SYSTEMS AND ASSOCIATED METHODS FOR ASSISTING THE DIAGNOSIS OF ADVANCED PROSTATE CANCER - Patent application

Abstract

The present invention relates to a system and method for computer-aided detection (CAD) of aggressive prostate cancer. Automated diagnosis of aggressive prostate cancer is difficult, and methods that work on some MRI scanners do not work well when performed on images from a different scanner. By using machine learning techniques to study a database containing MRI data and biopsy results of 265 patients taken with four different types of MRI scanners, the inventors established a method to automatically determine the presence of aggressive cancer, and when the method was performed on another database containing MRI data of 270 patients taken with a different MRI scanner than the first database, it showed high sensitivity and specificity. The method involves providing the MRI images to a system that calculates a score for a portion of the prostate gland and determining that portion contains aggressive cancer based on whether a criterion dependent on the score is verified.

Description

[Technical field]

[0001] The present invention relates to a system for aiding in the diagnosis of aggressive prostate cancer and a method for aiding in the diagnosis of aggressive prostate cancer.

[0002] With over one million new cases each year, prostate cancer is the most common cancer in men worldwide and the second leading cause of cancer-related deaths. Diagnosis relies on ultrasound-guided biopsy. However, tumors are often not visible on ultrasound, and ultrasound's role is limited to guiding the biopsy needle into the prostate.

[0003] Typically, doctors perform 12 systematic biopsies. It is not uncommon for clinically significant tumors to be missed during biopsy, resulting in patients not receiving appropriate treatment.

[0004] In other patients, the biopsy needle may mistakenly enter an isolated site of cancerous cells when the actual tumor is small and mild and does not pose a threat to the patient's health, resulting in such patients undergoing invasive treatments with potentially serious side effects such as incontinence and impotence, but without any medical benefit.

[0005] To improve the targeting of biopsies, several image processing methods have been proposed that attempt to recognize signs of advanced cancer in MRI images, such as the papers by Au Hoang Dinh et al., “Quantitative Analysis of Prostate Multiparametric MR Images for Detection of Advanced Prostate Cancer in the Peripheral Region: A Multi-Imager Study,” published in Radiology, Vol. 280, No. 1, p. 117, 2016, and “Characterization of Prostate Cancer with Gleason Score 7 or Higher Using Quantitative Multiparametric MR Imaging: Validation of a Computer-Aided Diagnosis System in Patients Applied for Prostate Biopsy,” published in Radiology, Vol. 287, No. 2, p. 525, 2018.

[0006] However, the reliability of known methods for detecting advanced cancer remains lower than that performed by experienced radiologists. In addition, these existing methods are usually developed on the basis of MR images acquired with a limited number of MR scanners, and their reliability drops significantly when they are used to detect cancer on MR images acquired with a different type of MR scanner than that used on the original basis.

[0007] Thus, there is a need for a system that can help radiologists diagnose advanced prostate cancer, that is more robust than existing systems, yet can reliably detect advanced prostate cancer, particularly across MR images acquired using different types of MR scanners. Summary of the Invention

[0008] overview In this respect, the present specification relates to a system for assisting in the diagnosis of aggressive prostate cancer as claimed in claim 1.

[0009] According to a particular embodiment, the system is according to any one of claims 2 to 11.

[0010] The present specification also relates to a method for aiding in the diagnosis of aggressive prostate cancer as claimed in claim 12.

[0011] The present specification also relates to a computer program product as defined in claim 13.

[0012] The present description also relates to an information medium according to claim 14.

[0013] The present specification further relates to a method for diagnosing aggressive prostate cancer, comprising steps for carrying out the method of claim 12. [Brief description of the drawings]

[0014] The characteristics and advantages of the invention will appear from reading the following specification, given by way of non-limiting example only and made with reference to the associated drawings, in which: [Figure 1] 1 is a schematic diagram of a system for aiding in the diagnosis of aggressive prostate cancer. [Diagram 2] 2 is a flowchart of steps of an example method for aiding in the diagnosis of aggressive prostate cancer performed by the system of FIG. 1. [Diagram 3] 2 is a graph showing an exemplary specificity of the system according to FIG. 1 as a function of several parameters; [Figure 4] 2 is a graph showing another example of specificity of the system according to FIG. 1 as a function of several parameters; [Diagram 5] 2 is a graph showing yet another example specificity of the system according to FIG. 1 as a function of several parameters; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] A schematic diagram of a system 10 for aiding in the diagnosis of aggressive prostate cancer is shown in FIG.

[0016] The system 10 includes a control module 15 , a human-machine interface 20 , and a computer program product 22 .

[0017] More generally, system 10 is an electronic computer that can process and / or convert data depicted as electronic or physical quantities in registers and / or memories of system 10 into other like data that corresponds to the physical data in the memory, registers, or other type of display, and transmit or store it in a storage device.

[0018] The system 10 is configured to perform a method for aiding in the diagnosis of aggressive prostate cancer.

[0019] The control module 15 is configured to receive at least one magnetic resonance image of the prostate of a subject 30 from a separate device 25 , such as a magnetic resonance imaging scanner 25 .

[0020] The control module 15 comprises, for example, a data processing unit, such as a processor 35 , a memory 40 and a reader 45 for a readable information medium 50 .

[0021] The computer program product 22 includes an information medium 50 .

[0022] The readable information medium 50 is a medium that is readable by the data processing unit 35. The readable information medium 50 is a medium suitable for storing electronic instructions in a memory and is capable of being coupled to a bus of a computer system.

[0023] As an example, the readable information medium 50 is an optical disk, a CD-ROM, a magneto-optical disk, a ROM memory, a RAM memory, an EPROM memory, an EEPROM memory, a magnetic card, or an optical card.

[0024] The readable information medium 50 has a program stored in memory, the program including program instructions.

[0025] The computer program can be loaded onto the data processing unit 35 and is adapted to apply the method for aiding in the diagnosis of aggressive prostate cancer when the computer program is applied on the processor 35.

[0026] The human-machine interface 20 is configured to enable the transmission of information between a human, such as a physician, operating the system 10 and the control module 15 .

[0027] The human machine interface 20 comprises a display screen, for example a touch screen. Optionally, the human machine interface 20 further comprises a mouse and / or a keyboard and / or a speaker and / or a microphone.

[0028] The scanner 25 may, for example, be separate from the system 10, in particular located in a separate room or building, and connected to the system 10 by a wired or wireless network.

[0029] In a possible variant, the scanner 25 is part of the system 10. For example, the control module may instruct the scanner 25 to acquire a magnetic resonance image(s).

[0030] In another variant, the separate device 25 is a server or database in which magnetic resonance image(s) are stored, which in this case are acquired, for example, by one or more scanners and transmitted to the separate device 25 via a network.

[0031] The scanner is a multi-parametric magnetic resonance imaging ("MRI") scanner and is capable of obtaining images of the subject's organs using different MRI techniques, in particular by measuring T1-weighted and T2-weighted MRI signals emanating from the organs, as described below.

[0032] The operation of the system 10 in interaction with a computer program product will now be described with reference to FIG. 2, which illustrates an exemplary application of a method for assisting in the diagnosis of aggressive prostate cancer.

[0033] The method for aiding in the diagnosis of aggressive prostate cancer includes an acquiring step 100 , a transmitting step 110 , a selecting step 120 , a calculating step 130 , a determining step 140 , and a signaling step 150 .

[0034] During an acquisition step 100 , at least one magnetic resonance image, for example a set of magnetic resonance images, is acquired by the scanner 25 .

[0035] Each image is of the subject's prostate.

[0036] The subject may be, for example, a patient exhibiting symptoms likely due to prostate cancer. In a variant, the subject does not exhibit any symptoms.

[0037] Each image is an image of at least a portion of the subject's prostate peripheral zone ("PZ"). In particular, each image is an image of at least a portion of the PZ and at least a portion of the transition region ("TZ") of the prostate.

[0038] The set of magnetic resonance image(s) includes at least one first image and optionally one second image or a plurality of second images.

[0039] Each image contains a set of image elements or "pixels."

[0040] The or each first image may, for example, be an apparent diffusion coefficient mapping. Such a mapping is a two-dimensional representation of the apparent diffusion coefficient values ​​of the imaged region of the prostate. In such a mapping, each pixel contains the apparent diffusion coefficient value of a corresponding region of the prostate.

[0041] The apparent diffusion coefficient is a measure of the magnitude of diffusion of water molecules in tissue and can be obtained by diffusion weighted imaging (DWI). The apparent diffusion coefficient value is calculated, in a manner known per se, from a first T2*-weighted image of the prostate and a set of at least two diffusion weighted images, each corresponding to a spatial direction and being a mapping of the T2*-signal attenuated as a function of how easily water diffuses in the corresponding direction.

[0042] The apparent diffusion coefficient is in millimeters per second (mm2 / s).

[0043] The at least one second image is, for example, a T1-weighted image including a set of pixels, each pixel associated with an intensity value of a T1-weighted signal originating from a corresponding region of the prostate, and in particular, a set of successive T1-weighted images are acquired during a predetermined time period.

[0044] The predetermined time is the time it takes for the contrast agent to enter or leave the prostate, in other words the volumetric concentration of the contrast agent monotonically increases or decreases during that predetermined time.

[0045] For example, a contrast agent is injected into a subject, and the concentration of the contrast agent increases in the prostate, particularly in each of the first or second regions, until it reaches a maximum value, and then decreases as the contrast agent is discharged from the prostate. The T1-weighted signal intensity of each region of the prostate is a function of the contrast agent concentration, and increases significantly as a function of the contrast agent concentration, but this increase is non-linear. Thus, the T1-weighted signal intensity follows a trend of increasing and then decreasing over time, similar to the contrast agent concentration.

[0046] The images are acquired for a predetermined time after injection during which the contrast agent enters the prostate gland. In a variation, the images are acquired for a predetermined time following the moment when the contrast agent in the prostate gland reaches a maximum concentration during which the contrast agent leaves the prostate gland after the maximum concentration of the contrast agent has been reached.

[0047] The contrast agent is, for example, gadolinium.

[0048] During a transmission step 110 , all the first or second images are transmitted to the control module 15 .

[0049] For example, every first or second image captured by the scanner 25 is immediately transmitted to the control module 15 via the network.

[0050] In a possible variant, the first or second image is transmitted by the scanner 25 to a database or server, stored thereon for some time and transmitted after the storage time has elapsed to the control module 15, for example when a physician decides to check whether the subject's prostate contains aggressive cancer. In this case, the method for assisting in the diagnosis of aggressive prostate cancer can be considered not to include the acquisition step 100.

[0051] During a selection step 120, at least a first portion of the PZ is selected.

[0052] For example, at least one image, e.g., a first image(s), is presented to a physician on the human-machine interface 20, and the physician delineates a portion of the first image that the physician considers to be of interest, in particular a portion of the PZ that, from the perspective of the image, the physician considers may contain an aggressive cancer.

[0053] Optionally, a set of different first portions of the PZ is selected, for example by the physician, by delineating a number of regions in the first image.

[0054] In a possible variant, an image, such as the first image or one of the first images, is automatically divided into multiple parts by control module 15, for example by overlaying a square grid on the image, and control module 15 considers each area of ​​PZ separated from one another by the grid to be one of such first parts.

[0055] Optionally, at least one second portion of the prostate is selected, each second portion being a portion of the TZ.

[0056] Each second portion may, for example, be contoured by a physician or automatically selected in the same manner as the first portion(s).

[0057] During a calculation step 130, the control module 15 calculates a first score P based on the at least first image(s) received. Optionally, as described in more detail below, the control module further calculates a second score Y from the at least first image(s) received.

[0058] In particular, a first score P is calculated for each selected first portion and / or a second score Y is calculated for each selected second portion. It is calculated.

[0059] Each first score P is a function of at least one of a first quantity x1, a second quantity x2, and a third quantity x3.

[0060] A first amount x1 is calculated for the first portion considered according to the following formula:

number

[0061] The normalized wash-in rate for each region is calculated in a known manner by generating a curve of the T1-weighted signal intensity of the region as a function of time from the intensity values ​​of the pixels corresponding to the region in successive second images, and estimating the slope of that curve.

[0062] The normalized wash-in or wash-out rate is, in particular, estimated from a normalized T1-weighted signal, where the intensity of the T1-weighted signal of a pixel at a given time is divided by the intensity of the T1-weighted signal of the same pixel before the contrast agent enters the prostate, e.g., before the contrast agent is injected or before the contrast agent reaches the prostate (i.e., before the intensity increase begins).

[0063] It should be noted that different normalization methods can be used and are equivalent to each other, since the corresponding values ​​of parameter a1 and wash-in rates or cash-our rates corresponding to one method of normalization can be estimated from the values ​​corresponding to another normalization.

[0064] In a possible variant, instead of dividing all values ​​of the T1-weighted signal by the value of this signal before contrast injection, the wash-in or wash-out rate is calculated from the gradient expressed as a % of the original value.

[0065] In such a modified example, a1 is mm 2 Instead of expressing it in mm 2 / percentage, so the value of a1 shown in 4 is mm 2 This is 100 times smaller than when expressed as

[0066] It should be considered that in carrying out the present invention, all types of normalization of the wash-in or wash-out rates can be used indiscriminately, with corresponding adaptations to the value of a1 and the units used to express the wash-in or wash-out rates.

[0067] The slope of the increasing part of the curve, in particular the maximum slope (corresponding to the gradual increase in the contrast agent concentration in the considered region), is the normalized wash-in rate of that region. In a possible variant, the normalized wash-in rate of a region is the average slope of the increasing part of the curve.

[0068] A wash-in or wash-out rate is referred to as a "normalized" wash-in or "normalized" wash-out rate when the intensity curve is constructed with normalized intensity values, i.e., the intensity value of the T1-weighted signal of a pixel or of a region of a second image at a different time, divided by the intensity of the T1-weighted signal of the same pixel or region before the contrast agent was injected. Such normalization allows easy comparison of wash-in or wash-out rates obtained from different MRI devices.

[0069] It should be noted that other normalization methods can also be used, as will be described below.

[0070] The slope, such as the average slope or maximum slope, of the decreasing part of the curve (corresponding to the gradual decrease of the contrast agent concentration in the considered region after it has reached a maximum) is the normalized washout rate of that region. In practice, the normalized washout rate is often the average slope.

[0071] When the parameter W is a normalized wash-in rate, the parameter W is, for example, the arithmetic mean of the normalized wash-in rates of the area corresponding to each pixel of the first portion.

[0072] When the parameter W is a normalized washout rate, the parameter W is, for example, the arithmetic mean of the normalized washout rates of the areas corresponding to each pixel of the first portion.

[0073] In a variant, the normalized wash-in or wash-out rate is calculated for each first portion of each second image by calculating the average value of the T1-weighted signal of all pixels in the first portion and generating a normalized curve of the average values ​​of the first portion calculated as a function of time. The generated time curve has the same overall shape as that of each pixel, since the average value of the contrast agent concentration in the first portion considered increases and then decreases with time. In this case, the wash-in rate value is the slope, in particular the maximum slope or the average slope, of the increasing portion of the generated time curve, and the normalized wash-out rate is the slope, for example the maximum slope or the average slope, of the decreasing portion of the curve.

[0074] It should be noted that the normalized wash-in or wash-out rates can also be calculated for the second portion(s) in a similar manner as above.

[0075] The parameter ADC is calculated by the control module 15 by ordering in increasing order the apparent diffusion coefficient values ​​of the pixels corresponding to the first portion considered, the parameter ADC being the percentile of the apparent diffusion coefficient values ​​of the first portion.

[0076] By "percentile" it is meant that the parameter ADC is the value of the apparent diffusion coefficient value that appears at a given position within a set of ordered apparent diffusion coefficient values.

[0077] For example, the tenth percentile of the apparent diffusion coefficient values ​​is the apparent diffusion coefficient value such that 90 percent of the apparent diffusion coefficient values ​​are strictly above the tenth percentile and 10 percent of the diffusion coefficient values ​​are less than or equal to the tenth percentile.

[0078] The second percentile is the value at which 98 percent of the apparent diffusion coefficient values ​​are strictly greater than the second percentile and 2 percent of the diffusion coefficient values ​​are inferior to or equal to the second percentile.

[0079] It should be noted that if the number of pixels corresponding to the portion considered is not divisible by 100, the percentile can be calculated by interpolation between the two pixels that fall within that percentile.

[0080] Unless otherwise specified, a value "included between" two limits corresponds to a range that includes both limits.

[0081] ADC values ​​are often expressed in millimeters per second (mm 2 / s).

[0082] If W is the normalized wash-in rate, the first constant a1 can be, for example, −6×10 -3 mm 2 and -10 -3 mm 2 In some embodiments, the first constant a1 is between −5×10 -3 mm 2 and -1.5×10 -3 mm 2 For example, between -5×10 -3 mm 2 and -3×10 -3 mm 2 is included between.

[0083] In this specification, the sign "x" between two numbers indicates multiplication.

[0084] The parameter ADC is in particular the 40th or lower percentile, in particular the 30th or lower percentile, in particular the 10th or lower percentile, where W is the normalized wash-in rate value.

[0085] If W is the normalized washout rate, then the first constant a1 can be, for example, 2 x 10 -2 mm 2 and 15×10 -2 mm 2 is included between.

[0086] The parameter ADC, for example, indicates that the percentile is less than or equal to 15, where W is the normalized washout rate value.

[0087] The second amount x2 is calculated for the first portion considered according to the following formula:

number

[0088] The second constant b1 is, for example, 10 -7 mm 2 / square second and 14×10 -7 mm 2 / square second (mm 2 / s 2 ) in some cases, the second constant b1 is between 10 -7 mm 2 / s 2 and 9×10 -7 mm 2 / s 2 is included between.

[0089] The time-to-peak value is the time between a first moment and a second moment on the curve of the value of the T1-weighted signal as a function of time. The first moment is the time when the value of the T1-weighted signal starts to increase after the injection of contrast agent into the subject (i.e. the moment when the injected contrast agent starts to enter the first portion under consideration). The second moment is the time of maximum concentration of contrast agent in the first portion and corresponds to the maximum intensity of the T1-weighted signal.

[0090] The time to peak value can be obtained, for example, by calculating the average value of the T1-weighted signals of all pixels for each second image, generating a curve of the average value as a function of time, and measuring the time to peak value on the curve.

[0091] In a variant, the time-to-peak value is the average value of the time-to-peak values ​​calculated for each pixel corresponding to the first portion.

[0092] The third amount, x3, is calculated for the corresponding first portion according to the following formula:

number

[0093] The third constant c0 is for example comprised between 2.41 and 6.44, in particular between 3.5 and 5.0. In particular, the third constant c0 is equal to 4.2.

[0094] The fourth constant c1 is -7570 sec / mm 2 and -4020 sec / mm 2 Especially during -7150 sec / mm 2 and -5270 sec / mm 2 is included in between.

[0095] A first score P for each first portion is calculated from at least one of the first quantity x1, the second quantity x2, and the third quantity x3.

[0096] The first score P may, for example, be equal to one of a first amount x1, a second amount x2, and a third amount x3.

[0097] The first score P is, in a possible variant, calculated using logistic regression.

[0098] In statistics, logistic regression models are used to model the probability that an event will occur, for example, the probability that a first fraction will include aggressive cancer.

[0099] The first score P is calculated, for example, using logistic regression according to the following formula:

number

number

[0100] In this case, the score P is the probability that the first fraction considered contains aggressive cancer.

[0101] For example, in our data, the scores are

number

[0102] In another variant, the first score P is obtained by calculating two or three probabilities, one for each of the first quantity x1, the second quantity x2, and the third quantity x3, using logistic regression according to equation 4, and by summing the three calculated probabilities.

[0103] It should be noted that in general, many types of scores P are possible, provided there is a bijective relationship between the calculated score P and either the first quantity x1, the second quantity x2, and the third quantity x3, or any combination of these quantities (such as a sum or weighted sum).

[0104] For example, the score P may be obtained by dividing or multiplying one of the first quantity x1, the second quantity x2, and the third quantity x3 by a constant, which may in some cases result in a dimensionless quantity x, which is optionally input into Equation 5 to calculate the score P.

[0105] As mentioned above, the units used to express the parameters ADC, WI, WO, and TTP involved in the calculation of the first amount x1, the second amount x2, and the third amount x3 can also be changed and the parameters a and b1 can be adapted accordingly.

[0106] Optionally, during the calculation step 130, the control module 15 further calculates a second score Y for the, or each, second portion of TZ.

[0107] The second score Y is calculated as a function of a fourth quantity y, where the fourth quantity y is dimensionless.

[0108] The fourth quantity, y, is calculated as a function of the following equation:

number

[0109] The fifth constant d0 is contained between 2.25 and 13.48, and in particular is equal to 7.9.

[0110] The sixth constant d1 is -23820s / mm 2 and -4490s / mm 2 In particular, the sixth constant d1 is equal to -10740.

[0111] The second score Y is calculated, for example, using the following formula:

number

[0112] However, as with the first score P, many different methods can be used to calculate the second score Y from the fourth quantity y. For example, the second score Y is equal to the fourth quantity y.

[0113] During a decision step 140, the control module 15 determines, for each first portion, whether the first portion of the PZ being considered contains an aggressive cancer or not.

[0114] In particular, the control module 15 determines whether a first criterion based on the first score P is verified, and determines that the first portion contains aggressive cancer if the criterion is verified.

[0115] According to one embodiment, the criterion is the fact that the first score P is less than or equal to a first threshold, in which case the control module compares the first score P to the first threshold and determines that the first portion contains aggressive cancer if the first score P is less than or equal to the first threshold, or that the first portion does not contain aggressive cancer if the first score P is strictly above the first threshold.

[0116] This is evident when the first score P is equal to one of the first amount x1, the second amount x2, and the third amount x3. In this case, the first threshold is referred to as, for example, a "score threshold."

[0117] The score threshold may be, for example, −0.8×10 when the first score P is equal to the first amount x1. -3 mm 2 / s and 2.3 × 10 -3 mm 2 / s, where W is the normalized wash-in rate. In particular, the score threshold is 0.55 × 10 -3 mm 2 / s and 0.8 × 10 -3 mm 2 Included between / s.

[0118] If W is the normalized washout rate, then the score threshold is, for example, -1.1 x 10 -3 and 3.2 x 10 -3 mm 2 / s. In particular, the score threshold is 0.85 × 10 -3 mm 2 / s and 1.1 × 10 -3 mm 2 Included between / s.

[0119] If the first score P is equal to the second amount x2, then the score threshold may be, for example, -1.1 x 10 -3 mm 2 / s and 3.3 × 10 -3 mm 2 / s, especially 0.9×10 -3 mm 2 / s and 1.1 × 10 -3 mm 2 Included between / s.

[0120] The score threshold is, for example, equal to 0.40 when the score P is equal to the third amount x3.

[0121] It should be noted that other criteria may also be considered.

[0122] If the first score P is calculated using Equation 4 and Equation 5, the criterion is the fact that the first score P is equal to or greater than a first threshold, in which case the control module compares the first score P to the first threshold and determines that the first portion contains aggressive cancer if the first score P is equal to or greater than the first threshold. If the first score P is strictly below the first threshold, the control module 15 determines that the first portion does not contain aggressive cancer. The first threshold is, for example, referred to as a "probability threshold."

[0123] During the decision step 140, the control module 15 further determines, for each second portion, whether the second portion of the TZ considered contains an aggressive cancer.

[0124] In particular, control module 15 checks whether a second criterion is verified and, if the criterion is verified, determines for each second portion that the considered second portion of the TZ contains aggressive cancer, and if the criterion is not verified, control module 15 determines that the second portion does not contain aggressive cancer.

[0125] According to one embodiment, the control module compares the second score Y to a second threshold and determines that the second portion contains aggressive cancer if the second score Y is equal to or less than the second threshold. In this case, the criterion is the fact that the second score Y is equal to or less than the second threshold. However, other criteria may be considered.

[0126] For example, control module 15 may calculate, in particular using Equation 6, the probability that the second portion includes aggressive cancer and determine that the second portion includes aggressive cancer if the probability is greater than or equal to a second probability threshold.

[0127] The second probability threshold is, for example, comprised between 0.70 and 0.77 when the value ADC used in Equation 6 is the 25th percentile of ADC.

[0128] If the control module 15 determines that one of the first and second portions (may be multiple) contains aggressive cancer, the control module 15 generates a message directed to a physician during a signaling step 150.

[0129] This message is intended to be transmitted by the human machine interface 20 to a physician.

[0130] The message informs the physician that at least one of the first portion or the second portion has been determined to contain aggressive cancer.

[0131] The message includes, for example, an indication of the aggressive cancer, or the first or second portion(s) containing aggressive cancer, such as one of the first or second images in which the aggressive cancer, or the first or second portion(s) containing aggressive cancer, is outlined in red thereon, or accompanied by a written message stating that the first or second portion under consideration is likely to be cancer.

[0132] This message may optionally be accompanied by a sound, such as an alarm signal, indicating a cancer finding.

[0133] It should be noted that this message can be in any format as long as the physician is able to understand the information.

[0134] The physician may then closely review the first or second portion indicated in the message, on the first or second image, or using other methods, to determine whether the physician shares the control module's assessment of the presence of aggressive cancer in the first or second portion.

[0135] Additionally, the method allows a physician reviewing the first and second images to have the analysis of the subject's prostate cross-checked by system 10, thereby enabling a diagnosis of cancer that may have been overlooked by the physician if the system directs the physician's attention to the possibility that cancer may be present in an area of ​​the prostate that may have been overlooked or where the cancer does not appear clearly.

[0136] In a possible variant, the signaling step 150 is not performed and the fact that the control module 15 has determined the presence of aggressive cancer is simply stored, for example in the memory 40 .

[0137] The method of assisting in the diagnosis of progressive prostate cancer may, for example, be performed as part of a method of diagnosing prostate cancer and include a diagnosis step in which prostate cancer is diagnosed as a function of a message(s) generated during the method of assisting in the diagnosis of progressive prostate cancer.

[0138] In particular, the method for diagnosing prostate cancer is part of a method for treating prostate cancer, which includes the step of treating prostate cancer diagnosed by carrying out the method for diagnosing prostate cancer.

[0139] Equations 1-3 and 5 were determined by the inventors in the course of a study using a training database containing MRI sequences from 265 patients who had been diagnosed with and treated for advanced prostate cancer and who consecutively underwent biopsy, multi-parametric MRI and radical prostatectomy.

[0140] Methods: T2-weighted, diffusion-weighted, and dynamic contrast-enhanced images were systematically recorded and prospectively reviewed by two independent radiologists. Regions suspicious for cancer were segmented in each sequence. A computer program extracted 23 quantitative parameters from these regions.

[0141] The training database contains MR images acquired with several different MR scanners of different manufacturers and types, using different magnetic field values. - Siemens Symphony scanner (1.5 Tesla, 63 patients, 233 suspicious lesions), - General Electric Medical Systems Discovery MR 750 scanner (3T, 124 patients, 272 lesions), - Achieve scanner manufactured by Philips medical systems (3T, 52 patients, 173 lesions). - Philips Healthcare Ingenia scanner (3T, 26 patients, 56 lesions)

[0142] The model was discovered by machine learning techniques, recursively browsing through different variables and testing their performance including them one by one. New variables were retained if their corresponding coefficient and the variance they explained were statistically significant in the model and had a correlation lower than 50% with each of the other variables in the model.

[0143] The threshold was determined using an intermediate database containing MR images of the prostate containing suspicious lesions whose benign or malignant nature had been confirmed by biopsy and was selected to give a sensitivity of 90%.

[0144] Like the training database, the intermediate database used MR data obtained from a GE MR750 scanner and a Philips Healthcare Ingenia scanner, with data from 101 and 11 patients, respectively.

[0145] From the selected parameters, Equations 1-3 and 5 were constructed and their performance was then evaluated on a test database containing data (MRI images and biopsies) from different patients than the training and intermediate databases. The training database contains data only from patients with confirmed prostate cancer, whereas the test database contains patients suspected of having prostate cancer but without confirmed cancer.

[0146] A method for aiding in the diagnosis of prostate cancer is performed on MRI images on a test database and includes two radiologists outlining a first portion and a second portion on the MRI images.

[0147] A total of 238 suspicious lesions were identified by radiologists in the data of 158 patients, of which biopsies identified 126 benign lesions (114 in the PZ and 12 in the TZ), 34 cancers with a Gleason score of 6 (30 in the PZ and 4 in the TZ), and 78 cancers with a Gleason score of 7 or higher (71 in the PZ and 7 in the TZ).

[0148] The MR images in the test database were acquired by four different MR scanners. - General Electric Medical Systems Discovery MR750 scanner (3T, 62 patients, 93 lesions), - Philips Healthcare Ingenia scanner (3T, 22 patients, 41 lesions), - GE medical systems' Optima MR450w (1.5T, 66 patients, 91 lesions), and - Philips Healthcare Ingenia (1.5T, 8 patients, 13 lesions).

[0149] The sensitivity and specificity of the systems using each of Equations 1, 2, 3, and 5 were estimated by comparing with the diagnostic results obtained from biopsies of the corresponding parts of the patient's prostate. The performance of the systems was measured from the specificity achieved when the sensitivity was set at 90% (referred to as "specificity sp90").

[0150] The accuracy of the system 10 using Equation 1, where W is the normalized wash-in rate, when used on a test database, consistently exhibits a specificity greater than 0.5 with an ADC of 40 or less percentile, and the first constant, a, is −6×10 -3 mm 2 and -10-3 mm 2 The specificity sp90 values ​​as a function of ADC and a1, where W is the normalized wash-in rate, for system 10 using Equation 1 in the test database are shown in FIG.

[0151] In Figures 3-5, contours are shown delimiting regions of specificity sp90 above a threshold, for example the line marked "0.55" includes all regions of data corresponding to sp90 above 0.55.

[0152] In particular, if ADC is less than or equal to 30 percentiles and the first constant a1 is -5×10 -3 mm 2 and -3×10 -3 mm 2 When the IL-10 expression level was between 0.01 and 0.25, the specificity was consistently above 0.55. ADC is the 10th percentile and the first constant a1 is -4.5×10 -3 mm 2 and -3×10 -3 mm 2 When the mean age was between 0.58 and 1.0, the specificity was substantially greater than or equal to 0.58.

[0153] On the test database, ADC is the second percentile and the first constant a1 is -3.96×10-5 mm 2 The best performance was obtained when the value of σ equals 0.01 and the specificity equals 0.606.

[0154] The accuracy of the system 10 using Equation 1 is the percentile below 15 ADC, where W is the normalized washout rate, and the second constant a1 is 2×10 -2 and 15×10 -2 mm 2 When used on a training database comprised between 0.01 and 0.25, it consistently demonstrated a specificity above 0.45. The specificity sp90 values ​​as a function of ADC and a1, where W is the normalized washout rate, within the training database of system 10 using Equation 1 are shown in Figure 4.

[0155] In particular, if ADC is a percentile between 2 and 10, and the first constant a1 is 4×10 -2 mm 2 and 12×10 -2 mm 2 When the mean age was between 0.5 and 1.0, the specificity was consistently greater than 0.50.

[0156] The best performance is achieved on the training database when ADC is in the second percentile and the first constant a1 is 10×10 -2 mm 2 , obtained when the specificity was equal to 0.549.

[0157] The accuracy of the system 10 using Equation 2 was measured on a test database where the ADC is a percentile between 20 and 45 and the second constant b1 is 10 -7 mm 2 and 14×10 -7 mm 2 When the mean was between 0.01 and 0.25, the specificity was consistently above 0.45.

[0158] The values ​​of specificity sp90 as a function of ADC and b1 in the test training database for system 10 using Equation 2 are shown in FIG.

[0159] In particular, if ADC is a percentile between 20 and 45, and the second constant b1 is 10 -7 and 9×10 -7 mm 2 When the mean age was between 0.5 and 1.0, the specificity was consistently greater than 0.50.

[0160] The best performance was achieved on the test database when ADC was in the 25th percentile and the second constant b1 was 5.1 x 10 -7 mm 2 ,Specificity was obtained when equal to 0.539.

[0161] Listed below are examples of the first threshold and determined constants c0 and c1 in different cases where system 10 uses Equation 4 and the quantity x is a third quantity x3.

[0162] a) In the case of the lowest percentile of ADC, the first threshold is between 0.18 and 0.37, the seventh constant c0 is between 2.41 and 3.61, and the eighth constant c1 is between -5880 s / mm 2 and -4020s / mm 2 (The combination that gives the best result is one in which the first threshold value is equal to 0.28, the seventh constant c0 is equal to 2.9, and the eighth constant c1 is equal to -4790 s / mm 2 (The value was equal to

[0163] b) If the ADC is in the second percentile, the first threshold is between 0.17 and 0.39, the seventh constant c0 is between 3.55 and 4.92, and the eighth constant c1 is between -7150 s / mm 2 and -5270s / mm 2 (The combination that gives the best result is one in which the first threshold value is equal to 0.27, the seventh constant c0 is equal to 4.17, and the eighth constant c1 is equal to -6080s / mm 2 (The value was equal to

[0164] c) If the ADC is in the ninth percentile, the first threshold is between 0.17 and 0.38, the seventh constant c0 is between 4.42 and 5.92, and the eighth constant c1 is between -7570 s / mm 2 and -5660s / mm 2 (The combination that gives the best result is one in which the first threshold value is equal to 0.27, the seventh constant c0 is equal to 5.09, and the eighth constant c1 is equal to -6500 s / mm 2 (The value was equal to

[0165] d) When the ADC is in the tenth percentile, the first threshold is between 0.17 and 0.38, the seventh constant c0 is between 4.47 and 5.96, and the eighth constant c1 is between -7560 s / mm 2 and -5650s / mm 2(The combination that gives the best result is one in which the first threshold value is equal to 0.26, the seventh constant c0 is equal to 5.13, and the eighth constant c1 is equal to -6490 s / mm 2 (The value was equal to

[0166] e) When the ADC is at the 25th percentile, the first threshold is between 0.14 and 0.36, the seventh constant c0 is between 4.59 and 6.44, and the eighth constant c1 is between -7430 s / mm 2 and -5260s / mm 2 (The combination that gives the best result is one in which the first threshold value is equal to 0.27, the seventh constant c0 is equal to 5.49, and the eighth constant c1 is equal to -6260 s / mm 2 (The value was equal to

[0167] Examples of the second threshold and determined constants d0 and d1 in different cases where the system 10 uses Equation 6 are listed below.

[0168] f) If the ADC is the second percentile, the second threshold is between 0.07 and 0.75, the ninth constant d0 is between 2.25 and 13.48, and the tenth constant d1 is between -23820 s / mm 2 and -4490s / mm 2 (The combination that gives the best result is between the second threshold value equal to 0.4, the ninth constant d0 equal to 6.5, and the tenth constant d1 equal to -11520 s / mm 2 (The value was equal to

[0169] g) If the ADC is the 9th percentile, the second threshold is between 0.07 and 0.71, the ninth constant d0 is between 3.83 and 12.5, and the tenth constant d1 is between -19500 s / mm 2 and -6280s / mm 2 (The combination that gives the best result is between the second threshold value equal to 0.38, the ninth constant d0 equal to 7.3, and the tenth constant d1 equal to -11320 s / mm 2 (The value was equal to

[0170] h) If the ADC is the 10th percentile, the second threshold is between 0.06 and 0.71, the ninth constant d0 is between 3.96 and 12.51, and the tenth constant d1 is between -19100 s / mm 2 and -6430s / mm 2 (The combination that gives the best result is one in which the second threshold is equal to 0.39, the ninth constant d0 is equal to 7.4, and the tenth constant d1 is equal to -11360 s / mm 2 (The value was equal to

[0171] i) When the ADC is the 25th percentile, the second threshold is between 0.14 and 0.64, the ninth constant d0 is between 4.9 and 11.19, and the tenth constant d1 is between -15030 s / mm 2 and -6910s / mm 2 (The combination that gives the best result is between the second threshold value equal to 0.4, the ninth constant d0 equal to 7.9, and the tenth constant d1 equal to -10740 s / mm 2 (The value was equal to

[0172] System 10 using Equation 4, where quantity x is the first quantity x1 and ADC is the second percentile, was found to have the following AUC and specificity (with the % sensitivity and specificity in parentheses when using contours drawn by one radiologist) for the following cases, when using contours drawn by one radiologist: - Case a: AUC between 76% and 86% (between 78% and 87%), specificity sp90 between 33% and 52% (between 28% and 63%), - case b: AUC between 77% and 86% (between 79% and 87%), specificity sp90 between 34% and 54% (between 29% and 66%), - case c: AUC between 78% and 88% (between 79% and 88%), specificity sp90 between 37% and 61% (between 32% and 74%), - Case d: AUC between 79% and 88% (between 79% and 88%), specificity sp90 between 37% and 62% (between 32% and 74%) - Case e: AUC was between 79% and 88% (between 80% and 89%), specificity sp90 was between 39% and 62% (between 38% and 72%).

[0173] "AUC" is the performance estimate from the area under the curve of the true positive rate as a function of the false positive rate. Such curves are well known in the art and are often referred to as "Receiver Operating Characteristic (ROC) curves."

[0174] System 10 using Equation 5 was found to have the following AUC and specificity (with the percentage of sensitivity and specificity in parentheses when using contours drawn by one radiologist) for the following cases, when using contours drawn by one radiologist: - Case f: AUC between 47% and 86% (between 35% and 83%), specificity sp90 between 0% and 80% (between 0% and 82%), - Case g: AUC between 46% and 88% (between 28% and 85%), specificity sp90 between 0% and 83% (between 0% and 82%), - case h: AUC between 46% and 89% (between 30% and 85%), specificity sp90 between 0% and 83% (between 0% and 82%), - Case i: AUC between 50% and 94% (between 29% and 82%), specificity sp90 between 0% and 83% (between 0% and 75%).

[0175] It should be noted that both quantities, x1 and x2, combine the centile of the ADC and a dynamic quantity in the MRI image (i.e., a quantity reflecting the change in value as a function of time while the contrast agent flows in or out of the prostate), i.e., the wash-in rate, wash-out rate, or time-to-peak value.

[0176] Thus, part of the invention is that the inventors have discovered ranges of values ​​a1, b1 and ADC percentiles that, when combined together, reflect the likelihood that cancer is present in a portion of the prostate, and that lead to efficient detection of cancer from images.

[0177] The computer-aided diagnostic methods described herein can be used to establish a clinical diagnosis based on the results of the diagnostic methods to aid in the diagnosis of aggressive prostate cancer.

[0178] In some embodiments, the methods of the invention are carried out in vitro or ex vivo.

[0179] To confirm the initial results of the method, additional validation studies were performed by the inventors. These results are presented below by way of example only.

[0180] In this example, evaluation was performed on another test dataset (named "external test dataset") that was independent of the test dataset described above.

[0181] The new database contains 104 patients and 126 lesions imaged with one scanner, the Signa Voyager scanner from GE medical systems (1.5T, 104 patients, 126 lesions).

[0182] These 104 patients were suspected of having cancer and underwent mpMRI and subsequent biopsy at other institutions.

[0183] As in the previous study, the PI-RADSv2 (Prostate Imaging-Reporting and Data System version 2) scores that were prospectively assigned to each lesion at the time of biopsy were retrieved from the patient's chart. Then, based on the mpMRI reports, two radiologists with 19 years of experience (OR, R1) and 5 years (PCM, R2), blinded to each other's biopsy findings, retrospectively contoured the target lesions at the slice levels that were most likely to be representative in T2-weighted, diffusion-weighted, and DCE images. The calculation of the lesion score was performed using the PZ or TZ model depending on its location. Thus, each lesion received a PI-RADSv2 score (prospectively assigned at the time of biopsy) and two CAD scores using the regions of interest (ROIs) contoured by R1 and R2. The PI-RADSv2 scores and the scores obtained by the present invention were then compared with the biopsy findings.

[0184] The table below details the distribution of PI-RADSv2 scores and scores derived from the present invention as a function of biopsy outcome.

[0185] [Table 1]

[0186] In the above table, the data indicates the number of patients. ISUP is the abbreviation for the International Society of Urological Pathology grade, which measures the severity of the cancer. CAD is the abbreviation for the computer-aided diagnosis system according to the present invention.

[0187] In the new test dataset, score AUCs using ROIs contoured in R1 and R2, respectively, were 82% (95% CI: 76-89) and 86% (95% CI: 79-93), which were not significantly different from the PI-RADSv2 scores assigned at biopsy (85%, 95% CI: 79-91, p=0.82 and p=1, respectively).

[0188] This new test faced challenges because all patients were imaged with a newer scanner different from the one used in the training dataset. Nevertheless, the system according to the invention provided robust results in terms of diagnostic thresholds (as quantified by sensitivity and specificity) as well as overall diagnostic performance (as quantified by AUC). Surprisingly, the threshold that showed 90% sensitivity in the pre-test dataset showed similar sensitivity in the internal test dataset (85%-89%) and the external test dataset (92%). At this level of sensitivity, the specificity of the computer-aided diagnosis system according to the invention was good (64%-76%), suggesting that it offers a better sensitivity / specificity trade-off than the classical thresholds of PI-RADSv2.

Claims

1. A system (10) for aiding in the diagnosis of aggressive prostate cancer, the system including a control module (15), the control module (15) comprising: receiving at least one magnetic resonance image of at least a first portion of a peripheral region of a prostate of a subject, said image comprising a set of pixels, said image comprising for each pixel an apparent diffusion coefficient value of a corresponding region of said prostate; - calculating a first score; and - determining that the first portion contains aggressive cancer if a first criterion based on the calculated first score is verified, or determining that the first portion does not contain aggressive cancer if the first criterion is not verified; and The first score is the quantity x １ and the amount x ２ is calculated as a function of at least one of [0090] where W is a normalized wash-in rate or normalized wash-out rate value of the first portion, the wash-in or wash-out value being expressed in Hertz and estimated from a set of magnetic resonance images of the first portion, ADC is a percentile of the apparent diffusion coefficient values ​​of the first portion, and mm 2 / s, １ is a first constant, and where W is the normalized wash-in rate value, the first constant is −6×10 －３ mm ２ and -10 －３ mm ２ and the percentile is less than or equal to 40, and W is a normalized washout rate value, the first constant is 2×10 －２ mm ２ and 15 x 10 －２ mm ２ and the percentile is less than or equal to 15; [0010] Here, b １ is 10 －７ mm ２ / s ２ and 14 x 10 －７ mm ２ / s ２ a second constant comprised between ADC and ADC+ADC, ADC being the percentile comprised between 20 and 45 of the apparent diffusion coefficient values ​​of all pixels corresponding to the first portion, TTP being a time to peak value of the first portion, expressed in seconds, the time to peak value being the time duration from a first instant to a second instant, the first instant being the arrival time of a contrast agent at the prostate and the second instant being the time of maximum contrast concentration of the MRI signal of the first portion.

2. W is the normalized wash-in rate, and the first constant a １ is -5 x 10 －３ mm ２ and -3 x 10 －３ mm ２ The system of claim 1 , wherein:

3. The system of claim 2 , wherein the percentile is comprised between 0 and 30.

4. The second constant b １ is 10 －７ mm ２ / square second and 9 x 10 －７ mm ２ 2. The system of claim 1, wherein the time is between 1 / 2 and 1 / 2 square seconds.

5. 2. The system of claim 1, wherein ADC is the 25th percentile of the apparent diffusion coefficient values ​​of all pixels corresponding to the first portion when the first score is calculated as a function of the time to peak value.

6. 2. The system of claim 1, wherein a wash-in or wash-out value is estimated for each pixel corresponding to the first portion, and W is an average value of the wash-in or wash-out values ​​estimated for each pixel.

7. 2. The system of claim 1, wherein a plurality of images of the first portion are acquired during a predetermined time, the predetermined time being a predetermined time for a contrast agent to enter the prostate or a predetermined time for a contrast agent to exit the prostate, each image including, for each pixel, an intensity value of a T1-weighted signal of the corresponding region of the prostate, and the control module is configured to calculate, for each image, an average value of normalized intensity values ​​of the pixels corresponding to the first portion, each normalized value being equal to a measured value divided by the value of the same pixel before the contrast agent entered the prostate, the normalized wash-in rate value being the upward slope of a curve of the calculated average values ​​as a function of time, and the normalized wash-out rate being the downward slope of the curve.

8. 2. The system of claim 1, wherein the control module is configured to compare the score to a threshold and determine that the first portion contains aggressive cancer if a first amount is less than or equal to the threshold.

9. The score is a first quantity x １ where W is the normalized wash-in rate, the threshold is −0.8×10 －３ Millimeters per second and 2.3 x 10 －３ or, if W is the normalized washout rate, the threshold is −1.1×10 －３ Millimeters per second and 3.2 x 10 －３ 9. The system of claim 8, wherein the axial speed is comprised between square millimeters per second.

10. The score is a second quantity x ２ and the threshold is equal to -1.1 x 10 －３ Millimeters per second and 3.3 x 10 －３ 9. The system of claim 8, wherein the axial speed is comprised between square millimeters per second.

11. Each image includes an image of a second portion of the transition region of the prostate, and the control module (15) further comprises: - calculating a second score Y as a function of a fourth quantity y, the fourth quantity y being according to the formula: ##EQU00011## Here, d ０ is the fifth constant, and d １ is -23820 sec / mm ２ and -4490 sec / mm ２ a sixth constant included between 2.25 and 13.48, the fifth constant being in particular included between 2.25 and 13.48, ADC being the 25th or lower percentile of the apparent diffusion coefficient values ​​of all pixels corresponding to the second portion, and - determining that the second portion contains aggressive cancer if a second criterion based on the calculated second score Y is verified, or determining that the second portion does not contain aggressive cancer if the first criterion is not verified; The system of claim 1 , configured as follows:

12. A method for aiding in the diagnosis of aggressive prostate cancer, the method being performed by a system (10) including a control module (15), the method being performed by the control module (15) and comprising the steps of: receiving (110) at least one magnetic resonance image of a first portion of a peripheral region of a prostate of a subject, said image comprising a set of pixels; the image comprising, for each pixel, an apparent diffusion coefficient value for a corresponding region of the first portion; - calculating a first score (130), - determining (140) that said first portion contains aggressive cancer based on said calculated first score; Including, The first score is calculated by the control module (15) using the following quantity x １ and the amount x ２ and ##EQU00012## where W is a normalized wash-in rate or normalized wash-out rate value of a first portion, the wash-in or wash-out value being expressed in Hertz and estimated from a set of magnetic resonance images of the first portion, ADC is a percentile of the apparent diffusion coefficient values ​​of the first portion, and mm 2 / s, １ is a first constant, and where W is the normalized wash-in rate value, the first constant is −6×10 －３ mm ２ and -10 －３ mm ２ and the percentile is less than or equal to 40, and W is a normalized washout rate value, the first constant is 2×10 －２ mm ２ and 15 x 10 －２ mm ２ and the percentile is less than or equal to 15; ##EQU00013## Here, b １ is 10 －７ mm ２ / s ２ and 14 x 10 －７ mm ２ / s ２ a first constant comprised between ADC and TTP; ADC being the percentile comprised between 20 and 45 of the apparent diffusion coefficient values ​​of all pixels corresponding to the first portion; TTP being the time to peak value of the first portion, expressed in seconds, the time to peak value being the duration from a first instant to a second instant, the first instant being the arrival time of a contrast agent at the prostate and the second instant being the time of maximum contrast concentration of the MRI signal of the first portion.

13. A computer program product (45) comprising software instructions configured to perform the method of claim 12 when the software instructions are executed on a processor (35).

14. An information medium (50) containing software instructions configured to perform the method according to claim 12 when the software instructions are executed on a processor (35).

15. A diagnostic method comprising using a method for assisting in the diagnosis of aggressive prostate cancer as described in claim 12 and establishing a clinical diagnosis based on the results of the method for assisting in the diagnosis of aggressive prostate cancer.

Images