System and method for detecting tumour tissue by means of real-time gas analysis

Abstract

The present invention belongs to the area of materials analysis to determine the physical or chemical properties thereof, in particular real-time gas analysis for detecting tumour tissue. The invention provides a system and a method for detecting tumour tissue by means of real-time gas analysis during a surgical procedure.

Description

[0001] SYSTEM AND METHOD FOR THE DETECTION OF TUMOR TISSUE BY REAL-TIME GAS ANALYSIS

[0002] TECHNICAL FIELD OF THE INVENTION

[0003] The present invention belongs to the area of ​​materials analysis to determine their physical or chemical properties, in particular, real-time gas analysis for the detection of tumor tissue.

[0004] BACKGROUND OF THE INVENTION

[0005] Cancer is the second leading cause of death in the United States and ranks among the top ten causes of death worldwide. A cancer diagnosis inflicts physical, psychological, and financial hardship on the individual. Cancer can be classified into solid tumors and hematologic malignancies. In certain solid tumor cancers, surgery and tumor removal may be proposed as part of curative therapy. One of the major challenges in curative therapy for solid tumor cancers is local recurrence, which can be very high for some types of cancer. One of the main factors influencing recurrence is the presence of positive margins after surgery—that is, the presence of tumor cells at the edge of the surgical specimen—which can double the risk of recurrence in certain cancers.The reasons for not achieving clear margins in surgery are varied, but the main factor is the lack of technology capable of identifying the margins in real time, that is, the possibility of detecting whether the margin achieved contains tumor cells or not.

[0006] The methods currently available in the state of the art for detecting surgical margins present several drawbacks. Methods that achieve rapid analysis speed lack the sensitivity and specificity necessary for accurate margin analysis. On the other hand, methods that achieve greater precision increase surgical time and require teams trained in the analysis of histopathological samples. Furthermore, both rapid detection methods and more precise methods are not cost-effective in relation to the primary objective of accurately detecting margins.

[0007] For example, document EP3265822 describes an analytical method using mass and / or ion mobility spectrometry, comprising: using a first device to generate aerosol, smoke, or vapor from one or more regions of a first target biological material; and analyzing the mass and / or ion mobility of said aerosol, smoke, or vapor, or of ions derived therefrom, to obtain the first spectrometric data. The method may utilize an ambient ionization method.

[0008] US20150313538 discloses a method that includes the evaluation of tumor margins and the discrimination between tumor and non-tumor tissues by analyzing the composition of the smoke produced during cauterization resection of tissues. However, this document has some of the aforementioned defects.

[0009] Document EP2435814 discloses a system, method, and device for analyzing, locating, and identifying tissue types. The method includes analyzing, locating, and identifying one or more tissue samples and is characterized by comprising: (a) generating gaseous tissue particles from a point in the tissue sample(s); (b) transporting the gaseous tissue particles from the point to an analyzer; (c) using the analyzer to generate tissue data based on the gaseous tissue particles; and (d) analyzing, locating, and identifying the tissue sample(s) based on such data. The invention can be used in close conjunction with a surgical procedure, where one or more surgical tools form an integral part of the ionization, or as a standalone mass spectrometry probe for analyzing one or more tissue samples.

[0010] Due to the complexity of currently available methods in the prior art, there remains a need to provide new solutions for the real-time detection of tumor tissue during surgical resection procedures. SUMMARY OF THE INVENTION

[0011] The present invention provides a first object of invention, which comprises a portable system for the detection of tumor tissue by means of real-time gas analysis, characterized in that it comprises:

[0012] • a processor;

[0013] • a storage module connected to the processor;

[0014] • at least two photoionization sensors configured with different ionization energies to sense gases produced in operating procedures, connected to a data acquirer, wherein the data acquirer is connected to the processor and is configured to acquire electrical signals from the at least two photoionization sensors and process them to transform the electrical signals into digital chemical concentration signals;

[0015] • a communication module connected to the processor;

[0016] • a real-time clock module connected to the processor;

[0017] • an aspiration system connected to the processor and at least two photoionization sensors; and

[0018] • an electrical power source connected to the processor; wherein the processor is configured to: a) acquire digital signals from the data acquirer and measurement time from the real-time clock module; b) record the digital signals and measurement time in the storage module; c) preprocess the digital signals, wherein preprocessing the digital signals comprises the steps of: i. interpolating between the digital signals to obtain at least two signal concentration curves;

[0019] (i) vector normalize the at least two concentration signal curves, obtaining at least two normalized curves; (iii) obtain at least one vector, where the vector comprises at least two components, where each component of the at least two components is selected from a time point on one of the at least two normalized curves and a time point on one of the at least two concentration signal curves; (d) calculate a statistical value for each component of the vector; (e) define an abnormality threshold comprising a value that is entered through an input medium connected to the processor or by means of artificial intelligence executed by the processor; (f) segment the at least two components of the vector into time windows of a duration that is selected from a time magnitude;(g) evaluate in the time windows whether in at least one segment of each of them the abnormality threshold is exceeded, where exceeding the threshold corresponds to the analyzed tissue being a tumor tissue with a confidence level associated with the value defined as the abnormality threshold.;

[0020] Additionally, the present invention provides a second object of invention, which comprises a method for detecting tumor tissue by means of real-time gas analysis, characterized in that it comprises the steps of: a) providing a portable system for detecting tumor tissue by means of real-time gas analysis as described above; b) activating the aspiration system; c) executing the following steps by means of the processor: i. acquiring digital signals from the data acquirer and measurement time from the real-time clock module;

[0021] i. record the digital signals and measurement time in the storage module; ii. preprocess the digital signals, wherein preprocessing the digital signals comprises the steps of: a. interpolating between the digital signals to obtain at least two concentration signal curves; b. vectorally normalizing the at least two concentration signal curves, obtaining at least two normalized curves; c. obtaining at least one vector, wherein the vector comprises at least two components, where each component of the at least two components is selected from a time point on one of the at least two normalized curves and a time point on one of the at least two concentration signal curves; d) calculating a statistical value for each component of the vector;e) defining an abnormality threshold comprising a value that is entered through an information input device connected to the processor or through artificial intelligence executed by the processor; f) segmenting the at least two components of the vector into time windows of a duration selected from a time magnitude; and g) evaluating in the time windows whether in at least one segment of each of them the abnormality threshold is exceeded, where exceeding the abnormality threshold corresponds to the analyzed tissue being tumor tissue with a confidence level associated with the value defined as the abnormality threshold.

[0022] BRIEF DESCRIPTION OF THE FIGURES

[0023] Figure 1 illustrates a top cross-sectional view of a preferred embodiment of the system for detecting tumor tissue by real-time gas analysis. Figure 2 illustrates a preferred embodiment of the system for detecting tumor tissue by real-time gas analysis.

[0024] FIG. 3 illustrates a preferred embodiment of the method for detecting tumor tissue by real-time gas analysis.

[0025] FIG. 4 illustrates a user making use of a preferred embodiment of the system for the detection of tumor tissue by real-time gas analysis.

[0026] FIG. 5 illustrates a preferred embodiment of the system for the detection of tumor tissue by real-time gas analysis.

[0027] DETAILED DESCRIPTION OF THE INVENTION

[0028] The present invention provides a first object of invention, which comprises a portable system (1) for the detection of tumor tissue by means of real-time gas analysis, characterized in that it comprises:

[0029] • a processor (11 );

[0030] • a storage module (12) connected to the processor (11);

[0031] • at least two photoionization sensors (13) configured with different ionization energies to sense gases produced in operating procedures, connected to a data acquirer (14), wherein the data acquirer (14) is connected to the processor (11) and is configured to acquire electrical signals from the at least two photoionization sensors (13) and process them to transform the electrical signals into digital signals (141) of chemical concentration;

[0032] • a communication module (15) connected to the processor (11);

[0033] • a real-time clock module (16) connected to the processor (11);

[0034] • an aspiration system (17) connected to the processor (11) and to at least two photoionization sensors (13); and

[0035] • an electrical power source (18) connected to the processor (11); wherein the processor is configured to: a) acquire digital signals (141) from the data acquirer (14) and measurement time (161) from the real-time clock module (16); b) record the digital signals (141) and measurement time (161) in the storage module (12); c) preprocess the digital signals (141), wherein preprocessing the digital signals (141) comprises the steps of: i. interpolating between the digital signals (141) to obtain at least two concentration signal curves (142);

[0036] (i) vector normalizing the at least two concentration signal curves (142), obtaining at least two normalized curves (143); (iii) obtaining at least one vector (144), wherein the vector (144) comprises at least two components, where each component of the at least two components is selected from a time point on a curve of the at least two normalized curves (143) and a time point on a curve of the at least two concentration signal curves (142); (d) calculating a statistical value (140) for each component of the vector (144); (e) defining an abnormality threshold (145) comprising a value that is entered through an information input means (19) connected to the processor or by means of an artificial intelligence (20) executed by the processor; (f) segmenting the at least two components of the vector (144) into time windows of a duration that is selected from a time magnitude;(g) evaluate in the time windows whether in at least one segment of each of them the abnormality threshold (145) is exceeded, where exceeding the abnormality threshold (145) corresponds to the analyzed tissue being tumor tissue with a confidence level associated with the value defined as the abnormality threshold (145). Figures 1, 2, 4 and 5 illustrate preferred embodiments of the present invention, although they do not necessarily illustrate all the elements described.

[0037] Within the scope of the present invention, without limiting it, chemical concentration shall be understood as the amount of solute present in a solution, usually expressed in terms of the proportion of solute with respect to the solvent or the total solution, which can be measured, for example, in parts per million (ppm), percentage (by mass or volume), molarity, molality, mole fraction, among others.

[0038] Within the scope of the present invention, without limiting it, the connection of the aspiration system (17) to the processor (11) and to the at least two photoionization sensors (13) correspond to two connections of different nature, where the aspiration system (17) allows air to enter the system and come into contact with the at least two photoionization sensors (13), while the processor (11) allows the control and operation of the aspiration system (17)

[0039] Within the scope of the present invention, without limiting it, a time point corresponds to a specific time of one or more analyzed curves, such as, for example, the at least two concentration signal curves (142), at least two normalized curves (143), or it may also correspond to the moment when a statistical value (140), such as a Z-score, is determined. For example, when obtaining a vector that has values ​​from two normalized curves (CN1 and CN2), from two concentration signal curves (CS1 and CS2), and from two statistical values ​​that correspond to the Z-scores obtained from the normalized curves (ZCN1, ZCN2), a time point is defined that corresponds to the 5th second of the curves, since obtaining the Z-score at the 5th second of the normalized curve, thus obtaining a 6-dimensional vector with the values ​​(CN1 (5s), CN2 (5s), CS1 (5s), ZCN1 (5s), ZCN2 (5s)), where 5s corresponds to the 5th second 5.

[0040] Within the scope of the present invention, without limitation, the vector (144) may have values ​​that correspond not only to a point in time on the curve of the at least two normalized curves (143), but may also correspond to a point in time on the at least two concentration signal curves (142), or may also correspond to a point in time of a statistical value (140). In a preferred embodiment of the present invention, the statistical value (140) corresponds to the z-score, p-value, confidence level, or other statistical indicator that allows for the statistical analysis of the normalized values ​​of the components of the vector (144).

[0041] In a preferred embodiment of the present invention, without limiting it, the time windows have a duration ranging from 0.5 seconds to 3600 seconds.

[0042] Within the scope of the present invention, without limiting it, the analyzed gas corresponds to gas emitted by the use of an electrosurgical unit or other electrosurgical equipment during an operating procedure. This is because, through the use of such equipment, tissues are cauterized, and this cauterization emits gases, which are captured and analyzed by the system (1). The system (1) allows for the real-time identification of tumor margins during surgical resection by detecting characteristic volatile organic compounds (VOCs), analyzed by specialized sensors, which in this case correspond, without limiting the invention, to at least two photoionization sensors (13). The system (1) allows for obtaining a composite response spectrum for the classification of the cauterized tissue.This condition, the simultaneous use of PID sensors with different ionization energies for VOC spectrum reconstruction and diagnostic testing, is the technical basis of the invention and constitutes the main object of protection.

[0043] In a preferred embodiment of the present invention, without limitation, the system (1) is further characterized by comprising a housing. In a more preferred embodiment of the present invention, without limitation, the housing is a physical structure comprising three main sections: a front section, a rear section, and side surfaces with openings for passive or active ventilation. The housing may be made of anodized aluminum, medical-grade stainless steel, or high-strength engineering polymers such as polycarbonate or medical-grade ABS, and may incorporate thermal and electromagnetic insulators if required. Internal assembly is achieved by means of metal or plastic fasteners, which may include M6 threaded screws (7) or functional equivalents, without limiting the invention to this metric.In a preferred embodiment of the present invention, without limiting it, the real-time clock module (16) corresponds to the DS3231 RTC real-time clock. The measurement time (161) corresponds to the time in which the acquisition of electrical signals by the data acquirer (14) is carried out, which is delivered by the real-time clock module (16), where said measurement time (161) is delivered in seconds, or in hours:minutes:seconds format.

[0044] Within the scope of the present invention, without limiting it, the term storage module (12) shall be understood to mean any device that allows the storage of information, such as, for example, an SD or microSD card, a flash drive, an SSD memory, an HDD memory, among others that may be available in the state of the art.

[0045] Within the scope of the present invention, without limitation, the aspiration module (17) is responsible for extracting the gases generated by the electrosurgical unit from the surgical site to the sensor system of the device. In a preferred embodiment of the aspiration module (17), without limitation, a sterile aspiration tube, preferably made of medical-grade polyvinyl chloride (PVC) or silicone, is connected to a spigot-type connector fixed to the housing. This connector may be replaced by quick-coupling, threaded, push-fit, or retaining ring systems, provided that leak-proofness and compatibility with surgical environments are guaranteed. Internally, in this preferred embodiment, the gases are transported through polytetrafluoroethylene (PTFE) tubing, selected for its chemical inertness to VOCs and its thermal resistance.The tubes can be arranged in a sequential or parallel configuration, distributing the gases towards the sensor modules. In a preferred embodiment of the present invention, the aspiration module (17) comprises a motor, where the motor can be selected from brushless, brushed, diaphragm, or membrane types, provided it guarantees a stable aspiration flow rate between 0.3 and 1.0 liters per minute, preferably 0.5 L / min. In an embodiment of the present invention, the aspiration module is connected to an electrosurgical unit, where said electrosurgical unit can be an electrosurgical unit, as illustrated in FIG. 2 and FIG. 4. In a preferred embodiment of the present invention, without limiting it, the at least two photoionization sensors (13) correspond to two photoionization sensors configured with ionization energies that are in the range between 10.0 electronvolts (eV) and 12.0 eV.In an even more preferred embodiment, the at least two photoionization sensors (13) correspond to three photoionization sensors configured with ionization energies of 10.0 electronvolts (eV), 10.6 eV and 11.7 eV.

[0046] Within the scope of the present invention, without limiting it, the at least two photoionization sensors (13) are connected to the data acquirer (14), which comprises the electronic components necessary for signal conditioning, voltage regulation, and connection to the acquisition system. In a preferred embodiment, without limiting the present invention, the outputs of the data acquirer (14) are of the differential analog type, accessible via block-type terminals with contacts labeled S+ (signal), S- (reference), VCC (power supply), and GND (ground).

[0047] In a preferred embodiment of the present invention, without limiting it, the system (1) is further characterized by comprising electrochemical sensors and near-infrared sensors connected to the data acquisition unit (14). These electrochemical sensors can be used to measure gases such as H2S, CO2, CO, NH3, and SO2, while the infrared sensors can be used to measure CO2. These sensors allow for the detection of combustion products from different tissues.

[0048] In a preferred embodiment of the present invention, without limiting it, the system (1) is further characterized by comprising a temperature sensor and a relative humidity sensor. In a more preferred embodiment, without limiting it, the temperature sensor is a PT1000 thermistor and the relative humidity sensor is selected from the DHT22 and BME280 models. Within the scope of the present invention, without limiting it, these sensors allow for thermal and environmental compensation of the signal from at least two photoionization sensors (13), increasing the stability and accuracy of the analysis. The signals from these sensors are digitized by means of an analog-to-digital converter, preferably an ADS1 1 15 or equivalent, within the data acquisition unit and sent by the communication module to the processor.

[0049] Within the scope of the present invention, without limitation, the communication module (15) and the data acquirer (14) may be connected to the processor (11) by wired or wireless means. In a preferred embodiment of the module (11) or the data acquirer (14), they may be connected to the processor (11) by a protocol selected from the following: Modbus synchronous protocol, I2C (Inter-Integrated Circuit), UART (Universal Asynchronous Transmission), SPI (Serial Peripheral Interface), RS-485, CAN Bus, or Ethernet.

[0050] In a preferred embodiment of the present invention, without limiting it, the processor (1 1 ) is selected from Raspberry Pi 5, Raspberry Pi 3, Raspberry Pi 4), Jetson Nano, Jetson Orin, BeagleBone Black, STM32, ESP32, Teensy, Arduino Due, or any microcontroller with digital reception capability, execution of analysis algorithms and generation of user interface.

[0051] Within the framework of the present invention, without limiting it, with the aim of reducing high-frequency noise and unifying the sampling rate, resampling can be performed based on the average of consecutive pairs of samples and it is possible to interpolate the data using third-order polynomial functions to obtain a continuous and smoothed representation, homogeneous in the time domain, resulting in the data having a stable sampling frequency of 20 Hz, guaranteeing greater uniformity in the analysis.

[0052] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that preprocessing the digital signals (141) further comprises the step of temporarily resampling the digital signals (141) by means of resampling based on the average of consecutive pairs.

[0053] In a preferred embodiment of the present invention, without limitation, the system (1) is characterized in that the interpolation step between the digital signals (141) comprises selecting one type of interpolation from among cubic interpolation, linear interpolation, and quadratic interpolation. It is also possible to use interpolation methods based on the Fourier transform of the signal. In this approach, the signal is transformed to the frequency domain, where as many zeros as values ​​to be interpolated are inserted (zero-padding), then the inverse transform is applied, obtaining the interpolated signal in the time domain.

[0054] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that the step of preprocessing the digital signals (141) further comprises the step of filtering the at least two concentration signal curves (142) by means of a low-pass filter.

[0055] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that the step of preprocessing the digital signals (141) further comprises the step of filtering the at least two concentration signal curves (142) using a Savitzky-Golay algorithm. In a further preferred embodiment, the Savitzky-Golay algorithm has a sliding window of 1 second.

[0056] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that the step of preprocessing the digital signals (141) further comprises the step of temporally aligning the at least two concentration signal curves (142). This alignment can be carried out by cross-correlation within a time window of 1 second, identifying time lags and correcting them to align the signals.

[0057] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that the step of preprocessing the digital signals (141) further comprises the step of applying a threshold to eliminate spurious values ​​from the at least two concentration signal curves (142). In a further preferred embodiment of the present invention, the threshold for eliminating spurious values ​​is 10 ppm, whereby, when none of the sample concentrations simultaneously exceeds 10 ppm, they are discarded since signals below this threshold behave nonlinearly, changing the proportions when normalized by L2.

[0058] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that the step of vector normalizing the at least two concentration signal curves (142) comprises the use of an L2 norm.

[0059] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that it comprises an alarm module connected to the processor (11), which is configured to deliver information about the analyzed tissue.

[0060] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that the alarm module comprises a visual interface and an auditory interface.

[0061] In a preferred embodiment of the present invention, without limitation, the system (1) is characterized in that the visual interface comprises a data presentation means selected from a monitor, a second processor, a tablet, and a smartphone. In a more preferred embodiment, without limitation, the visual interface corresponds to an LCD, OLED, or TFT monitor, with or without touch capability, and a variable size between 3.5” and 10”. The visual interface can display numerical and graphical diagnostic indicators, including color coding (e.g., red for tumor tissue and blue for healthy tissue), probability percentage, values ​​in parts per million (ppm), and real-time curves.

[0062] In a preferred embodiment of the present invention, without limiting it, the system (1) is characterized in that the auditory interface comprises a speaker, where said speaker may correspond to an acoustic transducer or piezoelectric loudspeaker that emits coded sound signals when the algorithm detects a probability higher than the established threshold.

[0063] In a preferred embodiment of the present invention, without limitation, the electrical power source (18) is an external 12-volt direct current (VDC) power source connected to an internal rechargeable battery with a preferred capacity of 20,000 milliampere-hours (mAh). In another preferred embodiment, the internal rechargeable battery has a capacity ranging from 15,000 to 30,000 mAh.

[0064] In a preferred embodiment of the present invention, without limiting it, the data acquirer (14) is configured to acquire electrical signals from at least two photoionization sensors (13) at a frequency in the range of 0.1 Hz to 1 kHz. Within the scope of the present invention, without limiting it, the acquired electrical signals are digitized by analog-to-digital conversion, generating discrete time sequences with a resolution ranging from 12 to 24 bits in a range of 0-10 V. In a further preferred embodiment, the voltages recorded by the sensors are converted to concentrations in parts per million using experimentally calibrated linear relationships, wherein the functions used are:

[0065] • Sensor 1: concentration = (voltage - 61.6) / 9.2

[0066] • Sensor 2: concentration = (voltage - 60.8) / 7.7

[0067] • Sensor 3: concentration = (voltage - 55.9) / 1.6

[0068] In a preferred embodiment of the present invention, without limiting it, the system (1) has been calibrated using three known concentrations of isobutylene (10 ppm, 100 ppm, and 1000 ppm), introduced from standard gas cylinders. Each concentration was measured six times independently, and the corresponding voltage signals for the three PID sensors were recorded. The PID sensors behave as linear time-invariant (LTI) systems when the concentration exceeds the 10 ppm threshold.

[0069] Within the framework of the present invention, without limiting it, it shall be understood that, when forming the concentration signal curves (142), they are interpolated between the digital signals (141) as a function of the time obtained by means of the measurement time (161), where the at least two concentration signal curves (142) have on one axis the ppm variable and on the other, the time variable (either in seconds, or in hours:minutes:seconds format).

[0070] In a preferred embodiment of the present invention, without limiting it, the time magnitude of the time windows has a duration in the range of 1 to 3600 seconds. In a preferred embodiment of the present invention, the time magnitude of the time windows has a duration of 5 seconds.

[0071] Within the scope of the present invention, without limiting it, the abnormality threshold (145) is a value of the same type as the statistical value (140), and therefore, both are comparable. That is, the abnormality threshold (145) may correspond to the z-score, confidence level, or other statistical indicator. In a preferred embodiment of the present invention, without limiting it, the abnormality threshold corresponds to a z-score value greater than 1.5. In a further preferred embodiment, this abnormality threshold (145) corresponds to a z-score value of 1.64. It should be noted that these values ​​were obtained through experimentation.

[0072] In a preferred embodiment of the present invention, without limiting it, the processor (1 1 ) is configured to display on the visual interface numerical and graphical indicators of the diagnosis, including color coding (e.g., red for tumor tissue (i.e., when the measurement exceeds the abnormality threshold) and blue for healthy tissue), probability percentage, values ​​in parts per million (ppm), and real-time curves.

[0073] In a preferred embodiment of the present invention, without limitation, the processor (11) is configured to trigger an alarm via the alarm module if the measurement exceeds the abnormality threshold. This alarm can be issued through a visual interface, such as a color-coded system where red indicates tumor tissue and blue indicates healthy tissue, or through an auditory interface, by means of a sound indicating the detection of tumor tissue. In a preferred embodiment, the sound is a progressive alert. No sound is emitted when healthy tissue is identified. In the presence of suspicious tissue, a low-intensity, intermittent sound signal is activated. If tumor tissue is detected, the device emits a continuous sound clearly distinguishable in pitch and frequency from the standard sound of a conventional electrosurgical unit.This auditory differentiation is intended to clearly alert the surgeon about the type of tissue being sectioned, without interfering with the usual surgical flow.

[0074] Within the scope of the present invention, without limitation, the abnormality threshold (145) can be defined by a user (e.g., a surgeon) via an input device, which may include a keyboard, mouse, or touchscreen, among other means for inputting information to a processor. Alternatively, the abnormality threshold (145) can be established by an artificial intelligence (20), which, based on data from a database with which it has been trained, can establish criteria for determining the threshold (145). The photoionization sensors can detect more than 400 volatile organic compounds. Each sensor has a different response factor for each compound, determined by its ionization energy.The signal delivered by the sensor, in voltage, corresponds to the linear sum of the mole fractions of the compounds present, weighted by the inverse of their response factor, as shown in the following equation:.

[0075] Where RFmix is ​​the response factor of the volatile organic compound mixture, X¡ the mole fraction of compound i, C¡ the concentration in ppm of compound i, Ct the total concentration of volatile organic compounds, and V is the voltage delivered by the sensor.

[0076] Among the relevant organic compounds are:

[0077] Table No. 1: Organic compounds that can be analyzed using photoionization sensors.

[0078] In a preferred embodiment of the present invention, without limiting it, the measurement time (161) corresponds to time data (hour, minute, second) and / or date data (day, month, year) in which the measurement is carried out, where time and / or date is assigned to each data of the digital signals (141).

[0079] Additionally, the present invention provides a second object of invention, which comprises a method for detecting tumor tissue by means of real-time gas analysis, characterized in that it comprises the steps of: a) providing a portable system (1) for detecting tumor tissue by means of real-time gas analysis as described above; b) activating the aspiration system (17); c) executing the following steps by means of the processor (11): i. acquiring digital signals (141) from the data acquirer (14) and measurement time (161) from the real-time clock module (16);

[0080] i. recording the digital signals (141) and measurement time (161) in the storage module (12); iii. preprocessing the digital signals (141), wherein preprocessing the digital signals (141) comprises the steps of: a. interpolating between the digital signals (141) to obtain at least two concentration signal curves (142); b. vector-normalizing the at least two concentration signal curves (142), obtaining at least two normalized curves (143); c. obtaining at least one vector (144), wherein the vector (144) comprises at least two components, where each component of the at least two components is selected from a time point on a curve of the at least two normalized curves (143) and a time point on a curve of the at least two concentration signal curves (142); iv. calculating a statistical value (140) for each component of the vector

[0081] (144); v. define an abnormality threshold (145) comprising a value that is entered through an information input means (19) connected to the processor or by means of an artificial intelligence (20) executed by the processor; vi. segmentation of the at least two components of the vector (144) into time windows of a duration selected from a time magnitude; and vii. evaluate in the time windows whether in at least one segment of each of them the abnormality threshold (145) is exceeded, where exceeding the abnormality threshold (145) corresponds to the analyzed tissue being tumor tissue with a confidence level associated with the value defined as the abnormality threshold

[0082] (145).

[0083] A preferred embodiment of the present method is illustrated in FIG. 3

[0084] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the step of preprocessing the digital signals (141) further comprises the step of temporarily resampling the digital signals (141) by means of a resampling based on the average of consecutive pairs;

[0085] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the interpolation step between the digital signals (141) comprises selecting an interpolation type from among cubic interpolation, linear interpolation, and quadratic interpolation. It is also possible to use interpolation methods based on the Fourier transform of the signal. In this approach, the signal is transformed to the frequency domain, where as many zeros as values ​​to be interpolated are inserted (zero-padding), then the inverse transform is applied, obtaining the interpolated signal in the time domain. In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the step of preprocessing the digital signals (141) further comprises the step of filtering the at least two concentration signal curves (142) using a low-pass filter.

[0086] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the step of preprocessing the digital signals (141) further comprises the step of filtering the at least two concentration signal curves (142) using a Savitzky-Golay algorithm. In a further preferred embodiment, the Savitzky-Golay algorithm has a sliding window of 1 second.

[0087] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the step of preprocessing the digital signals (141) further comprises the step of temporally aligning the at least two concentration signal curves (142). This alignment can be carried out by cross-correlation within a 1-second time window, identifying time lags and correcting them to align the signals.

[0088] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the step of preprocessing the digital signals (141) further comprises the step of applying a threshold to eliminate spurious values ​​from the at least two concentration signal curves (142). In a further preferred embodiment of the present invention, the threshold for eliminating spurious values ​​is 10 ppm, whereby, when none of the sample concentrations simultaneously exceeds 10 ppm, they are discarded since signals below this threshold behave nonlinearly, changing the proportions when normalized by L2.

[0089] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the step of vectorially normalizing the at least two concentration signal curves (142) comprises the use of an L2 norm.

[0090] Within the scope of the present invention, without limitation, it is possible to form a second vector (146) with a plurality of components, where each component of said plurality of components can be selected from: a time point of one of the at least two concentration signal curves (142), a time point of one of the at least two normalized curves (143), and from the statistical values ​​(140) at a time point of each of the at least two normalized curves (143), or it can include all of the foregoing variants. In a further preferred embodiment, it corresponds to a vector of 9 components at a time point, where 3 correspond to values ​​at said time point of 3 concentration signal curves, another 3 correspond to values ​​at said time point of 3 normalized curves, and another 3 correspond to statistical values ​​at said time point.

[0091] Within the scope of the present invention, without limiting it, it is possible to form a matrix (147) with a plurality of second vectors (148) (composed of more than one second vector (146)) over a time range. Within the scope of the present invention, without limiting it, it is understood that, when considering a time range for the creation of the matrix, a plurality of time points are considered, and therefore it is possible to obtain a plurality of second vectors (148).

[0092] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that it further comprises the step of applying a layer of artificial intelligence (20) onto the matrix (147).

[0093] In a preferred embodiment of the present invention, without limiting it, the method is characterized in that the artificial intelligence (20) is selected from a convolutional neural network, a short-term-long-term memory network, or a combination thereof. In another preferred embodiment, without limitation, the method is characterized by the selection of artificial intelligence from the following models or a combination thereof: Linear Regression, Logistic Regression, Polynomial Regression, Support Vector Machines (SVM), Decision Trees, Random Forests, Gradient Boosting (XGBoost, LightGBM, CatBoost), k-Nearest Neighbors (k-NN), Multilayer Perceptron (MLP), Naive Bayes, Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) Networks, Gated Recurrent Unit (GRU) Networks, Transformers for Classification, Ridge Regression, Lasso Regression, Elastic Net, Rule-based Classification (RuleFit, Decision Table), Ensemble Models (Stacking,Blending), Gaussian Process Regression (GPR), k-means, Hierarchical Clustering (Agglomerative, Divisive), DBSCAN, Gaussian Mixture Models (GMM), PCA (Principal Component Analysis), SVD (Singular Value Decomposition), t-SNE (t-disthbuted Stochastic Neighbor Embedding), UMAP (Uniform Manifold Approximation and Projection), Autoencoders (unsupervised version), Self-Organizing Maps (SOM), Isomap, Spectral Clustering, Multiple Correspondence Analysis (MCA), Density-Based Clustering (HDBSCAN), Deep Embedded Clustering (DEC), Factor Analysis, Topic Models (LDA - Latent Dirichlet Allocation), Hierarchical Heatmaps (heatmap + dendrograms), Feature Agglomeration, Isolation Forest, Convolutional Neural Networks (CNN) [LeNet, AlexNet, VGG, GoogLeNet (Inception), ResNet, DenseNet, MobileNet, ShuffleNet, EfficientNet], RNN and variants [Elman RNN, LSTM, GRU], Transformers and Related Models [Transformer, BERT, GPT, GPT-2, GPT-3, GPT-4, XLNet, T5,DistilBERT, Vision Transformer (ViT)], Generative Adversarial Networks (GAN) [GAN, DCGAN, WGAN, WGAN-GP, StyleGAN, StyleGAN2, CycleGAN, Pix2Pix, BigGAN, SAGAN], Autoencoders Profundos [Autoencoder Clásico, Autoencoder Vahacional (VAE), Denoising Autoencoder (DAE), Sparse Autoencoder, Convolutional Autoencoder, Contractive Autoencoder], Redes Neuronales de Grafo (GNN) [Graph Convolutional Network (GCN), Graph Attention Network (GAT), GraphSAGE], Capsule Networks (CapsNet), Modelos de Difusión [Denoising Diffusion Probabilistic Models (DDPM), Stable Diffusion, Imagen, DALL- E], Modelos Auto-supervisados (Self-Supervised) [SimCLR, BYOL, MoCo], Modelos Híbridos y Otros [Siamese Networks, Meta-Learning (MAML, Reptile), Zero-Shot Learning, Few-Shot Learning],

Claims

CLAIMS 1. A portable system (1) for the detection of tumor tissue by real-time gas analysis, CHARACTERIZED in that it comprises • a processor (1 1 ); • a storage module (12) connected to the processor (1 1 ); • at least two photoionization sensors (13) configured with different ionization energies to sense gases produced in operating procedures, connected to a data acquirer (14), wherein the data acquirer (14) is connected to the processor (1 1 ) and is configured to acquire electrical signals from the at least two photoionization sensors (13) and process them to transform the electrical signals into digital signals (141 ) of chemical concentration; • a communication module (15) connected to the processor (1 1 ); • a real-time clock module (16) connected to the processor (1 1 ); • an aspiration system (17) connected to the processor (11) and to at least two photoionization sensors (13); and • an electrical power source (18) connected to the processor (11); wherein the processor is configured to: a) acquire digital signals (141) from the data acquirer (14) and measurement time (161) from the real-time clock module (16); b) record the digital signals (141) and measurement time (161) in the storage module (12); c) preprocess the digital signals (141), wherein preprocessing the digital signals (141) comprises the steps of: i. interpolating between the digital signals (141) to obtain at least two concentration signal curves (142); (i) vectorially normalize the at least two concentration signal curves (142), obtaining at least two normalized curves (143); (iii) obtain at least one vector (144), wherein the vector (144) comprises at least two components, where each component of the at least two components is selected from a time point on a curve of the at least two normalized curves (143) and a time point on a curve of the at least two concentration signal curves (142); (d) calculate a statistical value (140) for each component of the vector (144); (e) define an abnormality threshold (145) comprising a value entered through an information input means (19) connected to the processor or by means of an artificial intelligence (20) executed by the processor; (f) segment the at least two components of the vector (144) into time windows of a duration selected from a time magnitude;(g) evaluate in the time windows whether in at least one segment of each of them the abnormality threshold (145) is exceeded, where exceeding the abnormality threshold (145) corresponds to the analyzed tissue being a tumor tissue with a confidence level associated with the value defined as the abnormality threshold (145).; 2. The system of claim 1, CHARACTERIZED in that it further comprises a housing.

3. The system of claim 1, CHARACTERIZED in that it further comprises electrochemical sensors and near-infrared sensors attached to the data acquirer (14).

4. The system of claim 1, CHARACTERIZED in that it further comprises a temperature sensor and a relative humidity sensor.

5. The system of claim 1, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of temporarily resampling the digital signals (141) by means of resampling based on the average of consecutive pairs.

6. The system of claim 1, CHARACTERIZED in that interpolating between the digital signals (141) comprises selecting a type of interpolation from cubic interpolation, linear interpolation, quadratic interpolation.

7. The system of claim 1, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of filtering the at least two concentration signal curves (142) by means of a low-pass filter.

8. The system of claim 1, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of filtering the at least two concentration signal curves (142) filtered by a Savitzky-Golay algorithm.

9. The system of claim 1, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of temporarily aligning the at least two concentration signal curves (142).

10. The system of claim 1, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of applying a threshold to eliminate spurious values ​​from the at least two concentration signal curves (142). 1 1. The system of claim 1, CHARACTERIZED in that vectorially normalizing the at least two concentration signal curves (142) comprises the use of an L2 norm.

12. The system of claim 1, CHARACTERIZED in that it comprises an alarm module connected to the processor (1 1 ), which is configured to deliver information about the analyzed tissue.

13. The system of claim 12, CHARACTERIZED in that the alarm module comprises a visual interface, and an auditory interface.

14. The system of claim 13, CHARACTERIZED in that the visual interface comprises a means for presenting data that is selected from a monitor, a second processor, a tablet, a smartphone.

15. The system of claim 13, CHARACTERIZED in that the auditory interface comprises a speaker.

16. The system of claim 1, CHARACTERIZED in that the time windows have a duration that is in the range between 0.5 seconds and 3600 seconds.

17. The system of claim 1, CHARACTERIZED in that the aspiration module is connected to an electrosurgical unit.

18. A method for detecting tumor tissue by real-time gas analysis, CHARACTERIZED in that it comprises the steps of: a. providing a portable system (1) for detecting tumor tissue by real-time gas analysis as described in claims 1 to 17; b. activating the aspiration system (17); c. executing the following steps by means of the processor (11): i. acquiring digital signals (141) from the data acquirer (14) and measurement time (161) from the real-time clock module (16); c.ii recording the digital signals (141) and measurement time (161) in the storage module (12); c.iii preprocessing the digital signals (141), wherein preprocessing the digital signals (141) comprises the steps of: .1 interpolate between the digital signals (141) to obtain at least two concentration signal curves (142); .2 vector normalize the at least two concentration signal curves (142), obtaining at least two normalized curves (143); and .3 obtain at least one vector (144), wherein the vector (144) comprises at least two components, wherein each component of the at least two components is selected from a time point of a curve of the at least two normalized curves (143) and a time point of a curve of the at least two concentration signal curves (142); c.iv calculate a statistic value (140) for each component of the vector (144); c.v define an abnormality threshold (145) comprising a value entered through an information input means (19) connected to the processor or by means of an artificial intelligence (20) executed by the processor; c.vi segment the at least two components of the vector (144) into time windows of a duration selected from a time magnitude; and c.vii evaluate in the time windows whether in at least one segment of each of them the abnormality threshold is exceeded (145), where exceeding the abnormality threshold (145) corresponds to the fact that the analyzed tissue corresponds to a tumor tissue with a confidence level associated with the value defined as the abnormality threshold (145).

19. The method of claim 18, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of temporarily resampling the digital signals (141) by means of a resampling based on the average of consecutive pairs.

20. The method of claim 18, CHARACTERIZED in that interpolating between the digital signals (141) comprises selecting a type of interpolation from cubic interpolation, linear interpolation, quadratic interpolation.

21. The method of claim 18, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of filtering them. less two concentration signal curves (142) using a low-pass filter.

22. The method of claim 18, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of filtering the at least two concentration signal curves (142) using a Savitzky-Golay algorithm.

23. The method of claim 18, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of temporarily aligning the at least two concentration signal curves (142).

24. The method of claim 18, CHARACTERIZED in that preprocessing the digital signals (141) further comprises the step of applying a threshold to eliminate spurious values ​​from the at least two concentration signal curves (142).

25. The method of claim 18, CHARACTERIZED in that vectorially normalizing the at least two concentration signal curves (142) comprises the use of an L2 norm.

26. The method of claim 18, CHARACTERIZED in that it further comprises a step of forming a second vector (146) with a plurality of components, wherein each component of said plurality of components can be selected from: a time point of a curve of the at least two concentration signal curves (142), a time point of a curve of the at least two normalized curves (143), and from the statistical values ​​(140) at a time point of each curve of the at least two normalized curves (143), or can include all of the foregoing variants.

27. The method of claim 26, CHARACTERIZED in that it further comprises a step of forming an array (147) with a plurality of second vectors (148) in a time range, wherein the plurality of second vectors (148) comprises more than one second vector (146).

28. The method of claim 27, CHARACTERIZED in that it further comprises the step of applying a layer of artificial intelligence (20) onto the matrix (147).

29. The method of claim 28, CHARACTERIZED in that the artificial intelligence (20) is selected from a convolutional neural network, a short-term-long-term memory network, or a combination thereof.

30. The method of claim 18, CHARACTERIZED in that the time windows have a duration in the range of 0.5 to 3600 seconds.

31. A computer-implemented method for analyzing chemical concentration data of gases for the detection of tumor tissue by real-time gas analysis, CHARACTERIZED in that it comprises the steps of: a. acquiring digital signals (141) from a data acquirer (14) connected to the computer and measurement time (161) from a real-time clock module (16) connected to the computer; b. recording the digital signals (141) and measurement time (161) in the storage module (12); c. preprocessing the digital signals (141), wherein preprocessing the digital signals (141) comprises the steps of: i. interpolating between the digital signals (141) to obtain at least two concentration signal curves (142); c.ii vector normalizing the at least two concentration signal curves (142), obtaining at least two normalized curves (143); and c.iii obtain at least one vector (144), wherein the vector (144) comprises at least two components, wherein each component of the at least two components is selected from a time point of a curve of the at least two normalized curves (143) and a point. temporal of a curve of the at least two concentration signal curves (142); d. calculate a statistical value (140) for each component of the vector (144); e. define an abnormality threshold (145) comprising a value entered through an information input means (19) connected to the processor or by means of an artificial intelligence (20) executed by the processor; f. segmentation of the at least two components of the vector (144) into time windows of a duration selected from a time magnitude; and g. evaluate in the time windows whether in at least one segment of each of them the abnormality threshold (145) is exceeded, where exceeding the abnormality threshold (145) corresponds to the analyzed tissue being tumor tissue with a confidence level associated with the value defined as the abnormality threshold (145).

Images