Thermography analysis device for detecting anomalies in breast tissues

Abstract

The invention relates to a thermography analysis device (1) by cutaneous contact for detecting anomalies in breast tissues, referred to as device (1). The device (1) comprises: two arrays (2) of temperature sensors (3) intended to be in contact with the skin of a respective breast and being arranged to homogeneously tile each of the breasts, and an acquisition unit (4) which over the course of time acquires temperature data coming from the temperature sensors and is arranged to communicate with a processing unit (5). The processing unit (5) is arranged to calculate, from the temperature data, a temperature deviation between the two breasts as a function of time, to select, from the temperature data as a function of time, an appropriate data processing model, and to determine information representative of the presence or not of an anomaly by processing the temperature data with the selected processing model.

Description

DESCRIPTION "Thermographic analysis device for detecting abnormalities in breast tissue" technical field

[0001] The present invention belongs to the field of detecting anomalies in breast tissue.

[0002] The invention relates, in particular, to the detection of anomalies in breast tissue by skin contact thermography.

[0003] The invention is aimed in particular at the detection of tumor tissue within breast tissue.

[0004] The invention also relates to support for the detection of breast abnormalities by skin contact thermography. State of the art

[0005] Tumor tissues are distinguished from surrounding healthy tissues by their temperature due to the inflammation caused and angiogenesis.

[0006] We know of commercial bras in the state of the art for the detection of cancerous tissue by thermography.

[0007] As an example, the "iTBra" bra offers tumor tissue detection based on temperature measurement over a circadian cycle. This model requires temperature data to be measured for at least one circadian cycle (at least 24 hours). The temperature of healthy breast tissue evolves according to circadian rhythms, whereas the temperature of cancerous tissue is not dependent on them.

[0008] The "Evabra" bra is also known to offer tumor tissue detection based on continuous temperature measurement. The temperature measurement must be uninterrupted and carried out for a minimum of 60 to 90 minutes.

[0009] One object of the present invention is to remedy at least one of the drawbacks of the prior art.

[0010] Another aim of the invention is to provide a bra and a method for detecting anomalies in breast tissue: - by skin contact measurement, and / or - by thermography, and / or - to detect the presence of malignant and / or benign abnormalities, and / or - to detect abnormalities, particularly the presence of tumor tissue, at an early stage, and / or - to distinguish the presence of a malignant anomaly from a benign anomaly, and / or - allowing for a reliable response, i.e., with few false positives, even for small temperature variations, - requiring a reduced temperature data acquisition time and / or shorter than that of prior art processes and bras, and / or - requiring reduced temperature data processing time. Description of the invention

[0011] To this end, the invention relates to a thermographic analysis device for detecting abnormalities in breast tissue, referred to as the device. The device comprises: - two networks of temperature sensors intended to be in contact with the skin of a respective breast and arranged, positioned or distributed to cover each breast evenly, - a unit for acquiring temperature data over time, from temperature sensors, arranged to communicate with a processing unit.

[0012] The processing unit is arranged to: - Calculate, based on temperature data, the temperature difference between the two breasts as a function of time. - select, preferably using the processing unit, from the temperature data, or from the temperature data and phenotypic data, a corresponding, appropriate, representative temperature profile and / or one associated with the temperature data, and / or - select, based on the temperature difference as a function of calculated time and / or the selected temperature profile, an appropriate, suitable, or representative data processing model, - determine information representative of the presence or absence of an anomaly by processing temperature data with the selected processing model.

[0013] Preferably, the processing unit and / or, respectively, the acquisition unit is arranged and / or configured to communicate and / or be connected, by wired or wireless link, to the acquisition unit and / or, respectively, to the processing unit.

[0014] The processing unit can be a remote unit. Preferably, the processing unit and / or the acquisition unit is arranged and / or configured to communicate with the acquisition unit.

[0015] The processing unit and / or the acquisition unit may include a transmitter / receiver. The processing unit and / or the acquisition unit may understand a transmitter / receiver arranged and / or configured to communicate and / or transmit: - temperature data from sensors, and / or - the data processed by the processing unit, i.e. the information representing the presence or absence of an anomaly.

[0016] Preferably, the processing unit is further arranged to determine, from temperature data and / or from the calculated temperature difference, a statistical signature of the temperature difference between the two breasts.

[0017] Preferably, the processing unit is further arranged to select, from the statistical signature of the determined temperature deviation and the selected temperature profile, the appropriate data processing model.

[0018] Preferably, each temperature sensor network includes a one-piece flexible support on which the temperature sensors are arranged.

[0019] Preferably, the flexible support comprises a set of distinct tabs or segments of actinomorphic geometry.

[0020] Preferably, the flexible support includes a central portion intended to be positioned on the mammary papilla of a breast.

[0021] Preferably, the tabs extend radially from a central portion of the support. Preferably, the tabs are arranged to and / or intended to extend radially from the mammary papilla towards the periphery of the breast.

[0022] Preferably, each of the tabs is separated from the two adjacent tabs by two slots located on either side of each tab.

[0023] Preferably, the central portion is arranged to and / or intended to extend primarily radially from the mammary papilla towards the periphery of the breast

[0024] Preferably, one or each sensor network includes temperature sensors arranged for and / or intended to be positioned on the breast, the mammary papilla and / or in the vicinity of the breast.

[0025] Preferably at least some, preferably only some, preferably all the sensors except those arranged for and / or intended to be positioned on the mammary papilla and / or in the vicinity of the breast, temperature sensors are distributed in a diamond pattern on the flexible support.

[0026] Preferably, each of the four vertices of a rhombus corresponds to a temperature sensor.

[0027] Preferably, for each diamond considered, three temperature sensors are placed on a given tab and one temperature sensor is placed on a tab adjacent to the given tab.

[0028] Preferably, for each sensor on each tab, three temperature sensors are arranged on a given tab and one temperature sensor is arranged on a tab adjacent to the given tab.

[0029] According to the invention, a bra for detecting anomalies in breast tissue by thermography is also proposed, comprising the device according to the invention.

[0030] The bra according to the invention further comprises two cups.

[0031] Each network of temperature sensors intended to be in contact with the skin of a respective breast and to cover one of the breasts evenly is arranged inside one of the respective cups.

[0032] Each of the caps is designed to thermally insulate the temperature sensors from the external environment.

[0033] Alternatively, the analysis device can be described as forming or being arranged to form the bra of the invention.

[0034] According to the invention, a method for detecting anomalies in breast tissue by thermography, called the method, is also proposed.

[0035] The process includes the following steps: - calculate, using the processing unit, a temperature difference between the two breasts as a function of time, based on temperature data from each breast. - select, using the processing unit, from the temperature data, or from the temperature data and phenotypic data, a selected temperature profile, and / or - select, using the processing unit, based on the temperature deviation as a function of calculated time and the selected temperature profile, an appropriate temperature data processing model, - determine information representative of the presence or absence of an anomaly by processing, using the processing unit, temperature data with the selected processing model.

[0036] The data processing model, called a model, can be chosen or selected from or among a set of models in a database.

[0037] Preferably, the temperature profile is selected or chosen from a set of temperature profiles, or from a set of temperature profiles and a set of phenotypic profiles, in a database.

[0038] Ideally, each temperature profile is based on: - temperature data from one or both breasts, and / or of phenotypic data.

[0039] Preferably, the process includes a step of acquiring temperature data using an acquisition unit.

[0040] Preferably, temperature data is measured using both temperature sensor networks.

[0041] According to the invention, the method may not include an acquisition step. The method can be implemented using temperature data from each breast that is provided, transmitted or received (for example by an external or remote element) or stored (for example in a database), or even using simulated data (which may be stored or transmitted data).

[0042] The acquisition step can be implemented independently of the other steps in the process.

[0043] Preferably, the process further includes a step of determining, from temperature data and / or from the calculated temperature difference, a statistical signature of the temperature difference between the two breasts.

[0044] Preferably, the temperature data processing model selection is implemented from the statistical signature of the determined temperature deviation and the selected temperature profile.

[0045] Preferably, the data processing model selection step is implemented, among other things, from a variation in the temperature difference between the two breasts over time and the selected temperature profile.

[0046] Preferably, the statistical signature and / or temperature difference corresponds to or is representative of a variation in the temperature difference between the two breasts over time or as a function of time.

[0047] Preferably, the temperature data form a set of temperature pairs corresponding to the temperature of each breast at a given time t or the average temperature of each breast over a given time interval.

[0048] Preferably, the temperature data form or comprise a data matrix that may be a function of time and an associated or relative position (preferably on and / or near the breast) for each individual temperature data point. Preferably, the positions of each temperature data point correspond to the position of each temperature sensor arranged to be in contact with a different portion or point of each breast, the mammary papilla, and / or the surrounding area of ​​the breast. Preferably, the temperature data form or comprise a separate matrix for each of the two breasts.

[0049] Preferably, the temperature difference and / or statistical signature is determined by comparing the temperature data of one breast, chosen as reference temperature data, with the temperature data of the other breast.

[0050] Preferably, the temperature data for each breast comprise or are dissociated into n sets of temperature data, each corresponding to a distinct area of ​​the breast, called a quadrant, where n is an integer.

[0051] Preferably, the calculation of the temperature difference and / or the selection of the data processing model and / or the determination of the statistical signature is implemented from or based on or according to the set of temperature data from the n quadrants.

[0052] Preferably, the process includes a step of determining, from the statistical signature or temperature difference and the selected temperature profile, one or more indicators representative of a trend in the temperature difference between the two breasts.

[0053] Preferably, the step of selecting the temperature data processing model is implemented from the determined indicator(s).

[0054] Preferably, temperature data processing is implemented by a statistical or deep learning model.

[0055] Preferably, the process includes a step of training the selected model from a training database. The training database includes: - input data consisting of a set of temperature data, each set including temperature data from each of the two breasts of the same subject, - output data consisting of information representing the presence or absence of an anomaly associated with each set of temperature data.

[0056] Preferably, the training step is implemented independently for each processing model using training data corresponding to the temperature deviation and / or associated statistical signature and the selected temperature profile.

[0057] According to the invention, a data processing device is also proposed comprising means arranged and / or programmed and / or configured to implement the process according to the invention.

[0058] According to the invention, a computer program is also proposed comprising instructions which, when the program is executed by a computer, lead the computer to implement the process according to the invention.

[0059] According to the invention, a computer-readable medium is also proposed comprising instructions which, when executed by a computer, lead the computer to implement the process according to the invention.

[0060] The invention, in particular the method according to the invention, including the information representing the presence or absence of an anomaly provided by the method according to the invention, does not relate to, nor is it intended to, nor does it permit or provide a therapeutic diagnosis.

[0061] The information, representative of the presence or absence of an anomaly, provided by the process according to the invention constitutes only information or an indicator not constituting and not having value as a therapeutic diagnosis.

[0062] Preferably, the thermographic analysis device for detecting anomalies in breast tissue and / or preferably the bra for detecting anomalies in breast tissue by thermography according to the invention is suitable, preferably is particularly adapted, more preferably is designed, and particularly advantageously is specially designed, for implementing the method of detecting anomalies in breast tissue by thermography according to the invention. Furthermore, any feature of the thermographic analysis device for detecting anomalies in breast tissue and / or the bra for detecting anomalies in breast tissue by thermography according to the invention is directly applicable to the method of determining and detecting anomalies in breast tissue by thermography, and vice versa. Brief description of the FIGURES

[0063] The invention will be better understood upon reading the following description, given solely by way of non-limiting example and made with reference to the accompanying drawings in which: - FIGURE 1 is a schematic representation of an embodiment of the thermographic analysis device for detecting anomalies in breast tissue according to the invention, - FIGURE 2 is a schematic representation, identical to FIGURE 1, of the thermographic analysis device for detecting anomalies in breast tissue according to the invention, - FIGURE 3 is a photograph, front view, of a bra comprising the thermographic analysis device according to the invention, - Figure 4 is a graph illustrating a particular statistical signature of the temperature difference between the two breasts, - Figure 5 is a schematic representation illustrating an example of the temperature profile selection step based on temperature data. - FIGURE 6 is a graph illustrating the average breast temperature per breast as a function of breast density, noted breast type on FIGURE 6.

[0064] It is understood that the embodiments described below are by no means exhaustive. In particular, variants of the invention may be conceived comprising only a selection of the features described below, isolated from the other features described, if this selection of features is sufficient to confer a technical advantage or to differentiate the invention from the prior art. This selection includes at least one preferably functional feature without structural details, or with only a portion of the structural details if this portion alone is sufficient to confer a technical advantage or to differentiate the invention from the prior art.

[0065] In particular, all the variants and embodiments described can be combined with each other if there are no technical obstacles to this combination.

[0066] In the figures and in the rest of the description, elements common to several figures retain the same reference. Detailed description of the FIGURES

[0067] According to the non-limiting embodiment, an example of implementing the method for detecting anomalies in breast tissue by thermography, referred to as the method, according to the invention, is described. The method comprises the step of calculating, using a processing unit 5, a temperature difference between the two breasts based on temperature data from each breast. The temperature difference between the two breasts can, for example, be determined by comparing the temperature data from each breast.

[0068] The process further includes the step of selecting, using the processing unit 5, from temperature data, or from temperature data and phenotypic data, a corresponding temperature profile.

[0069] The process further includes the step of selecting, using the processing unit 5, from the temperature deviation and the selected temperature profile, an appropriate temperature data processing model.

[0070] Finally, the process includes the step of determining information representative of the presence or absence of an anomaly by processing, using the processing unit 5, the temperature data with the selected processing model.

[0071] According to a first, non-limiting aspect of the invention, the temperature profile is a function of temperature data. Therefore, the temperature profile is selected, from measured temperature data, from a database comprising a set of temperature profiles.

[0072] According to a second, non-limiting aspect of the invention, the temperature profile is a function of measured temperature data and phenotypic data. Furthermore, the temperature profile is selected, from the measured temperature data and phenotypic data, from a database comprising a set of temperature profiles.

[0073] The temperature profile can be selected using any method, for example statistical, known to those skilled in the art. By way of non-limiting example, the temperature profile can be selected by analysis of variance between the temperature profiles in the database and: - measured temperature data, or - measured temperature data and phenotypic data.

[0074] As an illustrative example, Figure 5 shows an example of the temperature profile selection step based on measured temperature data and phenotypic data. According to the non-limiting embodiment, the phenotypic data considered include age, Body Mass Index (BMI), breast density, and breast size.

[0075] Phenotypic data refers to the data of a user of the analysis device according to the invention. Phenotypic data may include or relate to: morphological data and / or subjective data, for example related to habits, behaviors, perceptions or feelings.

[0076] Preferably, the phenotypic data are entered or provided by the user of the analysis device according to the invention (for example, in the context of a questionnaire).

[0077] The invention does not allow, nor is it intended to, provide a therapeutic diagnosis.

[0078] The ingenious use of the temperature difference between the two breasts and the selected temperature profile eliminates the need for reference data, as is the case in the prior art. Reference data used in the prior art can be standard or physiological data, or data previously measured over a significant period (typically more than one hour) on the individual whose temperature will be measured to determine the presence or absence of an abnormality.

[0079] The advantageous use of the temperature difference between the two breasts is based on the fact that there is only a tiny probability that abnormalities are present on each breast.

[0080] Determining the presence or absence of an anomaly using a pre-selected model also offers an advantage over prior art methods. Indeed, selecting the processing model based on the calculated temperature difference and the chosen temperature profile allows for the use of a processing model that corresponds to the individual's profile. The inventors observed a limited number of physiological breast temperature profiles. Each subject corresponds to a specific temperature profile. Therefore, processing temperature data with a particular processing model that matches the subject's / individual's profile helps to limit the number of false positives and improve the accuracy of the resulting representative information.

[0081] As an example, the selection of the treatment model can be carried out by any statistical method, for example correlation, known to a person skilled in the art.

[0082] The aforementioned technical effects of the method according to the invention, in particular but not exclusively the use of a treatment model corresponding to the subject's profile, make it possible to determine whether or not an anomaly is present from a limited temperature data acquisition time. Typically, the method is implemented from a data acquisition time of temperature between 20 and 40 minutes, typically an acquisition time of 30 minutes.

[0083] According to a first improvement, the method includes determining a statistical signature of the temperature difference between the two breasts. Preferably, the statistical signature is determined from temperature data from each breast. The selection of the appropriate data processing model is performed based on the determined statistical signature and the selected temperature profile.

[0084] Determining the statistical signature reduces the amount of temperature data required and / or the time required to select the processing model. Determining the statistical signature also helps to mitigate the effects of temperature measurement noise.

[0085] According to a second improvement, the process can include a step of subsampling the temperature data acquisition time into a subset of temperature data.

[0086] The process advantageously includes calculating an average temperature for each sub-assembly. Calculating an average temperature for each sub-assembly helps to limit the effects of temperature measurement noise and / or missing or aberrant temperature data.

[0087] According to a third improvement, the method for collecting temperature data for each breast comprises n sets of temperature data, each corresponding to a distinct area of ​​the breast, called a quadrant, where n is an integer. In this embodiment, the method is implemented using four quadrants 21, 22, 23, and 24 for each breast. The four quadrants are denoted QSI 21 (for Upper Inner Quadrant), QSE 22 (for Upper Outer Quadrant), QIE 23 (for Lower Outer Quadrant), and QII 24 (for Lower Inner Quadrant).

[0088] The embodiment presented, which includes subdividing or segmenting each breast into four zones / quadrants, is not limiting. The temperature data for each breast could be divided into only two or three quadrants. Advantageously, the temperature data for each breast could be divided into five or more quadrants.

[0089] The compartmentalization of temperature data into quadrants 21, 22, 23, and 24 allows, among other things, for the anomaly to be located in one of these quadrants when determining an anomaly. However, the calculation of the temperature deviation or the determination of the statistical signature is carried out using temperature data from each of the n quadrants.

[0090] Preferably, the temperature difference between the two breasts is calculated between each corresponding quadrant 21, 22, 23, 24. In other words, a temperature difference is calculated between quadrant 21 of the left breast and quadrant 21 of the right breast, a temperature difference is calculated between quadrant 22 of the right breast and quadrant 22 of the left breast, and so on.

[0091] With reference to Figure 4, and to the first, second, and third refinements, an example of a statistical signature of the temperature difference between the two breasts as a function of time is illustrated. The y-axis shows the temperature difference between the two breasts. The x-axis shows the average values ​​of the temperature difference between the two breasts for each of the four quadrants 21, 22, 23, and 24, across ten subsets.

[0092] We can observe that the temperature difference between the two breasts varies over time. The statistical signature of the temperature difference between the two breasts includes a variation in the temperature difference between the two breasts over time.

[0093] According to a fourth improvement, the temperature difference and / or statistical signature is determined by comparing the temperature data of one breast, chosen as the reference temperature data, with the temperature data of the other breast. The comparison can be performed by linear combination, for example by subtraction.

[0094] Thus, the process makes it possible to limit the amount of data required to implement the invention, in particular for the selection of the treatment model, while avoiding the need to use standard or physiological data or data previously measured on the subject.

[0095] According to a fifth improvement, the process includes a step of determining: - based on temperature data, one or more indicators representative of the quality of the measured data, and / or - based on the statistical signature or on the temperature difference and the selected temperature profile, one or more indicators representative of a trend in the temperature difference between the two breasts, and / or - based on the statistical signature or the temperature difference, the representative indicator(s) of the quality of the measured data and the selected temperature profile, one or more indicators of the reliability of the selected treatment model, and / or - from the indicator(s) representing the quality of the measured data and / or the indicator(s) of reliability of the selected processing model, an indicator of the reliability of the information representing the presence or absence of an anomaly, in particular the reliability of the information representing the presence or absence of an anomaly in a quadrant.

[0096] Each improvement can be implemented individually.

[0097] Each improvement can be combined with one, several, or all of the other improvements.

[0098] Advantageously, temperature data processing is implemented by a statistical or deep learning model.

[0099] A person skilled in the art will be able to choose and adapt the corresponding or most suitable statistical or deep learning model.

[0100] As a non-limiting example, the statistical or deep learning model could be a Long Short-Term Memory (LSTM) recurrent neural network. Learning begins with data sizing to make the data compatible with the LSTM layers. The recurrent neural network is then built with several LSTM layers. An LSTM layer consists of an LSTM cell, which is composed of several internal elements: input, output, and forget gates, as well as a memory cell to retain important features learned during iterations and to learn temporal sequences. A dense layer with sigmoid activation is added at the end for binary classification.

[0101] As a non-limiting example, regularization techniques such as dropout and L2 regularization (or ridge regression) are applied to prevent overfitting. This neural network captures temporal dependencies in temperature series and performs binary classification to detect the presence of breast abnormalities.

[0102] According to the invention, the method may not include a step of acquiring temperature data.

[0103] Temperature data can be stored data.

[0104] The stored data can originate, for example, from measurements taken on one or a group of individuals. Thus, the process can be implemented on a set of previously acquired stored data, independently of the process implementation.

[0105] The stored data can come, for example, from simulation(s). Thus, the process can be implemented on a set of stored data independently of the implementation of the process.

[0106] Advantageously, the process includes a step of training the statistical or deep learning model from a training database.

[0107] The training database includes input data consisting of a temperature dataset. Each temperature dataset includes temperature data from each of the two breasts of the same subject.

[0108] The training database also includes output data consisting of representative information on the presence or absence of an anomaly associated with each set of temperature data.

[0109] Learning from the statistical or deep learning model allows for the enrichment of temperature data processing models.

[0110] The method according to the invention does not provide a therapeutic diagnosis.

[0111] The method according to the invention aims to provide representative information on the presence or absence of an anomaly in breast tissues.

[0112] The method according to the invention is not intended to, nor does it allow for, providing a therapeutic diagnosis. The method according to the invention provides an indicator informing the individual of the relevance of consulting a practitioner, and if so, whether urgently (if an indication of the presence of a malignant anomaly is provided) or not (if an indication of the presence of a benign anomaly is provided).

[0113] With reference to FIGURES 1 and 2, a thermographic analysis device 1 is also proposed for the detection of anomalies in breast tissues, referred to as device 1.

[0114] Device 1 comprises two arrays 2 of temperature sensors 3 intended to be in contact with the skin of a respective breast. Each array 2 is arranged to cover each breast evenly.

[0115] According to the non-limiting embodiment, the two arrays 2 of temperature sensors 3 are symmetrical with respect to each other. Preferably, the two arrays 2 of temperature sensors 3 are symmetrical with respect to a median or sagittal plane of the individual and / or with respect to a transverse plane of the individual.

[0116] Device 1 further includes a temperature data acquisition unit 4 for temperature data over time, from temperature sensors 3, arranged to communicate with a processing unit 5. The acquisition unit 4 and the processing unit 5 are connected by wire or wirelessly.

[0117] By way of example, according to the non-limiting embodiment, the device comprises 96 temperature sensors 3.

[0118] The processing unit 5 is arranged and / or configured and / or programmed to implement the method of detecting anomalies in breast tissue by thermography according to the invention.

[0119] The device includes a one-piece flexible support 7.

[0120] The flexible support 7 includes the two networks 2 of temperature sensors 3.

[0121] The flexible support 7 includes a set of 8 separate tabs.

[0122] The 8 tabs have actinomorphic geometry.

[0123] The 8 tabs include the 2 network of 3 temperature sensors.

[0124] The flexible support 7 includes a central portion 9 intended to be positioned on the mammary papilla of a breast.

[0125] According to the embodiment, at least some of the temperature sensors 5 are arranged in a diamond and / or triangle and / or circle on the flexible support 7.

[0126] Preferably, the 5 temperature sensors are distributed in concentric circles.

[0127] According to the embodiment, the temperature sensors 5 are arranged in a rhombus. Each of the four vertices of a rhombus corresponds to a temperature sensor 5. For each rhombus considered and for each sensor 5 of each tab 8, three sensors 5 are arranged on a given tab 8 and one sensor 5 is arranged on a tab 8 adjacent to the given tab 8.

[0128] Advantageously, each tab 8 is separated from the two adjacent tabs 8 by two slots located on either side of each tab 8.

[0129] Preferably, each tab 8 and / or each slot, or the border or edge or side or perimeter separating two tabs 8 includes curves and / or curvature and / or is not straight or rectilinear or planar.

[0130] Preferably, the temperature sensors 3, preferably even more preferably the temperature sensors 3 of the same tab 8, are not aligned, in particular radially.

[0131] Thus, compared to a radial and / or axial or rectilinear distribution of the sensors and / or to a separation and / or to slits only radial and / or axial or rectilinear between two tabs 8, the tiling of the temperature sensors 3 and / or the arrangement of the tabs 8 makes it possible to limit the noise of temperature measurement and / or missing or aberrant temperature data and / or false positives.

[0132] It can also be seen in FIGURES 1 and 2 that device 1, in particular the arrangement of the temperature sensors 3, and advantageously the presence and / or arrangement of the tabs 8, makes it possible to obtain data of temperature originating from areas located between the breasts and / or from the armpits.

[0133] With reference to FIGURE 3, a bra 10 is also proposed for the detection of abnormalities in breast tissue by thermography. The bra comprises the device 1 according to the invention.

[0134] The bra 10 also includes two cups 6. Inside each cup 6 is arranged one of the respective temperature sensor arrays 2 of device 1.

[0135] Each of the 6 cups is arranged to thermally insulate the temperature sensors 3 from the environment outside the bra 10.

[0136] Figure 6 illustrates a graph representing the average breast temperature per breast as a function of breast density and breast condition.

[0137] The term "breast condition" refers to the state, condition, or status of the tissue: "physiological" or "breast" as opposed to "abnormal," which may be "pathological." The term "ACR" stands for American College of Radiology breast density classification.

[0138] It has been observed that dense breasts exhibit higher overall temperatures than less dense breasts. Types A through D each correspond to distinct phenotypic data. This phenotypic data, and in particular, but not exclusively, breast density, influences the temperature profile and is therefore associated with different temperature profiles.

[0139] The results presented in FIGURE 6 are from a clinical study conducted on a cohort of 70 individuals, consisting of 50 patients with the condition and 20 controls. Device 1 and bra 10 according to the invention made it possible to determine the presence of abnormalities with an effective prediction greater than 90% and with over 90% sensitivity. These preliminary results demonstrate the effectiveness and reliability of the invention for detecting abnormalities in breast tissue.

[0140] Of course, the invention is not limited to the examples just described and many modifications can be made to these examples without departing from the scope of the invention.

[0141] In particular, all the variants and embodiments described can be combined with each other if there are no technical obstacles to this combination.

Claims

- 24 - DEMANDS 1. A skin contact thermography analysis device (1) for the detection of abnormalities in breast tissue, referred to as device (1); said device comprising: - two arrays (2) of temperature sensors (3) intended to be in contact with the skin of a respective breast and arranged to cover each breast homogeneously, - a temperature data acquisition unit (4) for temperature data over time, from temperature sensors, arranged to communicate with a processing unit (5), said processing unit being arranged to: • Calculate, based on temperature data, the temperature difference between the two breasts as a function of time. • select, from the temperature data, or from the temperature data and phenotypic data, a temperature profile corresponding to the temperature data, • Select, based on the temperature difference over time and the selected temperature profile, an appropriate data processing model, • determine information representative of the presence or absence of an anomaly by processing temperature data with the selected processing model.

2. Device (1) according to the preceding claim, wherein the processing unit is arranged to, from temperature data: - to determine a statistical signature of the temperature difference between the two breasts, - select, from the statistical signature of the determined temperature deviation and the selected temperature profile, the appropriate data processing model.

3. Device (1) according to claim 1 or 2, wherein each array (2) of temperature sensors (3) comprises a one-piece flexible support (7) on which the temperature sensors are arranged, the flexible support comprising a set of distinct tabs (8) of actinomorphic geometry and comprising a central portion (9) intended to be positioned on the mammary papilla of a breast.

4. Device (1) according to the preceding claim, wherein at least a portion of the temperature sensors are distributed in a rhombus on the flexible support (7); each of the four vertices of a rhombus corresponding to a temperature sensor, and, for each rhombus considered, three temperature sensors are arranged on a given tab and one temperature sensor is arranged on a tab adjacent to the given tab.

5. Bra (10) for the detection of abnormalities in breast tissue by thermography comprising: - the device (1) according to any one of claims 3 to 4, - two cups (6) inside which is arranged a network (2) of respective temperature sensors (3), each of the two networks of temperature sensors being intended to be in contact with the skin of a breast and being arranged to cover a breast homogeneously; each of the cups is arranged to thermally insulate the temperature sensors from the external environment.

6. A method for detecting abnormalities in breast tissue by thermography, referred to as the method, comprising the steps of: - to calculate, using a processing unit (5), a temperature difference between the two breasts as a function of time, based on temperature data from each breast, - to select, using the processing unit, from the temperature data, or from the temperature data and phenotypic data, a temperature profile corresponding to the temperature data, - select, using the processing unit, from the temperature deviation as a function of time and the selected temperature profile, an appropriate temperature data processing model, - determine information representative of the presence or absence of an anomaly by processing, using the processing unit, temperature data with the selected processing model.

7. A method according to the preceding claim, comprising a step of acquiring, by means of an acquisition unit (4), temperature data during the time, measured by means of two arrays (2) of temperature sensors (3) in contact with the skin of a respective breast and arranged to form a homogeneous tiling of each of the breasts.

8. Method according to claim 6 or 7, further comprising a step of determining, from temperature data, a statistical signature of the temperature difference between the two breasts; the selection of the temperature data processing model is implemented from the statistical signature of the temperature difference determined and the user profile.

9. A method according to any one of claims 6 to 8, wherein the data processing model selection step is implemented from a variation of the temperature difference between the two breasts over time and the selected temperature profile.

10. A method according to any one of claims 8 or 9, wherein the temperature difference and / or statistical signature is determined by comparing the temperature data of one of the breasts, chosen as reference temperature data, with the temperature data of the other breast.

11. A method according to any one of claims 8 to 10, wherein: - the temperature data for each breast comprises n sets of temperature data, each corresponding to a distinct area of ​​the breast, called a quadrant, where n is an integer, and - the calculation of the temperature difference and / or the selection of the data processing model and / or the determination of the statistical signature is implemented from the set of temperature data from the n quadrants.

12. A method according to any one of claims 8 to 11, comprising a step of determining, from the statistical signature, one or more indicators representative of a trend in the temperature difference between the two breasts; the step of selecting the temperature data processing model is implemented from the determined indicator(s).

13. A method according to any one of claims 6 to 12, wherein the temperature data processing is implemented by a statistical or deep learning model. - 27 - 14. A method according to the preceding claim, comprising a step of training the selected model from a training database, said training database comprising: - input data consisting of a set of temperature data, each set including temperature data from each of the two breasts of the same subject, - output data consisting of information representing the presence or absence of an anomaly associated with each set of temperature data.

15. A method according to the preceding claim taken in combination with any one of claims 8 to 12, wherein the learning step is implemented, independently, for each processing model, from the training data corresponding to the temperature data.

16. Data processing device comprising means arranged and / or programmed and / or configured to implement the method according to any one of claims 1 to 15.

17. Computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 15.

18. Computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 15.

Images