms based on tumor burden determinations from two, three, and four biopsies. Specifically, the x-axis shows tumor burden values, and the y-axis shows the percentage of biopsies with estimated tumor burden values ​​compared to all biopsies performed. Thus, in Figure 17, 10% of biopsy procedures with two biopsies had tumor burden values ​​of 1 mm. 3 This results in an estimated tumor burden of 600,000 cells per 1000. In these figures, a reference tumor burden value of 170 is also shown, indicating the ideal target value to be estimated.

[0071] To illustrate the benefits of the present invention compared to existing methods, a histogram derived from biopsies selected by a device according to the present invention is shown in FIG. 17, while FIG. 18 shows a histogram created from randomly performed punctures that satisfy the set of constraints 300.

[0072] These figures show that the puncture parameter set data 52 determined by the imaging device 1 allows punctures to be performed to determine tumor volume that is much closer to the actual tumor volume than random punctures. These figures also show that tumor volume measurements made based on the puncture parameter set data 52 have much less scatter than measurements made by random punctures.

[0073] Therefore, by using the puncture parameter set data 52 identified by the imaging device 1, the correlator 60, constructor 70 and estimator are able to identify a much more reliable tumor volume, i.e., a tumor volume that is closer to the reference value 170.

[0074] The imaging device 1 enables a surgeon or radiologist to accurately and robustly measure tumor volume with significantly fewer punctures, typically 3-5 fewer punctures than conventional procedures.

[0075] The correlator 60, constructor 70 and estimator are programs executed by a computer processor. The alternatives described for the upsampler 20, slicer 30, selector 40 and guide 50 can also be applied to the correlator 60, constructor 70 and estimator.

[0076] In the above, slicer 30 identifies region 34 having a size greater than or substantially equal to the puncture zone of the biopsy needle. Alternatively, slicer 30 can identify region 34 having a dimension less than the puncture length but greater than the puncture diameter. In a biopsy performed based on this, histological measurements are performed only on the portion of the sampling corresponding to regions 540, 542, 544.

[0077] Alternatively, the device may also receive data in memory 10 from other types of non-invasive imaging in which the pixels of raster image 14 are associated with parameters reflecting cell density instead of diffusion parameters as in the case of DWI. In order to maintain the disclosure of the present application as originally filed, the contents of claims 1 to 14 as originally filed are added below. (Claim 1) A memory (10), imaging data (12) including at least one raster image (14) derived from cross-sectional diffusion-weighted magnetic resonance imaging of a tumor, each pixel of the raster image (14) being associated with a diffusion parameter whose value represents the mobility of water molecules within the tumor; biopsy data (16) including needle data (160) defining the dimensions of the puncture to be performed by the biopsy needle; procedure data (162) including two or more numbers of punctures to be performed in a biopsy procedure and a set of constraints (300) relating to intervention constraints that must be satisfied to perform said number of punctures; a memory (10) configured to receive the an upsampler (20) configured to upsample the raster image (14) into a processed image (22), the spatial resolution of the processed image (22) being such that each pixel of the processed image (22) corresponds to a substantially square cross-sectional image area having sides that are shorter than a needle diameter dimension of the needle data (160); a slicer (30) configured to divide the processed image (22) from the upsampler (20) into a set of regions (32), each region (34) in the set of regions comprising: at least two pixels of the processed image (22) immediately adjacent to each other; associated with a diffusion parameter whose value is substantially equal to the average of the values ​​of the diffusion parameter associated with its constituent pixels; the variance of the values ​​of the diffusion parameter associated with the pixels of the region is less than a given variance value; a portion of the cross-sectional diffusion-weighted magnetic resonance image of the tumor whose dimensions are greater than or substantially equal to the dimensions of a puncture made by a biopsy needle of the biopsy data (16); a slicer (30) in which the diffusion parameters associated with each region (34) of the set of regions (32) form a set (36) of diffusion parameters; A selector (40), identifying a subset (42) of diffusion parameters within the set (36) of diffusion parameters, the subset (42) including the same number of diffusion parameters as the number of punctures performed in the procedure data (162), wherein the subset (42) of diffusion parameters includes at least a first diffusion parameter (420) selected from a first decile of the diffusion parameter values ​​of the set (36) of diffusion parameters, a second diffusion parameter (426) selected from a last decile of the diffusion parameter values ​​of the set (36) of diffusion parameters, and a diffusion parameter (424) whose respective diffusion parameter values ​​are between the value of the first diffusion parameter (420) and the value of the second diffusion parameter (426) when the number of punctures performed in the procedure data (162) is greater than two; returning a subset (44) of regions derived from said set (32) of regions, each region (46) being associated with a diffusion parameter substantially equal to one of said parameters of said subset (42) of diffusion parameters; a selector (40) configured to: a guide (50) configured to identify, from the subset (44) of regions, a number of puncture parameter sets (52) equal to the number of punctures to be performed in the procedure data (162), each set of puncture parameters including a puncture depth (520), a puncture direction (522), and a puncture entry point (524), the guide (50) defining, together with the needle data (160), a puncture zone (526) that is substantially included in one of the regions (544) of the subset (54) of regions such that together they satisfy the set of constraints (300) of the procedure data (162); An imaging device comprising: (Claim 2) 2. The imaging device of claim 1, wherein the upsampler (20) is configured to upsample the raster image (14) to the processed image (22) by performing interpolation. (Claim 3) 3. The imaging device of claim 2, wherein the upsampler (20) is configured to perform bicubic interpolation to upsample the raster image (14) to the processed image (22). (Claim 4) 4. The imaging device according to claim 1, wherein the slicer (30) is configured to divide the processed image (22) from the upsampler (20) by a superpixel method. (Claim 5) 5. The imaging device according to claim 1, wherein the number of punctures performed in the biopsy procedure of the procedure data stored in the memory is four or less. (Claim 6) 6. The imaging device of claim 1, wherein the selector selects the first diffusion parameter from a 2nd percentile of the diffusion parameter values ​​of the set of diffusion parameters and selects the second diffusion parameter from a 98th percentile of the diffusion parameter values ​​of the set of diffusion parameters. (Claim 7) 7. The imaging device according to claim 1, wherein the first diffusion parameter (420) and the second diffusion parameter (426) are each equal to the median of their respective quartiles. (Claim 8) 8. The imaging device according to claim 1, wherein when the number of punctures performed in the procedure data is greater than two, the selector selects the diffusion parameters of the subset of diffusion parameters so that the diffusion parameters are equidistant from each other, two at a time. (Claim 9) When the number (N) of punctures performed in the procedure data (162) is greater than two, the selector (40) identifies the median (M) of the diffusion parameters in the set (36) of diffusion parameters and identifies the diffusion parameter (Dj), where j is a natural number in the subset (42) of diffusion parameters that is between 1 and the number (N) of punctures performed in the procedure data (162) according to the following formula: Dj=Dini+Delta(j-1) Here, Dini satisfies the following equation: Dini = M-min(|M-Dmin|,|M-Dmax|) Delta satisfies the following formula: Delta=2(M-Dini) / (N-1) 8. The imaging device of claim 1, wherein Dmin is equal to the value of the first diffusion parameter (420) identified by the selector (40), and Dmax is equal to the value of the second diffusion parameter (426) of the subset (42) of diffusion parameters identified by the selector (40). (Claim 10) The puncture depth (520), the puncture direction (522), and the puncture entry point (524) of each of the set of puncture parameters (52) specified by the guide (50), together with the biopsy data (16), define the puncture trajectory of the biopsy needle, and the set of constraints (300) of the procedure data (162) stored in the memory (10) includes: the punctures of the biopsy procedure must have the same entry point (524) into the tumor (200); The puncture trajectory must be a straight line and must be within the maximum puncture angle; The puncture depth (520) is less than the maximum puncture depth; the puncture zone (526) being spaced from the edge of the tumor (200) by a predetermined distance; If the tumor (200) includes a necrotic zone (500), the puncture trajectory does not penetrate the necrotic zone (500); if the tumor (200) includes a fatty tissue zone, none of the puncture zones (526) intersect with the fatty tissue zone; At least one of the entry points (524) of the puncture parameter set data (52) is pre-fixed; 10. The imaging device according to claim 1, further comprising one or more constraints selected from the group comprising: (Claim 11) the memory (10) is further configured to receive puncturing data (58) including a set of cell densities, each cell density being associated with one of the puncturing zones (526) of the puncturing parameter set data (52), each cell density having a value representing a cell density of the puncturing zone (526); The imaging device (1) a correlator (60) configured to associate each cell density value associated with one of the puncture zones (526) with the diffusion parameter value of the region (544) in which the puncture zone (526) is substantially contained for the set of regions (54), and to derive correlation data (62) associating cell density values ​​with the diffusion parameter values ​​of the set of diffusion parameters (36); a constructor (70) configured to identify density data (72) from the processed image (22) and the correlation data (62), the density data (72) comprising a density image in which each pixel corresponding to a region of the cross-sectional image is associated with the diffusion parameter value of the pixel in the processed image (22) corresponding to the region of the cross-sectional image of the pixel and the cell density with which that value is associated in the correlation data (62); 11. The imaging device according to claim 1, further comprising: (Claim 12) 12. The imaging device of claim 11, wherein the correlator (60) performs a strictly monotonic linear or nonlinear interpolation based on pairs formed from diffusion parameter values ​​and cell density values ​​associated with the puncture zone (526), ​​and derives the correlation data (62) from the results of the interpolation. (Claim 13) 13. The imaging device according to claim 11, wherein the cell density of the puncture data and the density data (72) corresponds to a total cell density or a cancer cell density. (Claim 14) 14. The imaging device of claim 11, further comprising an estimator configured to determine a tumor mass of the tumor based on the density data.

[0078] Annex A (1) Dj = Dmin+(j - 1)(Dmax - Dmin) / (N - 1) (2) Dj = Dini + Delta(j - 1) (3) Dini = M - min(|M - Dmin|,|M - Dmax|) (4) Delta = 2(M - Dini) / (N - 1)

Claims

1. A memory (10) comprising: imaging data (12) including at least one raster image (14) derived from cross-sectional diffusion-weighted magnetic resonance imaging of a tumor, each pixel of the raster image (14) being associated with a diffusion parameter value representing the mobility of water molecules within the tumor; biopsy data (16) including needle data (160) defining the dimensions of the puncture made by the biopsy needle; procedure data (162) including two or more numbers of punctures to be performed in a biopsy procedure and a set of constraints (300) relating to intervention constraints that must be satisfied to perform said number of punctures; a memory (10) configured to receive an upsampler (20) that upsamples the raster image (14) into a processed image (22), the processed image (22) having a spatial resolution such that each pixel of the processed image (22) corresponds to a substantially square cross-sectional image area, the sides of the area having a length that is shorter than a needle diameter dimension of the needle data (160); A slicer (30) configured to divide the processed image (22) from the upsampler (20) into a set of regions (32), each region (34) in the set of regions comprising: comprising at least two pixels of the processed image (22) immediately adjacent to each other; the value of each region (34) in the set of regions (32) is associated with a diffusion parameter value substantially equal to the average of several diffusion parameter values ​​associated with the pixels comprising said region; the variance of some of the diffusion parameter values ​​associated with the pixels of the region is less than a given variance value; a dimension of each region (34) in the set of regions (32) corresponds to a portion of the cross-sectional diffusion-weighted magnetic resonance image of the tumor that is greater than or substantially equal to a dimension of a puncture made by a biopsy needle of the biopsy data (16); a slicer (30), wherein the diffusion parameter values ​​associated with each region (34) of the set of regions (32) form a set of diffusion parameters (36); A selector (40) comprising: identifying a subset (42) of diffusion parameters within the set (36) of diffusion parameters, the subset (42) of diffusion parameters including the same number of diffusion parameters as the number of punctures performed in the procedure data (162), the subset (42) of diffusion parameters including at least a first diffusion parameter value (420) selected from a first decile of the diffusion parameter values ​​of the set (36) of diffusion parameters, a second diffusion parameter value (426) selected from a last decile of the diffusion parameter values ​​of the set (36) of diffusion parameters, and a diffusion parameter value (424) between the first diffusion parameter value (420) and the second diffusion parameter value (426) when the number of punctures performed in the procedure data (162) is greater than two; returning a subset (44) of regions derived from the set (32) of regions, each region (46) of the subset (44) of regions being associated with a diffusion parameter value substantially equal to one of the diffusion parameter values ​​of the subset (42) of diffusion parameters; a selector (40) configured to: a guide (50) configured to identify, from the subset (44) of regions, a number of puncture parameter set data (52) equal to the number of punctures to be performed in the procedure data (162), each set of puncture parameters including a puncture depth (520), a puncture direction (522), and a puncture entry point (524), the guide (50) defining, together with the needle data (160), a puncture zone (526) that is substantially included in one of the regions (544) of the subset (54) of regions such that together they satisfy the set of constraints (300) of the procedure data (162); An imaging device comprising:

2. 2. The imaging device of claim 1, wherein the upsampler (20) is configured to upsample the raster image (14) to the processed image (22) by performing interpolation.

3. 3. The imaging device of claim 2, wherein the upsampler (20) is configured to perform bicubic interpolation to upsample the raster image (14) to the processed image (22).

4. The imaging device of any one of claims 1 to 3, wherein the slicer (30) is configured to split the processed image (22) from the upsampler (20) by a superpixel method.

5. 5. The imaging device according to claim 1, wherein the number of punctures performed in the biopsy procedure of the procedure data (162) stored in the memory (10) is four or less.

6. 6. The imaging device of claim 1, wherein the selector (40) selects the first diffusion parameter value (420) from a 2nd percentile of the diffusion parameter values ​​of the set of diffusion parameters (36) and selects the second diffusion parameter value (426) from a 98th percentile of the diffusion parameter values ​​of the set of diffusion parameters (36).

7. The imaging device of any one of claims 1 to 6, wherein the first diffusion parameter value (420) and the second diffusion parameter value (426) are each equal to the median of a quartile to which each diffusion parameter value belongs.

8. An imaging device as described in any one of claims 1 to 7, wherein when the number of punctures performed in the procedure data (162) is more than two, the selector (40) selects the diffusion parameters (424) of the subset (42) of diffusion parameters so that the diffusion parameter values ​​(424) are equidistant from each other in pairs.

9. When the number (N) of punctures performed in the procedure data (162) is greater than two, the selector (40) identifies a median (M) of the diffusion parameter values ​​of the set (36) of diffusion parameter values ​​and identifies the diffusion parameter value (Dj), where j is a natural number of the subset (42) of diffusion parameters that is between 1 and the number (N) of punctures performed in the procedure data (162) according to the following formula: Dj=Dini+Delta(j-1) Here, D satisfies the following equation: Dini=M-min (|M-Dmin|, |M-Dmax|) Delta satisfies the following formula: Delta=2(M-Dini) / (N-1) 8. The imaging device of claim 1, wherein Dmin is equal to the first diffusion parameter value (420) identified by the selector (40), and Dmax is equal to the second diffusion parameter value (426) of the subset (42) of diffusion parameters identified by the selector (40).

10. The puncture depth (520), the puncture direction (522), and the puncture entry point (524) of each of the set of puncture parameters (52) specified by the guide (50), together with the biopsy data (16), define the puncture trajectory of the biopsy needle, and the set of constraints (300) of the procedure data (162) stored in the memory (10) includes: the punctures of the biopsy procedure must have the same entry point (524) into the tumor (200); The puncture trajectory must be a straight line and must be within the maximum puncture angle; The puncture depth (520) is less than the maximum puncture depth; the puncture zone (526) being spaced from the edge of the tumor (200) by a predetermined distance; If the tumor (200) includes a necrotic zone (500), the puncture trajectory does not penetrate the necrotic zone (500); if the tumor (200) includes a fatty tissue zone, none of the puncture zones (526) intersect with the fatty tissue zone; At least one of the entry points (524) of the puncture parameter set data (52) is pre-fixed; 10. The imaging device of claim 1, further comprising one or more constraints selected from the group comprising:

11. the memory (10) is further configured to receive puncture data (58) including a set of cell densities, each cell density being associated with one of the puncture zones (526) of the puncture parameter set data (52), each cell density having a value representing a cell density of the puncture zone (526); The imaging device (1) a correlator (60) configured to associate each cell density value associated with one of the puncture zones (526) with the diffusion parameter value of the region (544) in which the puncture zone (526) is substantially contained for the set of regions (54), and to derive correlation data (62) associating cell density values ​​with the diffusion parameter values ​​of the set of diffusion parameters (36); a constructor (70) configured to identify density data (72) from the processed image (22) and the correlation data (62), the density data (72) including a density image for each pixel corresponding to a region of the cross-sectional image, the density data (72) being associated with a cell density value, the cell density value being associated in the correlation data (62) with the diffusion parameter value of a pixel in the processed image (22) corresponding to the region of the cross-sectional image of the pixel; The imaging device according to any one of claims 1 to 10, further comprising:

12. The imaging device of claim 11, wherein the correlator (60) performs a strictly monotonic linear or nonlinear interpolation based on pairs formed from diffusion parameter values ​​and cell density values ​​associated with the puncture zone (526), ​​and derives the correlation data (62) from the results of the performed interpolation.

13. 13. The imaging device of claim 11 or 12, wherein the cell density of the puncture data and the density data (72) corresponds to a total cell density or a cancer cell density.

14. The imaging device of any one of claims 11 to 13, further comprising an estimator configured to determine a tumor volume of the tumor (200) based on the density data (72).

Images