Breast tumour detection bra with advanced imaging

Abstract

The present invention describes a detection and image reconstruction system based on microwave radar technology, designed mainly to detect breast abnormalities. The system comprises a set of emitting and receiving transducers. The signals reflected by the tissue are processed using algorithms such as delay and sum (DAS) and delay, multiply and sum (DMS), applied iteratively, which significantly improves image resolution, reconstruction and quality.

Description

[0001] MABIS breast tumor detection bra with AI and advanced imaging.

[0002] TECHNICAL FIELD

[0003] The present invention falls within the technical field of early detection of breast tumors using portable medical devices, specifically bras equipped with ultra-wideband (UWB) radar technology. These devices employ non-ionizing radiation and advanced signal processing algorithms to identify abnormalities in breast tissue. Additionally, the invention integrates artificial intelligence tools and advanced imaging techniques, offering a non-invasive, safe, and efficient approach to early diagnosis, with potential for telemedicine applications.

[0004] BACKGROUND OF THE INVENTION

[0005] The state of the art includes the article “A Time-Domain Microwave System for Breast Cancer Detection Using a Flexible Circuit Board” by Santorelli, A., et al. (DOI: 10.1,109 / T| ,2015.2440565), which describes a breast cancer detection system using a flexible circuit board and microstrip technology. The system comprises a switching network designed using a 4x1 and 2x1 cascade. A switching core selects one of the quadrants, allowing each of the 16 antennas to act as either a transmitter or receiver. These antennas were selected for their low insertion loss (0.45 dB), bandwidth extending up to 6 GHz, and good linearity at the required signal power levels (P1dB = +30 dBm).An Agilent 8722ES or S5085 VNA can be used to collect sensor data. This system also includes antennas, two microcontrollers, a pulse generator, and software that captures 240 signals, processes differential signals using the delay-multiply-and-sum algorithm, and displays the data on a PC to send control bits and save the recorded TD data. However, this information does not mention LVTTL (low-voltage transistor-transistor logic) control for the switches, switches with a frequency of up to 8 GHz, an Arduino microcontroller, or a portable battery.

[0006] We also offer the "Smart Bra with Optical Sensors to Detect Abnormal Breast Tissue" system, a patented system designed to detect breast tissue abnormalities using optical light emitters and receivers (patent publication code US11304456B2). This system uses near-infrared light to penetrate breast tissue and analyze changes in light intensity or spectrum, allowing for the identification of potential signs of abnormal tissue. Its primary focus is on the early detection of conditions such as breast cancer, through a portable and non-invasive solution integrated into an everyday garment.

[0007] Technical Limitations

[0008] • Limited resolution and accuracy:

[0009] Changes in light intensity or spectrum can be affected by multiple uncontrolled variables, such as breast tissue density, blood flow, or even user movement, which can lead to false positives or negatives.

[0010] • Dependence on a uniform distribution of sensors:

[0011] To achieve accurate results, the sensors must be perfectly positioned and adjusted to the body, which can vary between users due to anatomical differences and breast sizes. This could affect the device's effectiveness in different situations.

[0012] • Data processing:

[0013] Although optical data is collected, its interpretation requires advanced algorithms and, in some cases, external support, which could lead to delays in real-time diagnosis if the connection or analysis software fails.

[0014] • Lack of integration with complementary biomarkers: The system is based solely on optical properties, which excludes other diagnostic methods, such as thermal measurements or chemical biomarkers, that could improve the accuracy of the analysis.

[0015] On the other hand, we have the press release: “MABIS: Bra designed for the early detection of breast tumors” which can be found in the following brasfer-

[0016] The MABIS (Microwave Analyzer for Breast Imaging Scanning) system, developed by a team of researchers at the Catholic University of San Pablo (Arequipa, Peru) led by researcher Ebert San Román Castillo, was initially presented as a bra with wearable technology and UWB (Ultra-Wide Band) radar for the early detection of breast cancer. This portable device uses microwave imaging to identify and quantify breast abnormalities, offering high sensitivity and specificity in diagnosis.

[0017] In its initial version, the system integrated 16 flexible antennas to capture electromagnetic signals and reconstruct 2D images of breast tissue. Furthermore, it stood out as an accessible and low-cost tool with the potential to decentralize breast cancer diagnosis, especially in rural and marginalized urban areas.

[0018] While the released version of the MABIS system represented a significant advance, it had areas for improvement in terms of applied technology and functionality:

[0019] 1. Image processing: It was based solely on standard algorithms such as DAS (Delay and Sum), with limitations in the quality and resolution of the generated images. Now, the algorithm is strengthened by implementing iterative code iTDAS and ITDMAS. 2. Bra adaptability: Although designed as a wearable, it was not optimized to fit diverse patient body shapes.

[0020] 3. Clinical report generation: It did not include automated tools to interpret and report diagnostic results, which limited its use in environments with non-specialized medical staff, and now AI is being implemented for clinical report generation.

[0021] DESCRIPTION OF THE INVENTION

[0022] The present invention relates to a microwave analysis system specifically designed for breast imaging and detection, based on a bra that incorporates ultra-wideband sensors. These sensors consist of omnidirectional antennas strategically positioned on the bra cups, optimized for neoplasm detection using near-field radar technology.

[0023] The system integrates high-precision electronic devices, including chip-type switches, a microcontroller, a portable battery, a vector network analyzer (VNA), and the aforementioned antennas. The switches digitally control the sensor network, which connects to the VNA to acquire data in the frequency domain, totaling 512 signals. This design is fully automated thanks to the Arduino Mega microcontroller, which enables communication between the VNA and the switching board, optimizing signal reception times.

[0024] The optimized sensor geometry is shown in Fig. 1. The sensors are modified monopoles with an adaptive superlayer specifically designed to fit the breast environment. These sensors feature linear polarization and are fabricated on a 1.5 mm thick FR4 substrate, with dimensions of 20 mm wide by 27 mm high.

[0025] They are arranged in two levels: eight sensors at the top and eight at the bottom, mounted on a silicone bra that acts as an impedance matching device. This bra, which acts as the aforementioned superlayer, minimizes reflections at the interface between the antennas and the biological tissue, as illustrated in Fig. 2.

[0026] The superlayer not only improves coupling between the antennas and the tissue, but also significantly reduces reflections at the air-tissue interface, thus optimizing measurement quality. This design increases the dynamic range of the received signals, facilitating greater tissue penetration and preserving the fidelity of the collected data for subsequent analysis. This configuration ensures more accurate and reliable measurements, promoting the early detection of breast abnormalities.

[0027] The antenna is designed to operate efficiently in a frequency range of 1.40 to 10 GHz, achieving optimal adaptation in a complex medium, as detailed in Fig. 3. This optimization avoids significant signal losses within breast tissue, eliminating the need for complex surface reflection removal techniques. This creates a technological advantage by providing an ultra-wide bandwidth that improves spatial resolution.

[0028] The hardware design includes a switching matrix implemented on a PCB designed with KiCad. This matrix uses ADRF5045 and ADRF5019 switches, selected for their low insertion loss, high isolation, 50-ohm internal termination, and ability to operate between 100 MHz and 13 GHz.

[0029] 1. ADRF5045 Switch: o Common port with four connected outputs. o Dual power supply of +3.3V and -3.3V. o LVTTL logic control.

[0030] 2. ADRF5019 Switch: o Common port with two connected outputs. o Dual power supply of +3.3V and -2.5V. o LVTTL logic control.

[0031] The sensor array is arranged in a 2x16 port configuration, divided into two groups, with 16 ports for each side of the bra: right and left. These are connected via SMA connectors, ensuring reliable and stable transmission of the captured signals. The system allows for both monostatic and bistatic configurations, expanding its flexibility and adaptability for various clinical applications.

[0032] The implementation of dual-power switches (+3.3V, -3.3V, and -2.5V) provides a significant advantage, as it allows the invention to be fully portable and compatible with external batteries. Furthermore, the use of 4x1 and 2x1 type switches instead of larger options, such as the 8x1 switch, reduces signal loss due to the smaller number of ports involved, thus improving system efficiency and measurement quality.

[0033] Furthermore, the bra's design is intended to adapt to any body shape. It features two crossed straps at the back and an adjustable lumbar strap secured with plastic snaps, as shown in Figure 4. This adjustable feature allows the bra to adequately cover the entire breast area, regardless of size, whether very small or very large. This minimizes detection errors and ensures that the sensors maintain optimal contact with the tissue.

[0034] Thanks to its adaptable design, the system allows for the precise detection of breast tissue abnormalities. It can identify the position and size of these abnormalities using advanced signal processing algorithms. This capability is complemented by the generation of high-resolution microwave-based images, providing detailed and critical information for early and effective diagnosis.

[0035] The system is simple and efficient to operate. Its design allows for a complete measurement of each breast in a quick process, followed by automated analysis that guarantees reliable results in a short time. Thanks to its compact design (30 cm x 30 cm x 10 cm) and integrated battery, the device is highly portable, making it easy to use in various settings, including hospitals, ambulances, and rural health centers.

[0036] Now, the modified confocal microwave technique using the inverse Chirp Z transform offers several significant advantages in the system compared to the systems mentioned above:

[0037] • Higher resolution in specific frequency bands.

[0038] • Reduction of noise in the processed signal.

[0039] • Computational optimization, by allowing more efficient processing than the traditional inverse Fourier transform and lower computational load.

[0040] The inverse Chirp Z-transform allows for efficient data conversion from the frequency domain to the time domain, facilitating more detailed signal analysis. Unlike other transforms, this technique improves the accuracy of temporal peak localization due to its ability to dynamically adjust the frequency range and spectral resolution. This contributes to better detection of breast tissue abnormalities, even in cases where they are small or in their early stages.

[0041] In image processing, after identifying the peaks using cross-correlation, the DAS (Delay And Sum) and DMAS (Delay Multiply And Sum) algorithms are implemented to reconstruct the images on a pixel grid.

[0042] These images undergo advanced error metrics, such as Signal-to-Noise Ratio (SNR), an innovative approach not implemented in traditional microwave imaging systems. Incorporating this metric provides multiple competitive advantages that optimize diagnosis and improve system accuracy:

[0043] 1. Objective validation of system quality:

[0044] Signal-to-noise ratio (SNR) allows for the quantitative evaluation of captured signal quality, clearly differentiating between relevant signals (originating from potential tissue abnormalities) and noise or interference present in the images. This analysis ensures accurate and consistent detections, minimizing errors associated with subjectivity in visual interpretation.

[0045] 2. More reliable diagnosis:

[0046] By ensuring a high signal-to-noise ratio, this metric optimizes the system's ability to identify relevant features, such as high-energy areas associated with potential anomalies. This reduces the possibility of false positives or negatives, increasing confidence in the results obtained.

[0047] 3. Optimization of automated analysis:

[0048] Implementing signal-to-noise ratio (SNR) in image processing allows for prioritizing regions of interest (ROIs) in images, improving anomaly detection while eliminating irrelevant signals (clutter) and background noise. This automated approach accelerates diagnosis, making it more efficient and accessible for medical personnel, even those without specialized training in imaging technologies.

[0049] 4. Competitive advantage over traditional systems:

[0050] Conventional systems typically rely solely on visual inspections performed by trained technicians, which introduces variability into the results. By using SNR, MABIS overcomes these limitations by providing a standardized, objective, and reproducible assessment that is independent of operator experience.

[0051] 5. Improved image quality for clinical decisions:

[0052] By applying SNR, a clearer and more accurate representation of suspicious areas in the tissue is achieved, facilitating informed clinical decision-making. This not only improves the detection of abnormalities but also optimizes treatment planning by providing higher-quality images.

[0053] Furthermore, the use of the modified Iterative Delay And Sum (ItDAS) algorithm provides a distinct advantage in image quality. This algorithm performs iterative processing that dynamically adjusts signal focus, reducing noise and improving spatial resolution. The competitive advantages of the ItDAS algorithm are presented.

[0054] 1. Innovation in iterative backprojection:

[0055] Unlike traditional methods, the use of DAS and DMAS as iterative operators progressively improves image reconstruction, offering greater resolution and contrast.

[0056] 2. Optimizing accuracy:

[0057] The iterative structure allows for the reduction of errors and artifacts, ensuring greater fidelity in the representation of the dielectric properties of the target.

[0058] 3. Integrated standardization:

[0059] The use of the matrix of ones in reprojection and backprojection ensures a balanced distribution of the data, facilitating the convergence of the algorithm and improving stability.

[0060] 4. Dual application (DAS and DMAS):

[0061] The flexibility of using DAS or DMAS allows the algorithm to be adapted to different requirements, prioritizing computational efficiency or resolution depending on the application.

[0062] 5. Applicability in modern imaging systems:

[0063] The ItDAS algorithm is presented as a versatile tool for advanced radar and tomography systems, achieving more detailed and reliable image reconstruction compared to conventional approaches. Once the images are reconstructed, the system uses artificial intelligence (AI) exclusively to generate detailed clinical reports. The AI ​​has been trained to analyze the data obtained, organize relevant information, and structure a report indicating whether anomalies have been identified, as well as their location and size. This clinical report is not intended to make diagnoses but rather serves as a complementary tool that streamlines and standardizes the documentation of the evaluation process. The automated generation of reports minimizes manual intervention, reducing human error and allowing physicians to access clear and organized information to support their work.The reports are automatically uploaded to a web platform, which improves data accessibility for both doctors and patients, ensuring an efficient and secure flow of information.

[0064] BRIEF DESCRIPTION OF THE FIGURES

[0065] (List the attached figures, including a description of no more than three lines each. Do NOT include the drawings in this section of the document.)

[0066] Figure 1: Design of modified linear polarization monopole sensor for ultra-wideband

[0067] Figure 2: Electronic board design and implemented components

[0068] Figure 3: MABIS system implemented with the Bra, signal generation and reception module

[0069] Figure 4: MABIS bra design

[0070] Figure 5: Iterative algorithm process

[0071] Figure 6: Process of the inverse chirp z transform, cross correlation, image reconstruction and AI PREFERRED EMBODIMENTS OF THE INVENTION

[0072] (In the case of a device or apparatus, provide a physical description of the parts that make up the product using reference symbols to the drawings. Example: “The invention consists of a chair (10), which has on its front legs (11) a plurality of connecting elements (12) ...”)

[0073] The MABIS (Microwave Analyzer for Breast Imaging Scanning) system is an innovative technology developed for the non-invasive detection of breast abnormalities, combining portability, accuracy, and accessibility (30). The invention integrates advanced hardware and software components, which work synergistically to capture, process, and analyze microwave signals reflected by breast tissue. The structure and operation of the main elements of the system are described in technical and physical detail below.

[0074] Emission and Reception Sensors

[0075] In Figure 1, the MABIS system uses ultra-wideband sensors (10) to transmit microwave waves at a frequency of 1.4 to 10 GHz (105) towards the mammary region and capture the reflected signals. These sensors are built on a 1.5 mm thick FR4 substrate (109), with a total height of 27 mm (101) and a width of 19 mm (102), composed of a stepped structure with sections of 7 mm, 1 mm, 2 mm, 5 mm, and 12 mm.

[0076] The sensor geometry (10) has been designed to optimize current distribution, improving signal radiation and reception by introducing modifications that control dispersion and reduce losses.

[0077] The amplitude (in dB) versus time (in nanoseconds) graph shows the signal behavior. The main peak near -15 dB at 0 ns (103) corresponds to the initial reflection of the signal off the antenna surface, while the secondary oscillations between 2 ns and 10 ns (104) indicate reflections within the stepped sections. The deep minimum around 7 ns (108) suggests attenuation associated with interference or absorption of the signal in the modified structure. This design improves the sensor's performance in terms of radiation and signal control, achieving a more efficient time-domain response.

[0078] The sensors are arranged on two levels (106): eight sensors at the top and eight at the bottom, fitted onto a silicone bra (107) that acts as an impedance adapter (mentioned super layer) to minimize reflections between the antennas and the biological tissue:

[0079] • The sensor (10) is designed to optimize the transmission of energy to the tissue, minimizing losses and ensuring high sensitivity in capturing reflected signals.

[0080] • Adaptive dielectric coupling (107): The dielectric superlayer adapts to the patient's specific morphology, providing effective coupling between the sensors and the breast tissue. This ensures uniform energy transfer and significantly reduces external signal interference, improving the quality of the collected data.

[0081] • These sensors are connected to the signal generation (304) and capture module via pre-calibrated coaxial cables, ensuring data transfer with electromagnetic integrity.

[0082] System design and operation

[0083] Figure 2 shows a four-layer electronic board (20) designed to expand two input ports (201), connected to a Vector Network Analyzer (VNA), into 2x16 output ports (202), designated Right Sensor (RS) and Left Sensor (LS). This is achieved by integrating two ADRF4045 switches (203) and one ADRF5016 switch (204). A four-layer configuration was chosen because it is sufficient for this design, assigning each layer to specific functions.

[0084] 1. First layer (205): Exclusively for radio frequency signals. 2. Second layer (206): Ground plane to ensure stability and isolation.

[0085] 3. Third layer (207): Power supply distribution.

[0086] 4. Fourth layer (208): Switch control.

[0087] The layers are separated by Rogers4003 dielectric materials (209), selected to maximize switch performance. The design includes coplanar transmission lines and vias (211) for impedance matching to 50 ohms, ensuring signal integrity. This approach guarantees excellent isolation and minimizes signal loss compared to traditional module-based systems.

[0088] The board is preferably controlled by a microcontroller (210). This sends voltage pulses to enable the switches and select the specific pins depending on which port is to be activated. The configuration allows the system to operate in two modes:

[0089] • Monostatic: Transmission and reception of the signal through a single port.

[0090] • Bistatic: Transmission through one port and reception through another.

[0091] The microcontroller is connected to the electronic board via cables for automation. The output ports are female SMA connectors (202), facilitating connection to external sensors. The RF board is powered by a portable battery (212), and two batteries can be used for added autonomy and portability.

[0092] Signal Generation and Capture Module (30)

[0093] Figure 3 shows the module for generating and recording electromagnetic signals for breast tissue evaluation. Its main features include: 1. Generation of electromagnetic signals:

[0094] The signals are generated by a low-cost VNA (301), controlled by a microcontroller, covering a wide frequency range for optimal spatial resolution.

[0095] 2. Switching system (302):

[0096] It allows expanding the capabilities of the VNA up to 2x16 ports, enabling the connection of up to 32 sensors (303), guaranteeing the accurate acquisition of reflected signals.

[0097] 3. Optimized communication:

[0098] The programmed microcontroller (210) manages communication between the VNA and the switching system, integrating both components into a single system that optimizes both operating time and data processing.

[0099] MABIS bra design (40)

[0100] In Figure 4, the designed bra features an advanced system of adjustable regulators (401), both at the front and back, allowing for precise adaptation to various patient body shapes, covering sizes from XS to XL. The straps have an ergonomic design that allows for adjustment of their length and position (402), ensuring optimal contact between the device and the breast tissue. It also incorporates an adjustable waistband (403) that facilitates adjustment to the patient's waist and torso, guaranteeing proper pressure distribution without compromising user comfort.

[0101] This innovative design ensures that the sensors integrated into the bra maintain constant and secure contact with breast tissue through the "silicon bra" (107), optimizing signal acquisition for accurate diagnoses. Furthermore, its structure allows for a uniform fit that eliminates gaps or irregularities in the contact surface, improving measurement quality and minimizing any interference. This solution is ideal for medical applications requiring efficient monitoring adaptable to the diverse anatomy of patients.

[0102] Image Processing and Formation (60)

[0103] In Figure 6, the MABIS processing system (60) constitutes the technical core of the invention, designed to perform advanced analyses and generate high-precision images of breast tissue. This system is based on confocal microwave radar techniques, adapted to maximize their efficiency and offer competitive advantages.

[0104] • Modification of confocal microwave imaging techniques: The inverse Chirp Z transform (601) was implemented to convert data from the frequency domain to the time domain, optimizing signal processing. The main advantages of this transform are: o Higher resolution in specific frequency bands (602). o Reduction of noise in the processed signal. o Computational optimization, by allowing more efficient processing than the traditional inverse Fourier transform.

[0105] • Cross-correlation:

[0106] Once the signals are in the time domain, cross-correlation (603) is applied to compare the captured signals with reference models. This allows: Accurate identification of tissue anomalies by detecting significant discrepancies between the measured and expected signals; and enhancement of the most representative pulses (604), eliminating interference (605) and improving the system's sensitivity.

[0107] The processed data are subjected to specialized reconstruction algorithms (606) such as Delay and Sum (DAS) and Delay Multiply and Sum (DMAS), which produce two-dimensional images (607) based on the dielectric properties of the tissue.

[0108] • DAS: Robust and efficient algorithm for summing aligned signals (613).

[0109] • DMAS: Better resolution and enhancement of details, allowing a clearer differentiation between healthy tissue and abnormalities (614).

[0110] In this process, the pixels with the highest energy are highlighted (608), facilitating the identification of suspicious areas in breast tissue.

[0111] Error metrics for quality assessment:

[0112] Innovative techniques have been incorporated to evaluate the accuracy and reliability of the generated images, including:

[0113] The Signal-to-Noise Ratio (SNR) metric (609) is applied directly to images generated by MABIS, reducing reliance on specialized visual inspections. This makes the collected information accessible to general medical staff, nurses, and patients without requiring advanced radiology training.

[0114] Key components in the images:

[0115] 1. Region of Interest (ROI): Area with higher energy, representing possible tumors or abnormalities (610).

[0116] 2. Clutter: Unwanted interference and signals that affect image clarity (611).

[0117] 3. Background: Area free of relevant features, where residual noise is considered part of the analysis (612).

[0118] This ensures reliable differentiation between healthy and malignant tissue, providing an additional level of validation that minimizes subjectivity in diagnosis. Analysis of these areas facilitates a more accessible and reliable diagnosis, overcoming the limitations of traditional systems. Iterative Algorithm

[0119] In Figure 5, the ItDAS Iterative algorithm is based on equation (1), derived from an iterative scheme inspired by the Maximum Expectation (ME) algorithm, which is rooted in the Poisson statistics used in Positron Emission Tomography (PET) imaging. The goal of this approach is to integrate the advantages of radar and tomography reconstruction techniques by applying the DAS (Delay and Sum) and DMAS (Delay Multiply and Sum) beamformers as backprojection operators.

[0120] A relationship is sought between a radar-based reconstruction technique and a tomography-based one, because the iterative structure will apply DAS and DMAS as backprojection operators. The basic structure of the maximum expectation algorithms is as follows:

[0121] In this method, the new estimate (50) will be the product of the current or last estimate and the backprojection operator. First, the backprojection of the division between the initial measurements in the time domain will be performed using the forward projections of the last or current estimate (503). In each update, B and F are involved for all projection channels. Based on this iterative approach, an algorithm is proposed to achieve improved reconstruction, using equation 2. where Bp is the new estimate, that is, the pixel intensities; Bp and Fp are the backprojection and reprojection operators (504), respectively; I is the data obtained in the processing stage; U is a matrix of ones to which an operator is applied; the reprojection and backprojection of the matrix of ones U serves as normalization. It is by using the DAS and DMAS beamformers as backprojection operators that we have their iterative forms: Iterative DAS and Iterative DMAS. Figure 8 shows the diagram of the functionality of DAS and DMAS in their iterative form, because the DAS or DMAS beamformer can be used as the backprojection operator Bp.Then, in the Figure, we have the initial estimate of the target (501), which is back-projected into the time domain (502). Next, the relationship between the processed data I (already normalized by multiplying with FpU) and the reprojection of the last estimate made, which in the first iteration of the algorithm is the initial estimate of the target (503), is determined. This relationship will be back-projected and could be displayed on the pixel grid, that is, it is in the target space (504). But to obtain the new, improved estimated image (new estimate), it is necessary to multiply the previously estimated image by the back-projection of the new normalized estimate (505).

[0122] Then the detailed process of the algorithm:

[0123] 1. Initial estimate:

[0124] It starts with an initial estimate of the target (501), which is back-projected in the time domain using the DAS or DMAS operator.

[0125] 2. Relationship between data and reprojection:

[0126] The relationship between the normalized processed data (I) and the reprojection of the last estimate is calculated. In the first iteration, this last estimate corresponds to the initial estimate (502).

[0127] 3. Rear projection and updating:

[0128] The resulting relationship is back-projected into the target space, being displayed on the pixel grid. The new, improved estimate is obtained by multiplying the previous estimate by the normalized back-projection (615).

[0129] 4. Iteration and convergence:

[0130] The process is repeated iteratively (506), refining the reconstruction with each step until the estimate converges, generating images of greater accuracy and quality.

[0131] User Interface and Display

[0132] User interaction with the system is carried out through an intuitive and easy-to-use graphical interface (616), accessible from mobile devices, tablets or computers, allowing for a smooth and efficient experience.

[0133] • Real-time visualization (617): The interface facilitates the visualization of reconstructed images and diagnostic metrics in real time, with clear and user-friendly representations. This feature simplifies the interpretation of results, even for non-specialized personnel, democratizing access to diagnosis.

[0134] • Automated generation of clinical reports: The system integrates artificial intelligence (AI) to automatically generate detailed clinical reports (618), optimizing the medical documentation process. This functionality: o Streamlines the creation of accurate and structured reports. o Minimizes the administrative burden for physicians and technicians, allowing them to focus on clinical decision-making. o Facilitates the understanding of results by summarizing key findings and relevant metrics.

[0135] • Cloud synchronization: Collected data is automatically synchronized with a cloud platform, enabling: o Secure and scalable storage of results. o Advanced analysis using tools based on specialized algorithms. o Remote access to data by physicians and specialists, enabling interdisciplinary collaboration and remote monitoring.

[0136] • Robust and secure connectivity: The system uses Wi-Fi to ensure a fast, stable, and secure connection between the MABIS device and the display terminals.

[0137] • Offline operation: In the absence of connectivity, the system has a local copy of the algorithm on the central computer (PC), which ensures its uninterrupted operation and the availability of the results without depending on the network.

[0138] Portable and Modular Design

[0139] The system is housed in a compact and robust casing (30), manufactured by 3D printing (302) to ensure the protection of the electronic components and facilitate its transport and use in clinical, rural or hard-to-reach environments.

[0140] • Minimization of electromagnetic interference: The housing is made of specialized dielectric materials that minimize external electromagnetic interference. This ensures optimal signal capture performance, improving the accuracy of the collected data.

[0141] • Modular and scalable design: The system's modular architecture offers multiple competitive advantages: o Easy replacement: Individual components can be quickly replaced in case of failure, minimizing device downtime. o Efficient upgrades: The modular structure allows for the integration of new components or technologies without requiring a complete system redesign, optimizing scalability and keeping the device up-to-date. o Simplified maintenance: The design facilitates access to internal components, reducing costs and time associated with preventive and corrective maintenance tasks.

[0142] Therefore, to recap the MABIS system, the following key points stand out:

[0143] 1. Emission and Reception Sensors (10):

[0144] 2x16 configuration with 32 sensors optimized to emit and receive electromagnetic signals, achieving precise coverage of breast tissue.

[0145] 2. System Design and Operation (20):

[0146] 4-layer electronic board with lines adapted to 50 ohms, guaranteeing stability, modularity and easy maintenance of the system.

[0147] 3. Image Processing and Formation (60):

[0148] Use of advanced techniques such as the Inverse Chirp Z Transform, cross-correlation and iterative DAS / DMAS algorithms to reconstruct two-dimensional images with high resolution and enhancement of suspicious areas.

[0149] 4. Error Metrics (609) and Iterative Algorithm (50):

[0150] Quality assessment using SNR and MSE metrics, combined with the ItDAS algorithm, which iteratively improves the accuracy and consistency of the images.

[0151] 5. Interface and Modular Design (30):

[0152] Intuitive interface with AI-powered automated clinical report generation, cloud synchronization, and a portable, modular design that facilitates its use in clinical and rural environments.

Claims

CLAIMS 1. A MABIS breast tumor detection bra with AI and advanced imaging, of the type comprising: - A plurality of emission (10a) and reception (10b) sensors, arranged on two levels (106) within the bra (303) that transmit and receive microwave signals in a plurality of positions; and - A signal generation and capture module (30) that includes a Vector Network Analyzer (301) and calibrated coaxial cables (304), guaranteeing electromagnetic integrity in data transfer; and - An image processing system (60), which applies iterative DAS and DMAS algorithms, together with the SNR metric (609), to guarantee the quality of the generated images; and - An adjustable and ergonomic design, with adjustable straps (402) and an adaptable waistband (403), ensuring optimal contact with breast tissue in different sizes; and - A graphical user interface (616) accessible from mobile devices, tablets or computers, which integrates AI (616) for the automatic generation of clinical reports (618) and allows synchronization in the cloud.

2. A MABIS breast tumor detection bra, according to claim 1, characterized in that the sensors (10a and 10b) comprise a detection range of 1.4 GHz to 10 GHz and are designed to maximize sensitivity by means of an adaptive dielectric material (107).

3. A MABIS breast tumor detection bra, according to claim 1 and 2, characterized in that the plurality of the emission and reception sensors (10a and 10b) comprises a quantity of 2 A n sensors, where n is an even number and evenly distributed on the bra (303), allowing an adaptive expansion or reduction depending on the application and coverage requirements.

4. A MABIS breast tumor detection bra, according to claim 1, characterized in that it can generate three-dimensional (3D) images of breast tissue by combining iterative DAS and DMAS reconstruction algorithms (615), applied to improve spatial resolution and contrast.

5. A MABIS breast tumor detection bra, according to claim 1, 2 and 3, characterized in that it employs thresholding techniques in image processing to improve the detection of abnormalities (609), reducing noise and highlighting regions of interest (ROI, 610).

6. A MABIS breast tumor detection bra, according to claim 1, characterized in that the signal generation and capture module (30) can operate with a Vector Network Analyzer (301), a signal generator or an Analog-to-Digital Converter (ADC), ensuring flexibility and adaptability in implementation.

7. A MABIS breast tumor detection bra, according to the preceding claims, characterized in that it allows the detection of breast abnormalities, including benign and cancerous tumors, as illustrated in the image evaluation by means of the metrics system (609-612).

8. A MABIS breast tumor detection bra, according to the preceding claims, characterized in that it can be extended to other technological areas (20 and 30), including: • Detection of drones, vehicles and moving objects; • Applications in military technology, such as surveillance and early detection systems; • Applications in the automotive industry, such as proximity sensors and advanced driving assistance systems.

9. A MABIS breast tumor detection bra, according to claims 1 and 4, characterized in that it uses advanced metrics such as the Signal-to-Noise Ratio (SNR, 609).

10. A MABIS breast tumor detection bra, according to claims 1, 6 and 8, characterized in that its modular and portable design is housed in a 3D printed casing (30), reducing electromagnetic interference and allowing for efficient component replacement and upgrade, thus ensuring system scalability.

11. A MABIS breast tumor detection bra, according to claim 1, characterized in that the adjustable system includes: • Ergonomic straps (402) with length and position adjustment to ensure uniform contact with breast tissue; and • An adjustable waistband (403) that ensures proper pressure distribution, allowing the bra to be customized to the specific morphology of each patient, thus optimizing signal reception.

12. A MABIS breast tumor detection bra, according to claim 1, characterized in that it incorporates a graphical user interface (616) comprising a transceiver that connects to mobile devices and the internet, allowing the automatic generation of clinical reports (617) by means of artificial intelligence (618).

13. A MABIS breast tumor detection bra, according to claim 1, characterized in that the image processing system (60) applies advanced filtering techniques (603) to eliminate low-frequency noise (601), optimizing the interpretation of the reflected signals (50) with an adjustable attenuation range between 10 dB and 30 dB.

14. A MABIS breast tumor detection bra, according to claim 1, characterized in that the image processing system (60) employs an iterative backprojection scheme based on DAS and DMAS algorithms (613 and 614), using as input the signals normalized by Reprojection and rear projection (50), optimizing the reconstruction of images in real time.

15. A MABIS breast tumor detection bra, according to claim 1, characterized in that the signal generation and capture module (30) integrates radio frequency switches (203 and 204), configured to handle up to 2x32 channels, allowing dynamic selection of the emission (10a) and reception (10b) sensors to optimize data acquisition.

16. A MABIS breast tumor detection bra, according to claim 1, characterized in that the radio frequency switches (203 and 203) are controlled by an embedded microcontroller (210) that performs the switching with a response time of less than 100 ms, ensuring precise synchronization between the sensors and the signal capture module

Images