Method for carrying out end-to-end screening for cervical cancer via an integrated and connected system combining cytology assisted by artificial intelligence, and an approved PCR test for oncogenic human papillomaviruses

Abstract

The present invention relates to a method for carrying out end-to-end screening for cervical cancer. The method according to the invention combines two technologies, cytology and PCR on the same computer platform connected to and shared by various stakeholders. The method comprises the steps of a qualified person having access to the computer platform taking a cervical-vaginal sample using a specific sampling kit comprising a reference code to be typed into the computer platform to identify the sampled patient, then a step of carrying out, within an analysis center connected to the computer system, a cytological examination using artificial intelligence to diagnose diseased cells within the cervical-vaginal sample, and an FDA-certified HR HPV PCR test. The results of the tests are then validated by expert cytotechnologists and pathologists connected to the platform, before their ultimate storage on said computing platform. The system sends a final diagnostic report to the consulting physician via the computer platform.

Description

End-to-end cervical cancer screening process via an integrated and connected system combining artificial intelligence-supported cytology and an approved oncogenic human papillomavirus PCR test Technical field

[0001] The present invention relates to the field of medical diagnosis. It particularly relates to a method for the end-to-end screening of cervical cancer via an integrated system combining cytology supported by artificial intelligence (AI) and PCR of the HPV-HR (high risk) type in order to carry out a rapid and accurate diagnosis of this cancer where the papillomavirus is involved in the vast majority of cases. Prior art

[0002] Cervical cancer represents a public health challenge. It is the 4 èmecancer that affects women worldwide with more than 600,000 new cases each year, more than 80% of which are in developing countries, and causes around 350,000 deaths / year (+90% in low- and middle-income countries).

[0003] It is believed that there is an association between the development of cervical cancer and human papillomaviruses, also called HPV (Human Papillomavirus) and more specifically the oncogenic strains called high-risk (HR HPV)

[0004] Despite the development of HPV vaccination in young girls, the silent nature of the disease requires early and systematic diagnosis of the vulnerable population, namely adult women.

[0005] In terms of diagnostic solutions, there are different solutions for diagnosing HPV in women. The first was the Pap test (1940-1954), liquid-based cytology (1996-1999) and finally automated screening using image analysis. This last technique, despite its interest for this type of diagnosis, remains very dependent on the expertise of practitioners and the quality of the equipment used (e.g., image quality).

[0006] Another factor that also poses a problem in the field of HPV diagnosis is the large number of variants of the virus (+200) and in particular the high-risk carcinogenic HPVs, also called oncogenic (HPV 16, 18, 31, 33, 45, 52, 58 and mainly HPV 16 and 18). They can cause the development of precancerous lesions that can develop into cancers after several years or even decades.

[0007] Early and systematic diagnosis is therefore a promising avenue for anticipating the development of the disease thanks to available treatments. Among the avenues for improving diagnosis, the use of artificial intelligence is a very important means for analyzing images and quickly assessing risks.

[0008] Patent application KR102237696_B1 (INDUSTRY ACADEMIC COOPERATION FOUNDATION KEiMYUNG UNIVERSITY) relates to an artificial intelligence computer-aided diagnosis system for automatic analysis of cervical cancer cell images and a control method therefor and, more specifically, it is capable of obtaining 3D visualization images and imaging processing of a pathological slide via an imaging processing unit of a microscope including an autofocus function and a high-speed scanning function, and enabling cervical cancer image diagnosis by deep learning via a cervical cancer diagnostic unit.

[0009] Patent application CN1 13506623 (HUNANLABSCI MEDICAL ROBOT CO LTD; HUNAN SUOLAI INTELLIGENT TECH co LTD) relates to an artificial intelligence cervical cancer detection system, comprising an analyzer which comprises an analyzer body and an intelligent detection system disposed in the analyzer body. The intelligent detection system comprises an information input module, an operation detection mode selection module, a glass sheet identification module, a detection mode selection module, an intelligent detection module, a cloud expert diagnosis module, a result judgment module, an information storage module, a module information output module, an information display module and an information printing module;

[0010] These solutions, despite their relevance in terms of ease and speed of diagnosis thanks to AI, remain less sensitive when the disease is in an early stage without apparent signs. It therefore remains less effective when seeking treatment of the disease at a stage where there are not yet any lesions. Such a problem can be addressed thanks to the solution of the present invention which on the one hand will facilitate mass screening thanks to a platform connected end-to-end from the patient's care by a prescribing doctor who will take the sample (smear), to the production of the diagnostic report, and on the other hand by combining the precise cytological diagnosis thanks to AI AND the HPV HR molecular diagnosis (co-testing), will generate for the prescribing doctor a rapid and reliable report since with a sensitivity which approaches 100% and therefore with an excellent negative predictive value. Statement of the invention

[0011] To achieve the objectives of the invention, mainly a wide geographical coverage of screening and a better prediction or diagnosis of cervical cancer, a screening method has been proposed combining a cytological test assisted by artificial intelligence (AI) and a PV HR PCR test to confirm probable cases (presence or absence of oncogenic strains of the virus). The method according to the invention takes place in six stages: - cervico-vaginal sampling by a prescribing doctor registered on the platform of the solution called “Al PAP”, - registration of a reference code for the patient sampled on the platform, - routing of the sample to the Data Pathology analysis center, - performance of the cytological examination assisted by an EU-certified AI and an FDA-certified HR PCR test, this on the same cervico-vaginal sample, - carrying out a pre-diagnosis by a cytotechnician and a pathologist to confirm the machine results, - Automatic preparation and sharing of the final diagnostic report with the prescribing physician within 48 to 72 hours from the date of the sample.

[0012] The method is based on a platform comprising human and technical resources to coordinate the flow of material and data across its various components. The first element of the screening process involves collecting a smear sample from the patient using a dedicated collection kit. The collection can be done by the prescribing physician, in a specialized laboratory or by a midwife. The sample is then transferred to the Data Pathology analysis center to carry out co-testing. Data pathology laboratories are pathological anatomy centers connected, as part of the method of the present invention, to the IT platform to feed the patient's file with the test results.

[0013] Co-testing includes a cytological test supported by a European Union (e.g., CE) certified artificial intelligence (AI) solution and an FDA-certified high-risk (HR) HPV PCR test. Both tests are performed using the same sample (Smear).

[0014] For the cytological test, the smear sample is placed on an in vitro slide to allow the slide to be scanned by a scanner. The scanning process takes place in three steps: - Preparation and coloring of the sample - Scanning the blade - Automatic analysis of the blade by artificial intelligence (AI)

[0015] The digitalization of the slide then allows the image data to be used via a processing unit including artificial intelligence to extract information on the color, size, nature of the contours and quantity of abnormal cells.

[0016] The PCR test aims to detect DNA from high-risk HPV strains in order to complement the results of the cytological analysis for better patient care.

[0017] The results of both tests are reviewed by at least one cytotechnician and one pathologist (the experts) to generate a report including both the results of the supported cytological analysis by AI, the presence or absence of high-risk HPV and recalls the patient's care protocol according to current guidelines

[0018] The platform is open to a number of local and international experts to verify, critique and validate the results of the analysis proposed by artificial intelligence.

[0019] Thanks to these two tests and the experts' opinions, the final report suggests to the prescribing physician the appropriate mode of management depending on the presence or absence of lesions, the presence or absence of oncogenic strains of the virus. It will also allow to define the appropriate frequency to repeat a cytological test and / or a confirmatory HR HPV PCR and a follow-up program.

[0020] The architecture of the DATA PATHOLOGY platform is described in detail below using the figures attached to this application. Summary description of the drawings

[0021] Other characteristics and advantages of the invention will appear in the detailed description which follows and which refers to the appended figures, given solely by way of non-limiting example. - Figure 1 represents a diagram of the operation of the Al PAP co-testing platform and the interactions with the various stakeholders from end-to-end. - Figure 2 illustrates as an example the procedure to follow in the event of a positive HPV PCR and negative cytology. Way(s) of carrying out the invention 1. Figure 1 illustrates the end-to-end cervical cancer screening process using the co-testing technique which combines two technologies, cytology and PCR on a single connected computer platform, the process includes the following steps: - Cervicovaginal sampling by the prescribing physician having access to said computer platform, using a specific sampling kit including a reference code, - Registration of the patient's reference code taken on the computer platform, - Routing of the cervico-vaginal sample to the Data Pathology analysis center connected to the computer system, - Carrying out, on the same cervico-vaginal sample, a cytological examination using an artificial intelligence algorithm to detect pre-cancerous or cancerous lesions and an FDA-certified HR HPV test, - Validation of said test results by expert cytotechnicians and pathologists who are partners of the Al PAP platform, - Recording of the results of the two validated tests from the previous step on the Al PAP computer platform using the reference code of the patient sampled, - Automatic notification of the final diagnostic report to the prescribing physician via the Al PAP IT platform.

[0022] The architecture of the said platform is based on six levels of intervention. The first level at the prescribing physician or the latter takes a sample of the cervical vaginal smear using the sampling kit made available by the Datapathology platform (fig. 1). The kit includes the material to take the sample and a bottle with a preservation liquid which allows the placing under a slide of said sample and digitization of the latter for cytological analysis by Artificial Intelligence (AI). The sample is taken in the gynecological position, quickly and painlessly, with the aim of recovering cells from the cervix.

[0023] The smear sample from step (1) is sent to the pathological anatomy laboratory (3) (Data Pathology analysis center) to carry out the co-testing (4). A first AI-assisted cytological analysis which consists of spreading the smear sample on a dedicated slide using an automaton to ensure homogeneous dispersion of the sample on the slide (4-a), after this operation the slide is digitized using a scanner (4-b) and an image is sent to a processing unit for image processing (4-c).

[0024] A processing unit (4-c) comprises an artificial intelligence algorithm capable of extracting information from the image data (pixels) information on the nature of the cells taken from the cervix. The images mainly concern the color of the cells, their size, the shape of their outline and their number. This information is then correlated with reference data at the treatment unit level to judge the nature of the cells, and to make a diagnosis of the presence or absence of signs of cancerous lesions.

[0025] The co-testing step (4) also includes a second diagnostic test (4-d) based on PCR (Polymerase Chain Reaction) aimed at diagnosing HPV - HR (high-risk human papillomavirus targeting genotypes 16, 18 and 45), said test is carried out on the same sample with an FDA-certified HPV -HR kit.

[0026] The processing unit then produces an initial diagnostic report (5) which will be accessible to experts (cytologists and pathologists) to ensure the reliability and accuracy of the conclusions. This exchange helps improve the quality of the AI ​​platform by continuously strengthening its learning and improving the reliability and completeness of the reference data.

[0027] Once the report has been validated by the experts, a final report is generated by the platform, which takes into account the cytology and the results of the HR HPV test to suggest a patient management protocol to the prescribing physician. This report is made accessible to the prescribing physician through an automatic notification system on the Al PAP IT platform.

[0028] According to a first aspect of the invention, co-testing comprises liquid-based cytology and an approved HPV-HR test (e.g. FDA, CE, etc.). As illustrated by Figure 2 as an example, as there are many different scenarios, the genotyping result can guide the prescribing physician on how the patient should be managed and monitored subsequently. If at least one of the three HPV genotypes 16, 18 or 45 is detected, the patient must repeat a confirmatory cytological examination every 6 months (Fig. 2).

[0029] According to a second aspect of the invention, the cytological examination is carried out on a digitalized slide. The pre-reading is carried out by a platform computer science including an artificial intelligence algorithm trained by reference data to identify and qualify abnormal cells in the cervico-vaginal sample. Identification is mainly based on the comparison of their size, color and the shape of their outline.

[0030] According to another aspect, the invention relates to an end-to-end cervical cancer screening system comprising a shared IT platform and connecting different stakeholders, namely the prescribing physician, the analysis center (Data Pathology) and experts, each patient is identified by a reference code entered at the level of said platform by the prescribing physician at the time of the cervicovaginal sample, said sample is processed by a data pathology analysis center connected to said platform by carrying out a cytological test supported by artificial intelligence for reading the digitalized slides relating to the patient's cervicovaginal sample, and an HPV HR PCR test,The results of the cytological and PCR HPV HR tests are notified to experts connected to the platform to validate them and produce a final diagnostic report for the prescribing physician, recalling the patient's care protocol according to the different scenarios (e.g., figure 2).

[0031] The advantages of the invention are multiple. The first is the accuracy of the solution, approaching 100% sensitivity, thus avoiding missing pre-cancerous and cancerous lesions; the second concerns the response times concerning the tests carried out, which do not exceed 48 to 72 hours at most. This allows for rapid medical care for patients and allows them to avoid unbearable waits of up to a month to find out whether or not they suffer from cervical cancer.

[0032] The accuracy of the solution lies in the combination of AI which offers better accuracy in reading digitalized cytological test slides, and PCR which offers high accuracy in detecting HR HPV, particularly strains 16, 18 and 45.

[0033] The platform will also help practitioners by dematerializing files and providing access to patient history.

[0034] The invention also provides a better screening solution to reduce the incidence of cervical cancer, especially in the context of middle- to low-income countries.

[0035] The table below concerns the risk linked to each diagnostic technique alone or in association (co-testing) as proposed in our solution, in relation to precancerous CIN3 (Cervical Intraepithelial Neoplasia) or cancerous lesions The table clearly shows through this large-scale study that it is co-testing which allows to have the lowest rate of hidden precancerous lesions (1.2%) and also the lowest rate of cancerous lesions (5.5%) compared to cytology alone or an HPV test alone.

Claims

Claims:

1. End-to-end screening method for cervical cancer characterized in that it combines two technologies, cytology and PCR on the same connected computer platform, said method comprises the following steps: - Cervicovaginal sampling by the prescribing physician having access to said computer platform, using a specific sampling kit including a reference code, - Registration of the patient's reference code taken on the computer platform, - Routing of the cervical-vaginal sample to an analysis center connected to the IT platform, - Carrying out, on the same cervico-vaginal sample, a cytological test using an artificial intelligence algorithm to detect pre-cancerous or cancerous lesions and a certified HR HPV test, - Validation of said test results by expert cytotechnicians and pathologists who are partners of a computer platform called “Al PAP”, - Recording of the results of the two validated tests from the previous stage on the said “Al PAP” computer platform using the reference code of the patient sampled, - Automatic notification of a final diagnostic report to the prescribing physician by the said “Al PAP” IT platform.

2. End-to-end screening method for cervical cancer according to claim 1, characterized in that the HPV-HR PCR test comprises a genotyping step in the event of the presence of the virus by detecting DNA from high-risk HPV strains.

3. End-to-end screening method for cervical cancer according to claim 1, characterized in that the cytological test is carried out by artificial intelligence trained by reference data to identify and qualify the abnormal cells of the cervico-vaginal sample according to criteria of size, color and shape of the outline.

4. End-to-end screening method for cervical cancer according to claim 3, characterized in that the identification of abnormal cells is based mainly on the comparison of their size, their color and the shape of their outline.

5. End-to-end cervical cancer screening method according to claim 4, characterized in that the updating of the reference data takes into account the results of the HPV-HR PCR test when the latter is positive.

6. End-to-end cervical cancer screening method according to claim 5, characterized in that the reference data are continuously updated using a learning model and thanks to the opinions of experts connected to the platform.

7. End-to-end screening method for cervical cancer according to claim 7, characterized in that in the event that the genotyping is positive for at least one of the three genes 16, 18 or 45 of the human papilloma virus, a confirmatory cytology is required within an interval of 4 months.

8. End-to-end cervical cancer screening system characterized in that it comprises a shared IT platform connecting different stakeholders, namely the prescribing physician (1), the analysis center called "data pathology" (3) and the experts (5), each patient is identified by a reference code (2) entered at the level of said platform by the prescribing physician at the time of the cervico-vaginal sample, said sample is analyzed by a center (3) connected to said platform by carrying out a cytological test supported by artificial intelligence for reading the digitalized slides relating to the patient's cervico-vaginal sample using a processing unit (4-c), and a PCR HPV HR test (4-d),the results of the cytological and PCR-HPV HR tests are notified to experts (5) connected to the platform to validate them and an artificial intelligence algorithm (6) subsequently compiles the test results and the experts' observations to produce and notify the prescribing physician of a final diagnostic report recalling a patient care protocol according to the different scenarios.,

Images