Method for determining breast density

Abstract

The invention relates to a method (100) for determining breast density (Dm) from a first digital representation (BGR, BGR') of a breast, comprising:- a step (110) of obtaining a second digital representation (FGT) of the mammary gland;- a step (120) of determining the gland volume (Vg) from said second representation (FGT);- a step (130) of determining the breast volume (Vm) from said first representation (BGR, BGR');- a step (140) of producing first and second morphological characteristics (Qi1, Qi2) of interest of the mammary gland, the first and second morphological characteristics (Qi1, Qi2) of interest being distinct;- a step (150) of producing the breast density (Dm) from the first and second morphological characteristics (Qi1, Qi2). Figure for abstract: Fig 5

Description

Title of the invention: Method for determining breast density

[0001] The present invention relates to a method for determining, and more precisely for quantifying, breast density from a medical image, resulting from an acquisition sequence by a medical imaging device, of a patient's breast.

[0002] In medical imaging, an acquisition sequence makes it possible to collect and store raw data on the anatomy and / or on the functioning of an organ, making it possible to reconstruct a three-dimensional visual representation of said organ, in other words an image of said organ.

[0003] The invention is based in particular, but in a non-limiting manner, on imaging techniques, for example, by Mammography ("Mammography" or "Mastography", according to English terminology), by Magnetic Resonance ("MRI", according to English terminology) or by Computed Tomography ("X-ray Computed Tomography" or "CT", according to English terminology), or even by any other medical imaging technology concerned with the acquisition of images of a patient's breast such as for example by ultrasound.

[0004] An imaging device 1 adapted to a given technique or technology, for example using the Nuclear Magnetic Resonance technique, as illustrated by Figures 1 and 2, is generally used. Said imaging device 1 can deliver a plurality of experimental data 10 related to one or more parts of a patient's body, in particular the breast in this application case. Such a medical imaging device 1 is controlled using an acquisition console 2. A user 6, for example an operator, practitioner or researcher, can thus choose commands 11 to control the device 1, from parameters or instructions 16 entered via a human-machine input interface 8 of the analysis system.Such a human-machine interface 8 may consist, for example, of a computer keyboard, a pointing device, a touch screen, a microphone or, more generally, any interface arranged to translate a gesture or an instruction issued by a human 6 into control or configuration data. From information 10 produced by said apparatus 1, a plurality of sequences of digital images 12 of a part of a human or animal body are obtained. The apparatus 1 applies for this purpose a combination of high-frequency electromagnetic waves to the part of the body considered and measures the signal re-emitted by certain atoms. The apparatus 1 thus makes it possible to determine the chemical composition and therefore the nature of the biological tissues in each . elementary volume of interest, commonly called a voxel, of the imaged volume.

[0005] The image sequences 12 may optionally be stored within a server 3, i.e. a computer equipped with its own storage means, and constitute a medical file 13 of a patient. Such a file 13 may comprise images of different types, such as structural images highlighting the activity of the tissues or anatomical images reflecting the properties of the tissues. The image sequences 12 or, more generally, the experimental data, are analyzed by a processing unit 4 arranged for this purpose. Such a processing unit 4 may, for example, consist of one or more microprocessors or microcontrollers implementing suitable application program instructions loaded into storage means of said imaging analysis system. The term "storage means" means any volatile or, advantageously, non-volatile computer memory.Non-volatile memory is a computer memory whose technology allows its data to be retained in the absence of an electrical power supply. It can contain data resulting from inputs, calculations, measurements and / or program instructions. The main non-volatile memories currently available are electrically writable, such as EPROM (Erasable Programmable Read-Only Memory), or electrically writable and erasable, such as EEPROM (Electrically-Erasable Programmable Read-Only Memory), flash, SSD (Solid-State Drive), etc. Non-volatile memories are distinguished from so-called "volatile" memories, the data of which is lost in the absence of an electrical power supply.The main volatile memories currently available are RAM (Random Access Memory), DRAM (dynamic random access memory, requiring regular updating), SRAM (static random access memory requiring such updating during a power shortage), DPRAM or VRAM (particularly suitable for video), etc. A "data memory", in the rest of the document, can be volatile or non-volatile depending on the intended application.

[0006] Said processing unit 4 comprises means for communicating with the outside world to collect the images. Said means for communicating further allow the processing unit 4 to deliver or output, ultimately, a rendering, for example graphic and / or sound, of an estimation or quantification of a biomarker or a pharmacokinetic parameter QI developed by said processing unit 4 from the experimental data 10 and / or 12 obtained by Magnetic Resonance Imaging, to a user 6 of the imaging analysis system via of an output human-machine interface 5. Throughout the document, the term "output human-machine interface" means any device, used alone or in combination, making it possible to output or deliver a graphic, haptic, audio or, more generally, human-perceptible representation of a reconstructed physiological signal to a user 6 of a medical imaging analysis system. Such an output human-machine interface 5 may consist, in a non-exhaustive manner, of one or more screens, loudspeakers or other suitable alternative means. Said user 6 of the analysis system can thus confirm or deny a diagnosis, decide on a therapeutic action that he deems appropriate, carry out further research work, refine adjustment parameters of measuring equipment, etc.Optionally, this user 6 can also configure the operation of the processing unit 4 or the output human-machine interface 5, by means of operating and / or acquisition parameters 16. For example, he can thus define display thresholds or choose the estimated or quantified indicators or parameters for which he wishes to have a representation. The user 6 uses for this the input human-machine interface 8 previously mentioned or a second input interface provided for this. Advantageously, the input 8 and output 5 human-machine interfaces can constitute a single physical entity. Said input 8 and output 5 human-machine interfaces of the imaging analysis system can also be integrated into the acquisition console 2. There is a variant, described in connection with [Fig.2], for which an imaging system, as described previously, further comprises a preprocessing unit 7 for analyzing the image sequences 12, deducing experimental signals 15 therefrom and delivering the latter to the processing unit 4 which is thus relieved of this task.

[0007] Quantification of breast density is important for the management of breast cancer screening in the population. Indeed, high breast density is a risk factor, which a senologist must take into account during radiological interpretation.

[0008] As such, breasts are composed of two types of tissue: - fibroglandular tissue, also known as dense tissue, which supports the breasts and contributes to the production and transport of milk to the nipple; - adipose tissue, the fat that gives the breasts their shape.

[0009] However, it is the distribution of fibroglandular tissue in relation to fatty tissue that determines breast density. A breast is considered dense when it is made up of a large amount of fibroglandular tissue. Breast density is divided into four categories: A, B, C and D, ranging from a completely fatty breast to a completely fibrous breast, according to the BLRADS® classification (acronym Anglo-Saxon for “Breast Imaging-Reporting And Data System”), as illustrated in [Fig.3]: - Category A: almost entirely fatty breasts, which means that the breasts are almost entirely made up of fatty tissue; - Category B: areas of scattered fibroglandular density, meaning that the breasts are mainly made up of fatty tissue, with some small areas of fibroglandular tissue; - Category C: heterogeneously dense breasts, meaning that the breasts are made up of a mixture of fatty and fibroglandular tissue; - Category D: Extremely dense breasts, meaning the breasts are almost entirely made of fibroglandular tissue.

[0010] Today, it is found that high breast density is linked to breast cancer in two ways: - On a mammogram, an X-ray image of the breast used to detect possible breast cancer, cancerous lesions and dense tissues both have a white appearance (like fatty tissues which appear black or gray), making the detection of breast cancer more difficult if the patient's breast density is high. As such, [Fig.4] shows on the left of the image (breast 1), an adipose breast (low breast density) with an easily visible tumor (good contrast): white point contained in the white circle, compared to a dense breast on the right of the image (breast 2) with a tumor that is difficult to detect (white on white); - Many studies have shown that people with high breast density have a higher risk of developing breast cancer than people with lower breast density.

[0011] Such findings demonstrate the importance of accurately determining breast density in order to provide assistance to all healthcare personnel, so that they can define a breast cancer risk level for each patient. For example, a radiologist, faced with a patient with very dense breasts, will look more precisely at the radiological images and will be able, if necessary, to recommend specific treatment, for example by prescribing additional examinations such as MRI as a first-line screening procedure.

[0012] To date, as mentioned in the latest 2013 edition of the BI-RADS® Atlas — III. Breast MRI Reporting System, the recommendation for breast density assessment relies on the visual interpretation of the practitioner, in this case the eye of the radiologist. Indeed, the four categories of breast composition (described previously and illustrated in [Fig. 3]) are defined by the visually estimated content of fibroglandular tissue in the breasts. If the breasts do not contain apparently equal amounts of fibroglandular tissue, the breast with the most fibroglandular tissue will be used to categorize breast composition.

[0013] As such, the human factor remains very present in the process of estimating breast density, which can lead to great variability in the results obtained. Indeed, the practitioner is the sole judge for determining the category of breast density according to the BI-RADS® classification by his own visual interpretation of radiological images.

[0014] In order to address this problem, several methods seeking to estimate breast density more precisely, independently of the practitioner's visual interpretation alone, have been proposed in scientific publications.

[0015] We can cite as an example the study by X. Jing, M. Wielema, AG Monroy-Gonzalez, entitled “Automated Breast Density Assessment in MRI Using Deep Learning and Radiomics: Strategies for Reducing Inter-Observer Variability” and published in the Journal of Magnetic Resonance Imaging 2023. Such a method uses radiomic parameters based on voxel intensities in a region of interest with the description of the use of fourteen parameters based on the three-dimensional shape of the breast. However, for such a method, the calculation of the parameters is carried out only on the global breast area without distinction of the composition of the tissues.

[0016] We can also cite studies that have sought to focus on the percentage volume fraction of mammary gland in the breast. However, the percentage volume fraction of mammary gland in the breast alone cannot be sufficient to categorize breast density within the meaning of the BI-RADS® classification. Indeed, such a volume fraction does not provide information on the dispersed (BI-RADS® category B) or heterogeneous (BI-RADS® category C) appearance of the mammary gland in the breast.

[0017] Thus, in the context of breast cancer diagnosis assistance, the development of a method for determining breast density in a precise, quantitative and automatic manner, based on morphological characteristics of interest of the mammary gland, is of great interest. As such, such a method would make it possible to determine breast density in a standardized and automated manner by eliminating any relative subjectivity and therefore provide valuable and robust assistance to any practitioner with a view to establishing a precise, objective, reliable and reproducible diagnosis in connection with a human pathology affecting the breast.

[0018] To this end, the invention provides, according to a first object, a method for determining breast density from a medical image resulting from an acquisition sequence by a medical imaging device describing a first digital representation of a patient's breast in the form of a plurality of elementary volumes of interest, called "voxels" Said method is implemented by a processing unit of a medical imaging analysis system. Said method comprises: - a step of obtaining a second digital representation of the mammary gland resulting from a segmentation of said first digital representation of the patient's breast, said second digital representation describing the fibroglandular tissue present inside the patient's breast; - a step of determining the volume of the mammary gland from said second digital representation; - a step of determining the breast volume from said first digital representation of the breast of said patient; - a step of producing a first and a second morphological characteristics of interest of the mammary gland, the first and second morphological characteristics of interest of the mammary gland being distinct from one another; - a step of producing breast density from the first and second morphological characteristics of interest of the mammary gland.

[0019] Preferably, one of said first or second morphological characteristics of interest of the mammary gland is taken from a set of morphological characteristics of interest of the mammary gland comprising: - the volume fraction of the mammary gland corresponding to the ratio between the volume of the mammary gland and the breast volume; - the solidity of the mammary gland corresponding to the ratio between the volume of the mammary gland and the volume of the convex envelope enclosing said mammary gland; - the compactness of the mammary gland consisting of a comparison of the area of ​​the mammary gland surface and the volume of the mammary gland corresponding to the ratio ^4 / Vg' being the surface area of ​​the mammary gland and Vg being the volume of the mammary gland; - the ratio of mammary gland surface area to mammary gland volume; - the fractal dimension of the mammary gland reflecting the capacity of the mammary gland to fill the space.

[0020] According to a first embodiment of said method, the step of determining the breast volume consists of starting from the first digital representation in: - a first step of selecting a first voxel translating the distal part of the breast of said patient on the first digital representation; - a second step of selecting a distance as being the shortest distance between a second voxel representing the thorax of said patient on the first digital representation and the first selected voxel; - a third step of determining a sphere having as its center the first selected voxel and as its radius said selected distance; - a fourth step of estimating the breast volume, said breast volume corresponding to the intersection between said determined sphere and a volume characterized by the number of voxels describing tissue multiplied by the elementary volume of each voxel.

[0021] For the first embodiment, the radius of said sphere can advantageously be reduced by a predetermined value, for example to limit the inclusion of unwanted organs in the breast area, such as the pectoral muscle.

[0022] According to a second embodiment of said method, the step of determining the breast volume consists of starting from a third plural digital representation of the patient's torso in: - a first step of selecting a first voxel representing the distal part of the breast of said patient on the third digital representation; - a second step of selecting a distance as being the shortest distance between a second voxel representing the thorax of said patient on the third digital representation and said first selected voxel; - a third step of determining a sphere having as its center said first selected voxel and as its radius said selected distance; - a fourth stage of production of the first digital representation describing the voxels captured by the determined sphere; - a fifth step of estimating breast volume from the first digital representation produced.

[0023] The step of obtaining a second digital representation of the mammary gland by segmenting said first digital representation of the patient's breast may consist of implementing an automatic segmentation technique or a manual segmentation technique.

[0024] In addition, said medical imaging analysis system may further comprise a human-machine output interface. Thus, said method further comprises a step, subsequent to the step of producing the breast density, of triggering an output of the breast density produced by said human-machine output interface.

[0025] According to a second object, the invention relates to a computer program product comprising one or more program instructions executable by the processing unit of a computer. Said program instructions are loadable in a non-volatile memory of said computer and the execution of which by said processing unit causes the implementation of a method according to the invention.

[0026] According to a third object, the invention relates to a storage medium readable by a computer comprising the instructions of such a computer program product.

[0027] According to a fourth object, the invention relates to a medical imaging analysis system comprising a processing unit arranged to implement a method according to the invention, said processing unit being configured to: - obtaining a second digital representation of the mammary gland resulting from a segmentation of said first digital representation of the patient's breast, said second digital representation describing the fibroglandular tissue present inside the patient's breast; - determining the volume of the mammary gland from said second digital representation of the mammary gland; - determining the breast volume from said first digital representation of said patient's breast; - producing a first and a second morphological characteristics of interest of the mammary gland, the first and second morphological characteristics of interest of the mammary gland being distinct from each other; - produce breast density from the first and second morphological characteristics of interest of the mammary gland.

[0028] Such a medical imaging analysis system may further comprise an output human-machine interface, said processing unit being arranged to implement a method according to the invention, said processing unit being further configured to trigger an output of the breast density produced, by said output human-machine interface.

[0029] Finally, the invention relates to a medical imaging analysis system comprising a program memory comprising the program instructions of said computer program product.

[0030] The invention will be better understood and other characteristics and advantages thereof will appear on reading the following description of particular embodiments of the invention, given as illustrative and non-limiting examples, and referring to the appended drawings, among which:

[0031] [Fig.l], already described, illustrates a simplified description of a system for analyzing images obtained by magnetic resonance,

[0032] [Fig.2], already described, illustrates a simplified description of a variant of a system for analyzing images obtained by magnetic resonance,

[0033] [Fig.3], already described, illustrates the four categories of breast density estimation according to the BLRADS® classification,

[0034] [Fig.4], already described, illustrates a comparison between a radiological image representing a breast categorized as adipose and a radiological image representing a breast categorized as dense,

[0035] [Fig.5] illustrates an example of a functional algorithm of a method according to the invention,

[0036] [Fig.6] illustrates a first embodiment of a functional algorithm of a method according to the invention,

[0037] [Fig.7] illustrates the first embodiment described in [Fig.6],

[0038] [Fig.8] illustrates a second embodiment of a functional algorithm of a method according to the invention,

[0039] [Fig.9] illustrates a morphological characteristic of interest produced by a method according to the invention.

[0040] In order to simplify the description, the same reference is used in different figures to designate the same object, element or step. Thus, when the description cites a referenced object, element or step, this object, element or step may be identified in several figures. In addition, the figures as well as the description are given as non-limiting examples of embodiment.

[0041] As a preamble, for the purposes of the present invention, an image is a BGR matrix digital representation of elements called “voxels”, of thickness k (along a transverse axis), BGR(i,j), i and j being indices of integer values ​​to identify the voxel located at row i and column j of the BGR matrix, i.e. a graphical representation in the form of a table or matrix, of BGR(i,j) elements each encoding a triplet of integer values ​​between zero and two hundred and fifty-five, according to the RGB color coding, acronym for “Red Green Blue”, also known by the acronym RGB, the Anglo-Saxon acronym for “Red Green Blue”. Such computer color coding is the closest to the hardware currently available for constituting human-machine output interfaces such as computer screens.In general, the latter reconstruct a color by additive synthesis from three primary colors, red, green and blue, forming on the screen a mosaic generally too small to be discriminated by a human eye. The RGB coding indicates a luminous intensity value for each of these primary colors. Such a value is generally coded on a byte and therefore belongs to an interval of integer values ​​between zero and two hundred and fifty-five. The invention cannot be limited to this type of coding. Other color codings could alternatively be used. In this case, each voxel or element of the BGR table would describe a set of numerical values ​​adapted to said coding instead of the triplet mentioned above for RGB coding.

[0042] A method 100 for determining breast density Dm in accordance with the invention and illustrated in [Fig.5], is advantageously translated into the form of a product computer program whose program instructions are intended to be implanted in the program memory of an element of a medical imaging system, such as the system S according to figures 1 and 2, for example a computer or a computer server or, more generally, any electronic object having sufficient computing power.

[0043] [Fig. 5] thus illustrates a method 100, in accordance with the invention, for determining the breast density Dm from a medical image of a patient's breast, resulting from an acquisition sequence by an imaging device, such as the device 1 of the medical imaging analysis system S illustrated by FIGS. 1 and 2, in connection with a plurality of n voxels. Such a method 100 can be implemented by the processing unit 4 of such a medical image analysis system S.

[0044] For illustrative but non-limiting purposes, such a method 100 according to the invention is advantageously described in the remainder of this document and implemented for the processing of a medical image, describing a first digital BGR representation of a patient's breast, resulting from an acquisition sequence by a medical magnetic resonance imaging device 1. Preferably but not limitatively, said medical image may be a so-called "Tiw-weighted" image, resulting from a so-called "anatomical" acquisition sequence.

[0045] Thus, such a method 100 according to the invention, illustrated in [Fig. 5], consists of quantifying the mammary density Dm from at least two morphological characteristics of interest Qil and Qi2 of the mammary gland resulting from the analysis and processing of said medical image. To do this, said method 100 according to the invention can be described, as suggested by the exemplary embodiment illustrated in [Fig. 5], as comprising five main steps respectively referenced 110, 120, 130, 140 and 150.

[0046] A first step 110 consists of producing, from the first digital representation BGR, a second digital representation FGT, of the same dimensions as said first digital representation BGR, to describe the fibroglandular tissue, otherwise called mammary gland, present inside the breast from which the first digital representation BGR is derived. This first step 110 thus consists of obtaining a new digital representation FGT by segmentation of the first digital representation BGR of the breast in order to highlight the mammary gland. In other words, such a segmentation step 110 makes it possible to detect and highlight the voxels BGR(i,j) of said first digital representation BGR identified as fibroglandular tissue contained in the analyzed breast. Such a segmentation step 110 can be implemented by a manual or automatic segmentation technique.

[0047] For example, the purpose of such a step 110 may consist of forming a second digital representation FGT of the same dimensions as the first digital representation BGR, each voxel or element FGT(i,j) of which describes a first characteristic value when the corresponding voxel BGR(i,j) in the first representation BGR describes fibroglandular tissue and a second characteristic value otherwise. Such a second digital representation FGT is a binary matrix representation: said first characteristic value translating the Boolean value “true” (“true” according to English terminology) and said second characteristic value translating the Boolean value “false” (“false” according to English terminology).Thus, such a second FGT digital representation can be, for example, in the form of a black and white image according to which each voxel describes an integer value equal to zero or two hundred and fifty-five to translate a minimum (black) or maximum (white) light intensity. According to such an example, the voxels take the value of "0" to describe the Boolean information "false" and the value "255" to describe the Boolean information "true". Thus, it results in the fibroglandular tissue (mammary gland) appearing in white and the rest of the breast constituents (adipose tissue, skin, nipple, ...) and the other peripheral organs in black.

[0048] However, the invention cannot be limited to the single example of segmentation technique described above. Any other segmentation technique could be used for the implementation of step 110, such as, for example: - a segmentation technique by thresholding according to the intensity of each voxel BGR(i,j) of the first digital representation BGR. Thus, a predetermined threshold is set automatically or manually. For example, such a threshold can be equal to 150. If the intensity value of the voxel is below the threshold, i.e. strictly less than 150, we associate it with the Boolean value “false” i.e. for example the value “0”, minimum intensity: black color. If the intensity value of the voxel is equal to or greater than the threshold, i.e. greater than or equal to 150, we give it the Boolean value “true”, characteristic of the fibroglandular tissue, i.e. for example the value “255”, maximum intensity: white color.Thus, such a second FGT digital representation can be in the form of a binary image, for example expressed in black and white, on which the fibroglandular tissue (mammary gland) appears in white and the rest of the breast constituents (adipose tissue, etc.) in black. Alternatively, such a threshold can be adaptive instead of being fixed. Using a cursor, it is possible to adjust said threshold and see the differences in rendering and results obtained; . - a contour segmentation technique of one or more areas of interest on the first BGR digital representation. With such a technique, we seek to delimit and isolate one or more areas of interest on said first BGR digital representation. The result is generally presented in the form of a set of voxel chains; - a region growth segmentation technique, otherwise known as "growing-region" in English terminology. This technique consists of progressively growing one or more regions of interest around a voxel defined as a starting point by agglomerating neighboring voxels using a similarity measure; - a deep learning segmentation technique, otherwise known as “Deep Learning”: statistical elements are used to automatically segment different parts, here the fibroglandular tissue, of the first digital BGR representation in order to obtain a second digital FGT representation.

[0049] A second 120 step of the method 100 according to the invention consists of using the second digital representation FGT to determine the volume of the mammary gland Vg. At the end of the segmentation step 110, a second digital representation FGT describing the mammary gland is obtained. Thus, we have complete spatial knowledge of the mammary gland, otherwise called fibroglandular tissue. Thus, for a given representation of the breast, the volume of the mammary gland is determined by counting the number of voxels representing the fibroglandular tissue and multiplying the number of voxels obtained by the elementary volume of a voxel.

[0050] A third 130 step of the method 100 according to the invention consists in using the first digital representation BGR to determine the breast volume Vm. In order to precisely determine such a breast volume Vm, in this case to free oneself in particular from the thorax and the axillary zone not comprising fibroglandular tissue, it is necessary to materialize a boundary delimiting the specific tissues within the “other” tissues such as for example those of the thorax and / or the axillary zone. Such a boundary can be symbolized, for example, by a contour line manually or automatically delimiting the voxels translating the breast tissues from the voxels translating the tissues of the thorax and / or the axillary zone on the first digital representation BGR.

[0051] A first preferred embodiment for this step 130 is illustrated in Figures 6 and 7. As such, such a step 130 may, initially, consist of a first sub-step 131 aimed at selecting a first voxel BGR(il,jl) on the first digital representation BGR. Such a first voxel BGR(il,jl) translates the distal part of the breast of said patient on the first digital representation BGR. In this document, the term “distal part” means the part of the breast furthest from the of the thorax, the thorax being the part located below the dotted line shown in [Fig.7]. Once this first sub-step 131 has been carried out, a second sub-step 132 is carried out in order to define the shortest distance between said first voxel BGR(il,jl) and a second voxel BGR(i2,j2) translating the patient's thorax on said first digital representation BGR. A third sub-step 133 is carried out, such a step 133 consisting of determining a sphere of which: - the radius R is the distance selected in the previous step 132, translating the shortest distance between said first voxel BGR(il,jl) and said second voxel BGR(i2,j2), and -the center C, is the first voxel BGR(il,jl) selected in step 131. Finally, a fourth step 134 is carried out to estimate the breast volume Vm. Such a volume Vm corresponds to the intersection between said sphere determined in the previous step 133 and the volume characterized by the number of voxels describing tissue on the first digital representation BGR, i.e. all the voxels except those describing empty space, multiplied by the elementary volume of the voxels.

[0052] In order to further optimize the determination of such a breast volume Vm, and in particular to free oneself from the pectoral muscle, it may be envisaged to reduce the radius R by a predetermined value in order to obtain a sphere smaller than that obtained in sub-step 133. Thus, said sphere obtained would not include the pectoral muscle.

[0053] Alternatively, a second embodiment for this step 130 is illustrated in [Fig.8]. As such, for such a second embodiment, the first digital representation BGR, which will be called BGR' for this embodiment, is produced from a third digital representation TGR corresponding to a plural digital representation, namely a digital representation not describing only the breast, for example the entire torso including in particular the thorax and the upper part of the abdomen. Thus, for this second embodiment, step 130 may preferentially, initially, consist of a first sub-step 131' aimed at selecting a first voxel TGR(il,jl) on the third digital representation TGR. Such a first voxel TGR(il,jl) translates the distal part of the breast of said patient on the third digital representation TGR.In this document, the term "distal part" means the part of the breast furthest from the thorax. Once this first sub-step 131' has been carried out, a second sub-step 132' is carried out in order to define the shortest distance between said first voxel TGR(il,jl) and a second voxel TGR(i2,j2) translating the patient's thorax onto said third digital representation TGR. A third sub-step 133' is carried out, such a step 133' consisting of determining a sphere of which: . - the radius R' is the distance selected in the previous step 132', reflecting the shortest distance between said first voxel TGR(il,jl) and said second voxel TGR(i2,j2), and - the center C' is the first voxel TGR(il,jl) selected at step 131'. A fourth step 134' consists of producing the first digital representation BGR' from the third digital representation TGR. To do this, the sphere obtained in the previous step 133' is applied to the third representation TGR. The first digital representation BGR' describes, in this case, all the voxels of the third digital representation TGR captured by said sphere. All the other non-captured voxels are replaced by a value assigned to describe empty space. Thus, the voxels supposed to describe organs other than the breast are masked or even erased, such as voxels describing for example the thorax and / or the axillary zone. Finally, a fifth sub-step 135' is carried out to estimate the breast volume Vm. Such a volume Vm is characterized by the number of voxels of the first digital representation BGR', obtained in the previous step 134', not describing empty space, multiplied by the elementary volume of the voxels.

[0054] A fourth step 140 of the method 100 according to the invention consists of jointly exploiting the first BGR or BGR' and second FGT digital representations as well as the mammary gland volume Vg and the mammary volume Vm determined respectively in steps 120 and 130 to produce at least two distinct morphological characteristics of interest Qil and Qi2 of the mammary gland.

[0055] Such morphological characteristics of interest Qil and Qi2 are preferentially drawn from a set of morphological characteristics of interest of the mammary gland including in particular: - the volume fraction of the mammary gland which corresponds to the ratio between the volume of the mammary gland Vg and the breast volume Vm; - the solidity of the mammary gland corresponding to the ratio between the volume of the mammary gland Vg and the volume of the convex envelope enclosing said mammary gland, the convex envelope being illustrated in [Fig.9], where the black dots represent the voxels describing the mammary gland and the dotted lines the convex envelope, in this case the smallest surface allowing the capture of all the voxels describing the mammary gland; - the compactness of the mammary gland consisting of the comparison of the area of ​​the mammary gland surface and the volume of the mammary gland Vg, such a comparison corresponds to the ratio yyg with A being the gland surface breast; - the ratio of mammary gland surface area to mammary gland volume Vg; - the fractal dimension of the mammary gland, reflecting the capacity of the mammary gland to fill space, based for example on the Minkowski-Bouligand dimension. Thus, the more all the voxels are concentrated in an area, the less the fractal character is considered and therefore the higher the fractal dimension. The set of morphological characteristics of interest of the mammary gland mentioned above is not limited to these examples alone and could include other characteristics likely to describe the morphology of said mammary gland.

[0056] A fifth step 150 of the method 100 according to the invention consists of aggregating or jointly exploiting the two distinct morphological characteristics of interest Qil and Qi2 produced in step 140 to precisely determine the breast density Dm relative to the breast analyzed with regard to the BLRADS® classification for example or any other classification defining an estimated scale of the breast density Dm. It may thus be envisaged an additional step aimed at converting said morphological characteristics of interest produced Qil and Qi2 with respect to a scale linked to the classification which may be retained.

[0057] Thus, by combining, according to one or more pre-established rules, two morphological characteristics of interest of the mammary gland Q11 and Qi2 distinct from each other, it is then possible to determine more precisely the mammary density Dm. As an illustrative but non-limiting example for a breast to be analyzed, step 140 results in the production of two morphological characteristics of interest of the mammary gland Q11 and Qi2 of said breast, such as for example the fractal dimension calculated at two point three, corresponding to a category C or D according to the BLRADS® classification, and the solidity of the mammary gland calculated at zero point six, corresponding to a category D according to the BLRADS® classification.Thus, by combining these two characteristics Qil and Qi2 according to, for example, a pre-established rule specifying that the most favorable category is retained, it emerges that the breast density Dm of the breast to be analyzed would be categorized D according to the BLRADS® classification.

[0058] Such a method 100 may further comprise an optional step 160, subsequent to step 150, consisting of causing an output of the breast density Dm produced in step 150 by a human-machine output interface 5 of the medical imaging system 1 implementing such a method 100, such as that described in connection with [Fig.5].

[0059] Upstream of the determination of the mammary gland volume Vm, when the acquisition phase by the imaging device results in n digital representations BGR describing respectively “slices” or sections of the breast, the mammary gland volume Vm can advantageously consist of the sum of the respective volumes of mammary gland calculated from the n binary digital representations FGT obtained respectively after segmentation of the n representations BGR. In this case, a step of selecting said sphere is carried out and consists of choosing the sphere of the digital representation for which the distance between the first voxel, representing the distal part of the breast, and the second voxel, representing the thorax, is the greatest and of applying said sphere to the n digital representations.

[0060] Alternatively, reasoning as described in the present description could be applicable to a three-dimensional digital representation for which the same steps described previously would be applied.

[0061] It will be appreciated by those skilled in the art that the present disclosure is not limited to what is particularly shown and described above. Other modifications may be envisaged without departing from the scope of the present invention defined by the appended claims. In particular, in the preferred example described herein, the method 100 is implemented for the processing of a medical image, the result of a so-called “anatomical” acquisition sequence by a medical magnetic resonance imaging device 1. However, any other type of acquisition sequence could be used for the implementation of the method 100 according to the invention. Similarly, illustrative examples of determining breast density Dm have been given with regard to the BI-RADS® classification recognized to date and presenting a four-level scale.However, any other classification (having a scale of one to n levels) could be used to give a result or an estimate of the breast density Dm at the end of step 150 of said method 100 according to the invention.

Claims

Claims

1. Method (100) for determining breast density (Dm) from a medical image resulting from an acquisition sequence by a medical imaging device (1) describing a first digital representation (BGR, BGR') of a patient's breast in the form of a plurality of elementary volumes of interest, called "voxels", said method (100) being implemented by a processing unit (4) of a medical imaging analysis system (S), said method (100) comprising: - a step (110) of obtaining a second digital representation (FGT) of the mammary gland resulting from a segmentation of said first digital representation (BGR, BGR') of the patient's breast, said second digital representation (FGT) describing the fibroglandular tissue present inside the patient's breast; - a step (120) of determining the volume of the mammary gland (Vg) from said second digital representation (FGT);- a step (130) of determining the breast volume (Vm) from said first digital representation (BGR, BGR') of the breast of said patient; - a step (140) of producing a first and a second morphological characteristics (Qil, Qi2) of interest of the mammary gland; the first and second morphological characteristics (Qil, Qi2) of interest of the mammary gland being distinct from one another; - a step (150) of producing the breast density (Dm) from the first and second morphological characteristics (Qil, Qi2) of interest of the mammary gland.;

2. Method (100) according to claim 1 for which one of said first or second morphological characteristics (Qil, Qi2) of interest of the mammary gland is taken from a set of morphological characteristics of interest of the mammary gland comprising: - the volume fraction of the mammary gland corresponding to the ratio between the volume (Vg) of the mammary gland and the mammary volume (Vm); - the solidity of the mammary gland corresponding to the ratio between the volume (Vg) of the mammary gland and the volume of the convex envelope enclosing said mammary gland; - the compactness of the mammary gland consisting of a comparison of the surface area (A) of the mammary gland and the volume (Vg) of the mammary gland corresponding to the ratio yy yyg, A being the surface area of ​​the mammary gland and Vg being the volume of the mammary gland; - the ratio of the surface area (A) of the mammary gland to the volume (Vg) of the mammary gland; - the fractal dimension of the mammary gland reflecting the capacity of the mammary gland to fill the space.

3. Method (100) according to one of the preceding claims for which the step (130) of determining the breast volume (Vm) consists of starting from the first digital representation (BGR) in: - a first step (131) of selecting a first voxel (BGR(il,jl)) translating the distal part of the breast of said patient on the first digital representation (BGR); - a second step (132) of selecting a distance as being the shortest distance between a second voxel (BGR(i2,j2)) translating the thorax of said patient on the first digital representation (BGR) and the first voxel (BGR(il,jl)) selected; - a third step (133) of determining a sphere having as its center (C) the first selected voxel (BGR(il,jl)) and as its radius (R) said selected distance;- a fourth step (134) of estimating the breast volume (Vm), said breast volume (Vm) corresponding to the intersection between said determined sphere and a volume characterized by the number of voxels describing tissue multiplied by the elementary volume of each voxel.;

4. Method (100) according to the preceding claim for which the radius (R) of said sphere is reduced by a predetermined value.

5. Method (100) according to one of claims 1 or 2 for which the step (130) of determining the breast volume (Vm) consists of starting from a third plural digital representation of the patient's torso (TGR) in: - a first step (131') of selecting a first voxel (TGR(il,jl)) translating the distal part of the breast of said patient on the third digital representation (TGR); - a second step (132') of selecting a distance as being the shortest distance between a second voxel (TGR(i2,j2)) translating the thorax of said patient on the third digital representation (TGR) and said first voxel (TGR(il,jl)) selected; - a third step (133') of determining a sphere having as its center (C') said first voxel (TGR(il,jl)) selected and as its radius (R') said selected distance; - a fourth step (134') of producing the first digital representation (BGR') describing the voxels captured by the determined sphere; - a fifth step (135') of estimating the breast volume (Vm) from the first digital representation (BGR') produced.

6. Method (100) according to one of the preceding claims, for which the step (110) of obtaining a second digital representation (FGT) of the mammary gland by segmenting said first digital representation (BGR, BGR') of the patient's breast consists of implementing an automatic segmentation technique or a manual segmentation technique.

7. Method (100) according to one of the preceding claims, for which said medical imaging analysis system (S) further comprises an output human-machine interface (5) and said method (100) further comprises a step (160), subsequent to the step (150) of producing the breast density (Dm), of triggering an output of the breast density (Dm) produced by said output human-machine interface (5).

8. Computer program product comprising one or more program instructions executable by the processing unit (4) of a computer, said program instructions being loadable into a non-volatile memory of said computer and the execution of which by said processing unit (4) causes the implementation of a method (100) according to any one of the preceding claims.

9. A computer-readable storage medium comprising the instructions of a computer program product according to the preceding claim.

10. Medical imaging analysis system (S) comprising a processing unit (4) arranged to implement a method according to any one of claims 1 to 6, said processing unit (4) being configured to: - obtain a second digital representation (FGT) of the mammary gland resulting from a segmentation of said first digital representation (BGR, BGR') of the patient's breast, said second digital representation (FGT) describing the fibroglandular tissue present inside the patient's breast; - determine the mammary gland volume (Vg) from said second digital representation (FGT) of the mammary gland; - determine the mammary volume (Vm) from said first digital representation (BGR, BGR') of the patient's breast;- producing a first and a second morphological characteristics (Qil, Qi2) of interest of the mammary gland, the first and second morphological characteristics (Qil, Qi2) of interest of the mammary gland being distinct from each other; - producing the mammary density (Dm) from the first and second morphological characteristics (Qil, Qi2) of interest of the mammary gland.;

11. Medical imaging analysis system (S) according to the preceding claim further comprising an output human-machine interface (5), said processing unit (4) being arranged to implement a method according to claim 7, said processing unit (4) being further configured to trigger an output of the breast density (Dm) produced, by said output human-machine interface (5).

12. Medical imaging analysis system (S) according to one of claims 10 or 11 comprising a program memory comprising the program instructions of a computer program product according to claim 8.

Images