Method for diagnosing renal cell carcinoma using ultrasonic signal analysis and ultrasonic images

Abstract

Ultrasound apparatus for performing a diagnostic examination of the kidney to diagnose renal cell carcinoma, comprising at least an ultrasound probe and relative control means, acquisition and storage means of the ultrasound signals reflected by the tissues, characterized in that said ultrasound apparatus is configured to perform the following acquisition procedure: - emission of at least one beam of broadband ultrasound pulses towards the kidney of the patient, reception and storage of the original ultrasound signals reflected by the kidney tissues and processing of said signals to obtain an ultrasound image of the kidney, - emission of a beam of narrowband ultrasound pulses towards the kidney of the patient while keeping the probe in the same position, reception and storage of the ultrasound signals reflected by the kidney tissues in response to the narrowband ultrasound pulse beam sent at point 130), and in that said ultrasound apparatus is configured to perform the following calculation method of diagnostic parameters: - analysis of the broadband reflected signals to individualize one or more suspicious areas by means of the analysis of the statistical distribution of the coefficients of the associated wavelet transform, - extraction of the ultrasound signals reflected in response to the narrowband pulse beam for each suspicious area, - calculation of diagnostic parameters indicative of the presence and type of tumor (renal cell carcinoma, benign tumor) as a function of values related to the important frequencies of the spectrum associated with the signals reflected in response to the narrowband pulse beam, or as a function of descriptive parameters of the spectrum associated with the signals reflected in response to the broadband pulse beam.

Description

Technical Field

[0001] This invention relates to a method for diagnosing renal cell carcinoma using ultrasound signal analysis and ultrasound images. Background Technology

[0002] The increasing use of diagnostic imaging has led to the detection of a wider range of lesions and / or abnormal masses, often discovered incidentally during testing for other reasons (casual tumors).

[0003] However, since many incidentally discovered lesions never cause any disease, the discovery of incidental tumors increases the risk of overdiagnosis. When an incidental tumor is found, the clinician must verify that the lesion is not dangerous so that he can perform other tests to determine the nature of the lesion.

[0004] In cases of kidney masses, distinguishing between tumorous and benign masses is particularly difficult.

[0005] According to Oostenbrugge et al., “Diagnostic Imaging for solid renal tumors: apictorial review”, Kidney Cancer 2 (2018) 79-93, 20% of incidentally discovered renal tumors are benign lesions that do not require further treatment, but the majority of them are renal cell carcinomas that require aggressive treatment.

[0006] Therefore, it is clearly essential to have an effective diagnostic method that can determine whether a kidney lesion is considered malignant and is inexpensive, rapid, and non-invasive for the patient.

[0007] In diagnostic imaging technology, a diagnostic imaging technique using ultrasound is well known for many different test types.

[0008] Referring to ultrasound scans used for the diagnostic classification of renal tumors, transabdominal B-mode ultrasound with frequencies between 3 and 6 MHz was used, based on what is known in the current state of technology.

[0009] The patient lies supine or on their side, and the kidneys are scanned longitudinally and laterally.

[0010] The effectiveness of ultrasound in detecting kidney tumors depends on the tumor's echogenicity, size, and location.

[0011] During ultrasound scanning, hypoechoic (dark areas on the ultrasound image relative to the surrounding renal parenchyma), hyperechoic (bright areas on the ultrasound image relative to the surrounding renal parenchyma), or isoechoic renal tumors were detected. These renal tumors are undetectable (or almost undetectable) by means of grayscale analysis of the corresponding ultrasound images because they appear similar to the surrounding renal parenchyma on the surface.

[0012] Isoechoic and hypoechoic tumors are more difficult to detect by ultrasound scans, especially if they are smaller than 20 mm and are isoechoic.

[0013] Furthermore, even when a tumor mass is detected, its classification remains problematic. The purpose of classification is to differentiate renal cell carcinoma from benign masses (angiomyolipomas, eosinophilic tumors, etc.). With current technology, the echogenicity of known tumors cannot be differentiated between histological subtypes, therefore benign tumors cannot be reliably distinguished from malignant tumors.

[0014] Doppler ultrasound is also known to be used with or without contrast media, but the effectiveness of these techniques in characterizing lesions as malignant or benign has not yet been demonstrated.

[0015] Other methods for analyzing ultrasound images or signals aimed at improving tumor detection using ultrasound scanning are also known. For example, WO2018 / 157130 describes a method for obtaining improved ultrasound images to allow physicians to make a diagnosis of liver or kidney tumors based solely on ultrasound images. Summary of the Invention

[0016] Technical issues

[0017] In any case, under the current state of technology, this and other known ultrasound analysis modalities are limited for various reasons.

[0018] First, they do not allow for the classification of various histological types of renal tumors. Second, they do not allow for the effective detection of isoechoic masses that are barely visible (or not visible at all) on ultrasound images. Finally, they do not allow for interpretation outside the subjective analysis of the physician performing the ultrasound scan.

[0019] Purpose of the invention

[0020] The object of the present invention is to provide a diagnostic method for individualizing renal tumor masses or classifying their type based solely on the analysis of ultrasound signals, which overcomes the limitations associated with known embodiments in the current state of the art.

[0021] Specifically, one approach allows for the equally efficient detection of isoechoic masses to classify various histological types and is independent of the subjective analysis of the physician performing the ultrasound scan.

[0022] According to another objective, the present invention provides an ultrasound device configured to perform such a diagnostic method. Detailed Implementation

[0023] The following is a reference to the appendix. Figures 1 to 3 The present invention is described, and appended Figures 1 to 3 A summary flowchart is shown, relating to the acquisition procedure, the calculation of diagnostic parameters, and the individualization of non-homogeneous regions using wavelet transform analysis.

[0024] First of all, it should be noted that the proposed method involves the analysis of ultrasound signals detected by transabdominal renal ultrasound scanning, which is performed according to the common situation regarding the acquisition location under the current state of technology.

[0025] Therefore, the acquisition location and ultrasound imaging plane are the commonly used acquisition location and ultrasound imaging plane.

[0026] Specifically, based on known and commonly used practices, preparation for a kidney ultrasound scan is not required because there is no proven evidence that fasting can improve kidney visibility.

[0027] Ultrasound scans are typically performed with the patient in a supine or semi-recumbent position, with one side of the kidney raised for scanning. In cases of bulging of the colic curve, it may be necessary to position the probe posteriorly to the axillary line or even further back. In these situations, the subcostal approach is not feasible, and the kidney scan must be performed using a window in the intercostal space; however, it is not permissible to represent the entire kidney in a single scan.

[0028] In children and thin individuals, the prone position can be used due to the reduced thickness of the paravertebral muscles on the back.

[0029] These instructions regarding the selection of the ultrasound acquisition plane can also be applied to ultrasound scans performed using the apparatus and method according to the invention.

[0030] The device according to the invention includes at least an ultrasound probe, a control device configured to guide the probe, a device for acquiring and storing ultrasound signals reflected by tissue, a data processing device for the signals configured to acquire B-mode ultrasound images, and a visualization device for the acquired images.

[0031] These are features provided in devices known in the current state of technology, so they will not be described in more detail here.

[0032] The device according to the invention is characterized in that it is configured to acquire and also store “raw” ultrasound signals, also known as radio frequency ultrasound signals (i.e., ultrasound signals received by the probe before their processing required to obtain ultrasound images), and is equipped with computing devices loaded with suitable computer programs configured to perform the diagnostic methods described below.

[0033] First of all, the following text describes the acquisition of renal ultrasound images and the transmission of ultrasound signals to the kidneys.

[0034] It is clear that the ultrasound signal will travel through propagation lines depending on the location of acquisition, and these propagation lines pass through multiple different tissues before reaching the kidney.

[0035] In the first embodiment, the selection of signals relating to the region of interest of the kidney will be accomplished by an expert in the field using tools known in themselves and known in the current state of the art.

[0036] Preferably, the device according to the invention is configured to segment the acquired ultrasound images to automatically determine regions of interest related to the kidneys.

[0037] The device according to the invention is also configured to perform the following acquisition process of ultrasonic signals for each acquisition location. For this purpose, the device includes an electronic computing device on which a computer program is loaded, configured to perform the methods described below.

[0038] 100) Firing at least a broadband ultrasound pulse beam toward the patient’s kidneys.

[0039] For clarity, it will be noted that the ultrasonic pulse beam is intended to be a pulse group, each pulse being emitted by a piezoelectric or CMUT transducer. In fact, it will be noted that where the invention is described with reference to a piezoelectric transducer for the sake of brevity, other transducers known in the present art, such as CMUT capacitive transducers, may also be used.

[0040] 110) Receive ultrasound signals reflected by kidney tissue in response to a pulse beam transmitted at point (100), and store the received raw ultrasound signals; 120) Process the signal stored at point 110) to obtain a kidney B-mode ultrasound image.

[0041] During the imaging process, the ultrasound probe operates in a broadband manner. When a probe with a nominal frequency of 3.5 MHz is preferably used, the effective bandwidth is between 0 and 7 MHz.

[0042] 130) While keeping the probe in the same position, a narrow-band ultrasound pulse beam is emitted toward the patient's kidney.

[0043] Preferably, the frequency used is close to, but slightly lower than, the probe's nominal frequency. When using a probe with a nominal frequency of 3.5 MHz, the frequency used for narrowband pulse transmission is preferably equal to 3 MHz.

[0044] It will be noted that only the fact that the device is configured to automatically perform the acquisition process just described allows for the transmission of both broadband and narrowband signals in the same acquisition plane, because narrowband signals can be transmitted immediately after the listening (i.e., receiving) time for the broadband signal has ended. This allows for numerical analysis in which the response to both broadband and narrowband pulses is certain for every part of the organization.

[0045] 140) Receive and store the raw ultrasound signal reflected by the kidney tissue in response to a narrow-band ultrasound pulse beam transmitted at point 130).

[0046] The device according to the invention is further configured to: after the acquisition process just described, perform a method for calculating diagnostic parameters, which includes the following steps: 200) By analyzing the statistical distribution of the coefficients of the associated wavelet transform, the original ultrasound signal stored at point 110) in response to the broadband pulse beam is analyzed to individualize one or more suspicious regions.

[0047] The individualization may also be performed by the operator using a suitable graphical interface to analyze other suspicious areas that were not identified at point 200.

[0048] For each suspicious region individualized at point 200, the following method is used to calculate diagnostic parameters indicating the presence of a renal tumor.

[0049] 210) Extract a signal associated with at least one segment contained in the suspicious region from the original ultrasound signal reflected by the kidney tissue in response to a narrow-band ultrasound pulse beam stored at point 140; 220) Perform a frequency transformation on the signal extracted at point 210) to at least calculate the spectrum obtained in response to the narrowband ultrasound signal. Preferably, the frequency transformation is an FFT; 230) Extract values ​​related to a plurality of important frequencies from the spectrum calculated at point 220). Preferably, the frequencies include: —The nominal frequency of the narrowband signal being transmitted. —First harmonic (i.e., twice the nominal frequency of the transmitted narrowband signal); —The first subharmonic of the signal (i.e., half of the nominal frequency of the transmitted narrowband signal).

[0050] In the case of a narrowband signal with a frequency of 3 MHz, in a preferred embodiment, the three extracted values ​​are then values ​​related to 1.5 MHz, 3 MHz, and 6 MHz.

[0051] 240) Extract a signal associated with at least one segment contained in the suspicious region from the original ultrasound signal reflected by the kidney tissue in response to a broadband signal stored at point 110); 250) Calculate the frequency transform of the signal extracted at point 240) to obtain at least one spectrum in response to the broadband ultrasound signal. Preferably, the frequency transform is an FFT.

[0052] 260) Extract one or more descriptive parameters from the spectrum calculated at point 250).

[0053] Preferably, the calculated parameters include one or more of the following parameters: —Average value of the spectrum; —The region corresponding to the spectrum at a defined frequency interval; —The spectral width at a predetermined intensity level (meaning the difference between the maximum and minimum frequencies), particularly at a level defined by an intensity value lower than the maximum value of the spectrum for the predetermined amount, particularly below 1 dB or 3 dB; —The frequency value corresponding to the maximum value of the spectrum; —The slope of the line that interpolates multiple points of the spectrum at predetermined frequency intervals; —The coefficients of a polynomial interpolated at frequency intervals containing the maximum value of the spectrum.

[0054] 270) is used as a function of the value extracted at point 230) and the descriptive parameter calculated at point 260) to calculate diagnostic parameters indicating the presence and type of tumor (renal cell carcinoma, benign tumor).

[0055] It is important to note that the diagnostic parameters calculated at point 270, in their various calculation modes, indicate the nonlinearity introduced into the ultrasound spectrum reflected by the cellular structure of the scanned region and its possible distortion.

[0056] It should also be noted that the calculation is possible only because the following is true: when the probe is in the same acquisition position, both the ultrasound signal reflected when emitting a broadband signal (which allows for the acquisition of significant ultrasound images and the individualization of hypoechoic and hyperechoic regions on the same probe) and the ultrasound signal reflected when emitting a narrowband signal (to highlight the nonlinearity introduced by the cellular structure of the region of interest and its possible deformation) have been acquired.

[0057] Preferably, the diagnostic parameters indicating tumor type are obtained as a function of the following: —The longitudinal dimension of the suspected area; —The lateral dimensions of the suspected area; —The surface area of ​​the suspected area; —The ratio between the surface area and the length of the edge of the suspected area; —The intensity of the first harmonic of the spectrum obtained in response to the narrowband ultrasonic signal calculated at point 220 ( 1) and the intensity of the nominal frequency ( The ratio between n); —The intensity of the first subharmonic of the spectrum obtained in response to the narrowband ultrasonic signal calculated at point 220. sub1) and the intensity of the nominal frequency ( The ratio between n); —The parameters extracted at point 260.

[0058] Furthermore, preferably, the diagnostic parameters indicating the tumor type are calculated by comparing the parameters just described with similar ratios relating to ultrasound signals corresponding to the following: —Multiple areas were confirmed to be cancerous by subsequent histological examination; —Multiple regions associated with homogeneous, rationally healthy kidney tissue.

[0059] In the following sections, preferred embodiments of a method for individualizing masses that are undetectable by ultrasound imaging will be described in detail.

[0060] Biological tissues can assimilate into aggregates of irregularly deployed "scatterers," generating non-stationary signals. Therefore, the signal's spectral content is locally modified and requires a time-frequency representation (TFR), as the TFR represents the time intervals on the signal in which specific spectral components exist. Specifically, wavelet transform (WT) extends the signal by positioning wavelet functions in both time and frequency, offering the possibility of flexible decomposition that processes the signal on a frequency-based basis with an adjustable filter window, rather than with a sliding window of constant time length (as occurs in the case of FFT, which is effectively a TFR with fixed resolution).

[0061] In this specific case, to individualize masses that are undetectable by ultrasound imaging, as an example, a specific technique of wavelet analysis known in the current state of technology as DWPT (Discrete Wavelet Packet Transform) can be used. This technique is chosen to reduce computational costs and is characterized, as an example, by using the following parameter configuration (which represents a suitable trade-off between time resolution and frequency resolution): —A biorthogonal and symmetric “mother wavelet” function (example “Daubechies 16”); —Decomposition level: 3° —Level of extraction: 1°.

[0062] In the parameter configuration, it is assumed that the “raw” ultrasound signal is sampled at 40 MHz (40 MS / s), and DWPT decomposes the signal corresponding to each considered propagation line in 8 frequency bands covering the entire spectral content of the signal being examined: 8 bands with an amplitude of 2.5 MHz, each ranging from 0 to 20 MHz, which is half the sampling frequency used. In this way, for each “raw” ultrasound signal (corresponding to the determined propagation line), 8 sets of DWPT coefficients are calculated; each set corresponds to a specific frequency band, and in each set, the number of DWPT coefficients is equal to the number of sampling times considered (because decimation is equal to 1, i.e., no decimation is performed).

[0063] In this way, at each moment of the initial signal, the eight DWPT coefficients have been correlated (each associated with a specific one of the eight frequency bands), which can be identified as k. ixy ,in

[0064] —i identifies the frequency band, —x identifies the signal under consideration (i.e., the propagation line) and —y indicates the time being considered (which actually corresponds to a certain distance from the probe and a certain depth inside the tissue).

[0065] so: —i varies between 1 and the number of frequency bands “n” (e.g., 8, in the case of third-order wavelet decomposition); —x varies between 1 and the number of available propagation lines, which is typically consistent with or twice the number of piezoelectric transducers in the probe (e.g., 128 or 256). —y varies between 1 and the number of samples that make up the signal (a value determined by the acquisition depth set on the ultrasound equipment once the sampling frequency is known).

[0066] In other words, k ixy The number of coefficients is associated with each point (x, y) in the ultrasound image, and it is equal to the number of frequency bands.

[0067] Therefore, k in each region of the ultrasound image ixy The statistical distribution of the coefficients can be used to characterize the corresponding parts of the kidney tissue through the following steps.

[0068] In each step, reference is also made to the description of preferred, rather than limiting, embodiments, as examples.

[0069] 499) The “raw” ultrasonic broadband reflection signal received by each of the piezoelectric transducers is decomposed in multiple (n) frequency bands using DWPT wavelet transform, and multiple (n) DWPT coefficients (k) are calculated for each sampling time of the signal. ixy ), 500) Defines the moving analysis window in the ultrasound image plane. 510) Define the forward step of the movement analysis window for each of the two directions. 520) Using a moving window positioned in the first location of the ultrasound image: 521) Identify the DWPT coefficient (k) associated with each of the points contained within the moving window. ixy The value of ) is used to define multiple (n) sets of DWPT coefficients associated with a specific position of the window; 522) For each set of DWPT coefficients defined at point 521), calculate at least one statistical descriptor selected from the following: mean, median, modal value, minimum, maximum, standard deviation, skewness, kurtosis.

[0070] Preferably, for each set of DWPT coefficients, multiple “m” enlisted descriptors are calculated (e.g., 8 for each enlisted group). At each window location, “n x m” statistical descriptors can then be calculated (64 in the example case).

[0071] 530) The amount by which the window defined at point 500 advances along one of the two forward directions by the same amount as the corresponding step defined at point 510). 540) Repeat steps 520) to 530) and move in one or the other direction until the entire region of interest corresponding to the kidney has been covered by the moving acquisition window.

[0072] To be clear, it should be said again that regions of interest can be identified either manually or through automatic identification algorithms.

[0073] For each of the calculated “m x n” descriptors, a value is then obtained that is associated with each point in the ultrasound image contained within the region of interest (the point associated with the descriptor value is the center point of the moving window at the location for which the descriptor was calculated). Thus, “m x n” maps have been obtained, each associated with a specific descriptor.

[0074] 550) For each of the descriptors calculated at points 520) to 540), an image is obtained by associating the value of the descriptor with a point of the ultrasound image associated with the window for which the descriptor was calculated; 560) Analyze each of the images obtained at point 550) to individualize one or more non-homogeneous regions relative to the surrounding area. This individualization can be performed by an operator using a suitable graphical interface, or it can be performed automatically using a suitable segmentation algorithm.

[0075] 570) The individualized heterogeneity at point 560) will be compared with the corresponding portion of the ultrasound image. If the anatomical differences visible on the ultrasound image do not demonstrate heterogeneity (but rather no obvious specificity in the region), the corresponding portion of the tissue will be identified as a suspicious area to be classified. Thus, these heterogeneous or suspicious areas can be classified as relating to healthy tissue or renal cell carcinoma according to the methods described in steps 210) to 240).

[0076] By using appropriate color mapping superimposed on a regular grayscale B-mode ultrasound image, the expansion and classification of these regions can be superimposed on the ultrasound image.

[0077] It should be noted that the method just described is only allowed to be implemented when retrieving the following items from the same position: —Ultrasound images, —The original broadband radio frequency ultrasound signal —Raw narrowband radio frequency ultrasound signal According to another embodiment, in order to calculate the diagnostic parameters of point 270, the process can be as follows: 400) For each segment of the propagation line of the ultrasound signal contained within the kidney tissue: 401) The original ultrasound signal reflected from this segment is decomposed using the DWPT (Discrete Wavelet Packet Transform) method; 402) Calculate the DWPT coefficients associated with each band for each point in the considered segment, thereby defining a set of coefficients for each said point; 403) For each set of DWPT coefficients defined at point 402), calculate at least one statistical descriptor selected from the following: mean, median, modal value, minimum, maximum, standard deviation, skewness, kurtosis; 410) Define a series of parameters relating to a common segment of the propagation line of ultrasound signals contained within the kidney tissue, including: —At least one of the statistical descriptors calculated at point 522 and associated with the segment of the examined propagation line; —Intensity of the first harmonic of the original ultrasound signal reflected by narrowband radio frequency ( 1) and the intensity of the nominal frequency ( The ratio between n); —Intensity of the first subharmonic of the original ultrasound signal reflected by narrowband radio frequency ( sub1) and the intensity of the nominal frequency ( The ratio between n); —The average value of the spectrum of the original ultrasound signal reflected by broadband radio frequency; —The region to which the spectrum of the original ultrasound signal is reflected by broadband at defined frequency intervals and / or defined amplitude intervals; —The spectral width at a predetermined intensity level (meaning the difference between the maximum and minimum frequencies), particularly at a level defined by an intensity value lower than the maximum value of the spectrum for the predetermined amount, particularly below 1 dB or 3 dB; —The frequency value corresponding to the maximum value of the spectrum of the broadband reflected original ultrasound signal; —The slope of the line interpolating multiple points of the spectrum of the original ultrasonic signal at predetermined frequency intervals; —The coefficients of the polynomial of the points of the spectrum of the original ultrasound signal that are broadband reflected are interpolated at frequency intervals containing the maximum value of the spectrum.

[0078] 420) A classification neural network was trained using a dataset containing parameters defined at point 400, which included multiple regions for which cancer had been confirmed by subsequent histological examination and multiple regions for which homogeneous, reasonably healthy kidney tissue was associated.

[0079] 430) Present the trained network with the dataset associated with each non-homogeneous region individualized at point 560).

[0080] 440) is a function used as the network output to classify non-homogeneous regions as renal cell carcinoma or as healthy tissue.

[0081] In another embodiment, the convolutional neural network can be trained using multiple images obtained at point 550, as well as corresponding ultrasound images associated with a patient whose cancer outcome has been confirmed by subsequent histological examination and multiple images associated with homogeneous, reasonably healthy kidney tissue.

Claims

1. An ultrasound device used for performing kidney diagnostic examinations to diagnose renal cell carcinoma, including: —At least one ultrasonic probe, including multiple piezoelectric or CMUT transducers, —A control device configured to drive the probe. —A device for acquiring and storing ultrasound signals reflected from tissue. The ultrasound device is configured to perform the following acquisition process: 100) Firing at least one broadband ultrasound pulse beam toward the patient's kidney; 110) Receive ultrasound signals reflected by kidney tissue in response to the pulse beam transmitted at step 100); 120) Process the signal stored in step 110) to obtain a kidney B-mode ultrasound image. The process is characterized in that, at step 110), it further involves storing the raw ultrasound signal received by the probe prior to the processing step of obtaining the B-mode ultrasound image. And the process further includes the following steps: 130) While keeping the probe in the same position, emit a narrow-band ultrasound pulse beam toward the patient's kidney; 140) Receive and store the raw ultrasound signal reflected by the kidney tissue in response to the narrow-band ultrasound pulse beam transmitted in step 130). A further feature is that the ultrasound device also includes a computing device on which a suitable computer program is loaded, the computing device being configured to perform a method for calculating diagnostic parameters, the calculation method comprising the following steps: 200) By analyzing the statistical distribution of the coefficients of the associated wavelet transform, the original ultrasound signal reflected in response to the broadband ultrasound pulse beam, stored at step 110), is analyzed to individualize one or more suspicious regions. For each suspicious area: 210) Extract a signal associated with at least one segment contained in the suspicious region from the original ultrasound signal reflected by the kidney tissue in response to a narrow-band ultrasound pulse beam stored at step 140); 220) Perform frequency transformation of the signal extracted in step 210) to at least calculate the spectrum obtained in response to the narrowband ultrasound pulse; 230) Extract values ​​related to multiple important frequencies from the spectrum calculated in step 220); 240) Extract a signal associated with at least one segment contained in the suspicious region from the original ultrasound signal reflected by the kidney tissue in response to the broadband signal stored at step 110); 250) Calculate the frequency transformation of the signal extracted at step 240) to obtain at least one spectrum in response to the at least one broadband ultrasonic pulse; 260) Calculate one or more descriptive parameters of the spectrum calculated in step 250). 270) Calculate diagnostic parameters indicating the presence and type of tumor as a function of the value extracted at step 230) and the descriptive parameter calculated at step 260).

2. The ultrasonic device according to claim 1, characterized in that, The important frequencies mentioned in step 230) include: —The nominal frequency of the narrowband ultrasonic pulses transmitted; —The first harmonic of the nominal frequency; —The first subharmonic of the nominal frequency.

3. The ultrasonic device according to claim 1, characterized in that, The diagnostic parameters calculated at step 270) are obtained as a function of one or more of the following parameters: —The longitudinal dimension of the suspected area; —The lateral dimensions of the suspected area; —The surface area of ​​the suspected area; —The ratio between the surface area and the length of the edge of the suspected area; —The intensity of the first harmonic of the spectrum obtained in response to the narrowband ultrasonic signal calculated at step 220) 1) and the intensity of the nominal frequency ( The ratio between n); —The intensity of the first subharmonic of the spectrum obtained in response to the narrowband ultrasonic signal calculated at step 220) sub1) and the intensity of the nominal frequency ( The ratio between n); —The one or more descriptive parameters extracted at step 260).

4. The ultrasonic device according to claim 3, characterized in that, The diagnostic parameters are calculated by comparing them with corresponding parameters in the ultrasound signals of the following: —Multiple areas were confirmed to be cancerous by subsequent histological examination; —Multiple regions associated with homogeneous, rationally healthy kidney tissue.

5. The ultrasonic device according to any one of the preceding claims, characterized in that, The probe has a nominal frequency of 3.5 MHz, an effective bandwidth for broadband acquisition between 0 and 7 MHz, and a frequency of 3 MHz for transmitting narrowband pulses.

6. The ultrasonic device according to any one of claims 1 to 4, characterized in that, The suspicious area is a hypoechoic or hyperechoic area relative to the surrounding renal parenchyma.

7. The ultrasonic device according to claim 6, characterized in that, The individualization of the suspicious region is obtained by means of the following steps: 499) The original ultrasonic broadband reflected signal received by each of the piezoelectric or CMUT transducers is decomposed in multiple frequency bands by means of DWPT wavelet transform, and multiple DWPT coefficients (kixy) are calculated for each sampling time of the signal. 500) Defines the moving analysis window in the ultrasound image plane. 510) Define the forward step of the movement analysis window for each of the two directions. 520) Using the moving analysis window positioned at the first location on the ultrasound image: 521) Identify the value of the DWPT coefficient (kixy) associated with each of the points contained within the motion analysis window, thereby defining multiple sets of DWPT coefficients associated with a specific location of the window; 522) For each set of DWPT coefficients defined in step 521), calculate at least one statistical descriptor selected from the following: mean, median, modal value, minimum, maximum, standard deviation, skewness, kurtosis. 530) Advance the window defined in step 500) along one of the two forward directions by an amount equal to the corresponding step defined in step 510). 540) Repeat steps 520) to 530) in one or the other direction until the entire region of interest corresponding to the kidney has been covered by the moving acquisition window; 550) For each of the descriptors calculated in steps 520) to 540), an image is obtained by associating the value of the descriptor with a point of the ultrasound image associated with the window for which the descriptor was calculated; 560) Analyze each of the images obtained in step 550) to individualize one or more non-homogeneous regions relative to the surrounding region.

8. The ultrasonic device according to any one of claims 1 to 4, characterized in that, The descriptive parameters calculated at step 260) include one or more of the following parameters: —Average value of the spectrum; —The region to which the spectrum corresponds at defined frequency intervals; — Spectral width at a predetermined intensity level; —The frequency corresponding to the maximum value of the spectrum; —The slope of the line that interpolates multiple points of the spectrum at predetermined frequency intervals; —The coefficients of a polynomial interpolated at frequency intervals containing the maximum value of the spectrum.

9. The ultrasonic device according to claim 7, characterized in that, The diagnostic parameters described in step 270) are calculated using the following steps: 410) Define a series of parameters relating to a common segment of the propagation line of ultrasound signals contained within kidney tissue, including: —At least one of the statistical descriptors calculated at step 522 and associated with the segment of the examined propagation line; —The intensity of the first harmonic of the original ultrasonic signal received by the probe ( 1) and the intensity of the nominal frequency ( The ratio between n); —The intensity of the first subharmonic of the original ultrasonic signal reflected in response to a narrowband pulse ( sub1) and the intensity of the nominal frequency ( The ratio between n); —The average value of the spectrum of the original ultrasound signal reflected in response to a broadband pulse; —The region to which the spectrum of the original ultrasound signal is reflected by broadband at defined frequency intervals and / or defined amplitude intervals; — Spectral width at a predetermined intensity level; —The frequency corresponding to the maximum value of the spectrum of the broadband reflected original ultrasound signal; —The slope of the line that interpolates multiple points of the spectrum of the original ultrasonic signal at predetermined frequency intervals by broadband reflection. —The coefficients of the polynomial of the points of the spectrum of the original broadband reflected ultrasound signal are interpolated at frequency intervals that include the maximum value of the spectrum. 420) The neural network is trained using a dataset containing the parameters defined in step 410) that relates to multiple regions that have been confirmed as renal cell carcinoma by subsequent histological examination and to multiple regions that are related to homogeneous, reasonably healthy kidney tissue. 430) Present the neural network with a dataset associated with each non-homogeneous region individualized in step 560), 440) is a function used as the network output to classify non-homogeneous regions as renal cell carcinoma or as healthy tissue.

Images