Artificial intelligence-based simultaneous diagnosis system for multiple cancers using exosome SERS signals

Abstract

The present invention relates to an artificial intelligence-based system and method for simultaneous diagnosis of multiple cancers using exosome SERS signals. According to the present invention, the artificial intelligence-based system for simultaneous diagnosis of multiple cancers using exosome SERS signals includes a signal acquisition unit that drops exosomes acquired from a subject onto a chip and then acquires multiple exosome SERS signals from the chip; a cancer diagnosis unit that inputs the acquired multiple exosome SERS signals into a trained cancer classification algorithm, acquires a signal value of 0 or 1 for each of the multiple exosome SERS signals, and diagnoses cancer or normal using the average of the acquired signal values; and a cancer information provision unit that, if cancer is diagnosed, inputs the multiple exosome SERS signals into multiple TOO discrimination algorithms, acquires a signal value of 0 or 1 for each of the multiple exosome SERS signals, predicts carcinoma using the average of the acquired signal values, and provides information about the predicted carcinoma. This invention is a technology developed through the Seoul Industrial Development Institute's 2021 Bio-Medical Technology Commercialization Support Project (BT210040) "Commercialization of in vitro diagnostic equipment for simultaneous screening of multiple cancers through artificial intelligence."

Description

[Technical field]

[0001] The present invention relates to an artificial intelligence-based simultaneous multiple cancer diagnosis system and method using exosome SERS signals, and more specifically, to a simultaneous multiple cancer diagnosis system and method that detects a SERS signal map that covers the entire exosome rather than a specific marker, and simultaneously diagnoses multiple cancers through an artificial intelligence algorithm trained using the detected SERS signal map. [Background technology]

[0002] Cancer remains the leading cause of death in modern humans, but early diagnosis improves cancer outcomes by providing appropriate treatment. Early cancer detection by in vitro diagnostics is one of the important goals in the biomedical field, allowing routine cancer management.

[0003] However, although various cancer biomarkers such as carcinoembryonic antigen (CEA) 5 and prostate-specific antigen (PSA) 6 are currently being used in the diagnosis and prognosis strategies of various cancers, they rarely exist for early-stage cancers, and many cancers still have the problem of a lack of effective pre-screening tools.

[0004] Recently, there have been attempts to use exosomes, which are rich in information about mother cells, as biomarkers for liquid biopsies, and research has been conducted to detect information about disease exosomes present in body fluids using various biomaterial detection technologies to diagnose or monitor diseases early.

[0005] Previously, there have been attempts to diagnose cancer using the SERS signal of exosomes in the blood, but this was only able to diagnose a single cancer and was not able to diagnose multiple cancers simultaneously, and there was a problem that classification errors in the algorithm could result in false positive diagnoses.

[0006] The technology behind the present invention is disclosed in Korean Patent Registration No. 10-2225231 (announced on March 9, 2021). Summary of the Invention [Problem to be solved by the invention]

[0007] Thus, the present invention provides a system and method for simultaneously diagnosing multiple cancers, which detects a SERS signal map that covers the entire exosome rather than a specific marker, and simultaneously diagnoses multiple cancers through an artificial intelligence algorithm trained using the detected SERS signal map. [Means for solving the problem]

[0008] According to an embodiment of the present invention for achieving the above technical objectives, the AI-based simultaneous multiple cancer diagnosis system using exosome SERS signals includes a signal acquisition unit that drops exosomes acquired from a subject onto a chip and then acquires a plurality of exosome SERS (Surface Enhanced Raman Spectroscopy) signals from the chip; a cancer diagnosis unit that inputs the acquired plurality of exosome SERS signals into a trained cancer classification algorithm, acquires a signal value of 0 or 1 for each of the plurality of exosome SERS signals, and diagnoses cancer or normal using an average of the acquired signal values; and a cancer information provision unit that, if cancer is diagnosed, inputs the plurality of exosome SERS signals into a plurality of TOO (tissue of origin) discrimination algorithms, acquires a signal value of 0 or 1 for each of the plurality of exosome SERS signals, predicts carcinoma using the average of the acquired signal values, and provides information about the predicted carcinoma.

[0009] The artificial intelligence-based simultaneous multiple cancer diagnosis system using exosome SERS signals may further include a SERS signal collection unit that acquires a plurality of first exosome SERS signals from exosomes acquired from a normal subject and a plurality of second exosome SERS signals from exosomes acquired from a cancer patient having a plurality of carcinomas, labels the first exosome SERS signals as 0, and labels the second exosome SERS signals as 1; and a first learning unit that inputs the labeled first exosome SERS signals and the labeled second exosome SERS signals into a cancer classification algorithm and trains the cancer classification algorithm to classify the input exosome SERS signals as 0 or 1, respectively.

[0010] The artificial intelligence-based multiple cancer simultaneous diagnosis system using exosome SERS signals may further include a second learning unit that inputs a second exosome SERS signal obtained from a cancer patient corresponding to a specific carcinoma among the multiple carcinomas and a second exosome SERS signal obtained from the remaining cancer patients excluding the specific carcinoma into a plurality of TOO discrimination algorithms and trains each TOO discrimination algorithm to determine whether the second exosome SESR signal corresponds to the specific carcinoma.

[0011] The signal acquisition unit can acquire an exosome SERS signal map including n*m exosome SERS signals from the chip including n*m dot arrays (where n and m are the same or different natural numbers).

[0012] The cancer diagnosis unit inputs the n*m ​​exosome SERS signals into the cancer classification algorithm and outputs a signal value of 0 or 1 corresponding to each of the n*m ​​exosome SERS signals. If the average of the output signal values ​​is close to 0, the cancer can be classified as normal, and if the average of the output signal values ​​is close to 1, the cancer can be classified and diagnosed.

[0013] The cancer information providing unit inputs the n*m ​​exosome SERS signals to the multiple TOO discrimination algorithms, and the multiple TOO discrimination algorithms output a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compare the average of the output signal values ​​with a classification standard value for the carcinoma to determine whether or not it corresponds to each carcinoma.

[0014] According to an embodiment of the present invention, a method for simultaneous multiple cancer diagnosis using a simultaneous multiple cancer diagnosis system includes the steps of: dropping exosomes obtained from a subject onto a chip, and then acquiring a plurality of exosome SERS signals from the chip; inputting the acquired plurality of exosome SERS signals into a trained cancer classification algorithm to acquire a signal value of 0 or 1 for each of the plurality of exosome SERS signals, and diagnosing cancer or normal using an average of the acquired signal values; and, if cancer is diagnosed, inputting the plurality of exosome SERS signals into a plurality of TOO discrimination algorithms to acquire a signal value of 0 or 1 for each of the plurality of exosome SERS signals, and predicting carcinoma using the average of the acquired signal values, and providing information about the predicted carcinoma. Effect of the Invention

[0015] As described above, according to the present invention, the exosome SERS signal map is input into a cancer classification algorithm to first diagnose the presence or absence of cancer, thereby reducing false positive diagnoses. Furthermore, the exosome SERS signal map diagnosed as cancer can be input into multiple TOO discrimination algorithms and reanalyzed to classify the carcinomas, thereby making it possible to simultaneously diagnose multiple cancers, which is useful for early cancer detection.

[0016] In addition, according to the present invention, the complex SERS signal of exosomes can be analyzed through artificial intelligence to overcome the conventional limitation of signal heterogeneity. This enables non-invasive liquid biopsy of cancer using exosomes. In addition, cancer diagnosis can be performed using only blood, without the risk of radiation exposure as with X-ray or CT, or invasive tissue biopsy, and it can be used not only for cancer diagnosis but also for patient treatment monitoring. [Brief description of the drawings]

[0017] [Figure 1] 1 is a configuration diagram illustrating a system for simultaneously diagnosing multiple cancers according to an embodiment of the present invention. [Diagram 2] 1 is a flowchart illustrating a method for simultaneous multiple cancer diagnosis using a system for simultaneous multiple cancer diagnosis according to an embodiment of the present invention. [Diagram 3] FIG. 3 is an exemplary diagram showing SERS spectra appearing for each carcinoma in step S210 shown in FIG. 2. [Figure 4] FIG. 3 is an exemplary diagram illustrating a method for labeling a first exosome SERS signal and a second exosome SERS signal in step S220 shown in FIG. 2. [Diagram 5] FIG. 3 is an exemplary diagram illustrating a method for training a cancer classification algorithm in step S220 shown in FIG. 2. [Figure 6] FIG. 3 is an exemplary diagram illustrating step S230 shown in FIG. [Figure 7] FIG. 3 is an exemplary diagram illustrating step S240 shown in FIG. [Figure 8] 1 is an exemplary diagram showing the accuracy of cancer diagnosis according to an embodiment of the present invention. [Figure 9] 11 is an exemplary diagram showing a false positive diagnosis rate when a cancer patient is diagnosed according to an embodiment of the present invention and then a carcinoma is classified using a TOO discrimination algorithm; [Figure 10] FIG. 1 is an exemplary diagram showing an ROC curve for an artificial intelligence-based simultaneous diagnosis method for multiple cancers using exosome SERS signals according to an embodiment of the present invention. [Figure 11]FIG. 13 is an exemplary diagram showing accuracy of multiple cancer diagnosis results using the TOO discrimination algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings, in which the thickness of lines and the size of components are exaggerated for clarity and convenience of description.

[0019] In addition, the terms described below are defined in consideration of the functions of the present invention, and may vary depending on the intention or practice of a user or operator. Therefore, the definitions of such terms should be based on the overall content of this specification.

[0020] Hereinafter, the multiple cancer simultaneous diagnosis system according to the embodiment of the present invention will be described in more detail with reference to FIG.

[0021] FIG. 1 is a configuration diagram illustrating a system for simultaneously diagnosing multiple cancers according to an embodiment of the present invention.

[0022] As shown in FIG. 1, the multiple cancer simultaneous diagnosis system 100 according to an embodiment of the present invention includes a SERS signal collecting unit 110, a first learning unit 120, a second learning unit 130, a signal acquiring unit 140, a cancer diagnosing unit 150, and a cancer information providing unit 160.

[0023] First, the SERS signal collecting unit 110 collects exosome SERS signals obtained by performing Raman spectroscopy on exosomes collected from the plasma of a normal individual and exosomes collected from the plasma of a cancer patient.

[0024] Hereinafter, for convenience of explanation, the exosome SERS signal obtained from a normal subject is referred to as the first exosome SERS signal, and the exosome SERS signal obtained from a cancer patient is referred to as the second exosome SERS signal.

[0025] Then, the SERS signal collecting unit 110 labels the multiple first exosome SERS signals as 0, and labels the multiple second exosome SERS signals as 1.

[0026] The first learning unit 120 constructs a cancer classification algorithm based on deep learning, and inputs the labeled first exosome SERS signal and the labeled second exosome SERS signal into the constructed cancer classification algorithm for learning. Then, the cancer classification algorithm outputs a signal value of 0 or 1 corresponding to the input exosome SERS signal.

[0027] The second learning unit 130 constructs a TOO discrimination algorithm based on deep learning. At this time, a plurality of TOO discrimination algorithms are constructed corresponding to carcinomas. Here, the carcinomas may include at least one of lung cancer, breast cancer, colon cancer, liver cancer, pancreatic cancer, and gastric cancer, but are not necessarily limited thereto, and may include various other carcinomas.

[0028] Next, the second learning unit 130 inputs the second exosome SERS signals acquired from each cancer patient into a plurality of TOO discrimination algorithms for learning. Here, the TOO discrimination algorithm for a specific carcinoma is input with the second exosome SERS signals acquired from the cancer patients corresponding to the specific carcinoma and the second exosome SERS signals acquired from the remaining cancer patients excluding the specific carcinoma, and is trained to determine whether the second exosome SESR signal corresponds to the specific carcinoma. Then, the TOO discrimination algorithm outputs a signal value of 0 or 1 for the input exosome SERS signal.

[0029] The signal acquisition unit 140 drops exosomes collected from the plasma of the subject onto a chip, and then performs Raman spectroscopy on the chip to acquire multiple exosome SERS signals.

[0030] The cancer diagnosis unit 150 inputs the acquired exosome SERS signals into a trained cancer classification algorithm and acquires a signal value of 0 or 1 for each exosome SERS signal. Then, the cancer diagnosis unit 150 diagnoses the subject as normal or cancer using the average of the acquired signal values.

[0031] If the measurement subject is diagnosed with cancer, the cancer information providing unit 160 inputs the multiple exosome SERS signals into the multiple trained TOO discrimination algorithms, respectively.

[0032] Then, the multiple TOO discrimination algorithms output signal values ​​of 0 or 1 for the multiple exosome SERS signals input, and the cancer information providing unit 160 compares the average of the output signal values ​​with the classification standard value for the carcinoma in question to determine whether or not it corresponds to each carcinoma.

[0033] For example, the lung cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for a plurality of input exosome SERS signals, and compares the average of the output signal values ​​with a lung cancer classification standard value to discriminate the presence or absence of lung cancer. The breast cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for a plurality of input exosome SERS signals, and compares the average of the output signal values ​​with a breast cancer classification standard value to discriminate the presence or absence of breast cancer. The colon cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for a plurality of input exosome SERS signals, and compares the average of the output signal values ​​with a colon cancer classification standard value to discriminate the presence or absence of colon cancer. The liver cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for a plurality of input exosome SERS signals, and compares the average of the output signal values ​​with a liver cancer classification standard value to discriminate the presence or absence of liver cancer. The pancreatic cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for multiple input exosome SERS signals, and compares the average of the output signal values ​​with a pancreatic cancer classification standard value to discriminate the presence or absence of pancreatic cancer. Finally, the gastric cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for multiple input exosome SERS signals, and compares the average of the output signal values ​​with a gastric cancer classification standard value to discriminate the presence or absence of gastric cancer.

[0034] Then, the cancer information providing section 160 provides information about the carcinomas that have been determined to be positive.

[0035] A method for simultaneously diagnosing multiple cancers using a system for simultaneously diagnosing multiple cancers according to an embodiment of the present invention will be described in more detail with reference to FIGS.

[0036] FIG. 2 is a flowchart illustrating a method for simultaneously diagnosing multiple cancers using the system for simultaneously diagnosing multiple cancers according to an embodiment of the present invention.

[0037] As shown in FIG. 2, a method for simultaneous diagnosis of multiple cancers using a system for simultaneous diagnosis of multiple cancers according to an embodiment of the present invention includes a step of learning a cancer classification algorithm and a TOO discrimination algorithm, and a step of diagnosing multiple cancers using the learned cancer classification algorithm and TOO discrimination algorithm.

[0038] First, the step of learning the cancer classification algorithm and the TOO discrimination algorithm will be described. The multiple cancer simultaneous diagnosis system 100 collects exosome SERS signals from a normal subject group and a cancer patient group (step S210).

[0039] Specifically, plasma samples from normal subjects were obtained and exosomes were isolated from the plasma using size exclusion chromatography (SEC), and exosomes were also isolated from plasma samples from cancer patients using the same method.

[0040] Next, an exosome solution isolated from a plasma sample from a normal individual and an exosome solution isolated from a plasma sample from a cancer patient are dropped onto each Au nanoparticle assembly array chip and then dried.

[0041] Here, the Au nanoparticle assembly array chip is prepared by precipitating AuNPs (Au nanoparticles) in a colloidal solution, and then coating the NPs on an APTES-functionalized glass surface. In order to increase the detection throughput and uniformity of the signal acquisition process on the APTES-functionalized glass surface, the Au nanoparticle assembly array chip includes an n*m (where n and m are the same or different natural numbers) dot array, and the exosome SERS signal is measured at each dot.

[0042] Next, the SERS signal collecting unit 110 performs Raman spectroscopy on the Au nanoparticle assembly array chip on which the exosome solution has been dried to collect an exosome SERS signal map including multiple exosome SERS signals corresponding to the dot array.

[0043] That is, the SERS signal collecting unit 110 collects n*m first exosome SERS signals from the exosome solution of a normal individual, and collects n*m second exosome SERS signals from the exosome solution of a cancer patient.

[0044] FIG. 3 is an example diagram showing SERS spectra appearing for each carcinoma in step S210 shown in FIG.

[0045] The multiple cancer simultaneous diagnosis system 100 according to the embodiment of the present invention configures a cancer patient group using patients diagnosed with at least one of lung cancer, breast cancer, colon cancer, liver cancer, pancreatic cancer, and gastric cancer.

[0046] As shown in FIG. 3, lung cancer, breast cancer, colon cancer, liver cancer, pancreatic cancer, and stomach cancer each have a different SERS spectrum.

[0047] Once step S210 is completed, the first learning unit 120 labels the first exosome SERS signal obtained from the normal group and the second exosome SERS signal obtained from the cancer patient group, and trains a cancer classification algorithm using the labeled first exosome SERS signal and second exosome SERS signal (step S220).

[0048] FIG. 4 is an exemplary diagram illustrating a method for labeling a first exosome SERS signal and a second exosome SERS signal in step S220 shown in FIG.

[0049] As shown in Figure 4, the exosome solution of a cancer patient contains both normal exosomes and cancer exosomes, so the n*m ​​dots arranged on the Au nanoparticle assembly array chip may contain only one of normal exosomes or cancer exosomes, or may contain both normal exosomes and cancer exosomes. In other words, the multiple exosome SERS signals corresponding to the dot array are output differently.

[0050] However, the first learning unit 120 according to an embodiment of the present invention does not classify exosome SERS signals according to the presence or absence of cancer exosomes or normal exosomes, but labels all of the multiple first exosome SERS signals obtained from normal subjects as 0, and labels all of the multiple second exosome SERS signals obtained from cancer patients as 1.

[0051] FIG. 5 is an exemplary diagram illustrating a method for training a cancer classification algorithm in step S220 shown in FIG.

[0052] As shown in FIG. 5, the first learning unit 120 randomly extracts learning data and test data from the labeled first exosome SERS signal and the labeled second exosome SERS signal.

[0053] Then, the first learning unit 120 uses the first exosome SERS signal and the second exosome SERS signal corresponding to the extracted learning data as input data and the labeled values ​​as output data to train a cancer classification algorithm.

[0054] That is, the cancer classification algorithm outputs a signal value of 0 or 1 corresponding to multiple input exosome SERS signals, and first diagnoses the presence or absence of cancer using the average of the output signal values.

[0055] Next, the second learning unit 130 learns a TOO discrimination algorithm using the second exosome SERS signals of the cancer patient group acquired in step S210 (step S230).

[0056] FIG. 6 is an example diagram illustrating step S230 shown in FIG.

[0057] As shown in FIG. 6, the second learning unit 130 constructs a plurality of TOO discrimination algorithms corresponding to lung cancer, breast cancer, colon cancer, liver cancer, pancreatic cancer, and stomach cancer, respectively.

[0058] Then, the second learning unit 130 inputs the second exosome SERS signal obtained from the lung cancer patient group and the second exosome SERS signal obtained from the remaining cancer patient group excluding the lung cancer patients into the lung cancer TOO discrimination algorithm, and trains the lung cancer TOO discrimination algorithm to output a signal value of 0 or 1 corresponding to the input exosome SERS signal.

[0059] In addition, the second learning unit 130 inputs the second exosome SERS signal obtained from the breast cancer patient group and the second exosome SERS signal obtained from the remaining cancer patient group excluding the breast cancer patients into the breast cancer TOO discrimination algorithm, and trains the breast cancer TOO discrimination algorithm to output a signal value of 0 or 1 corresponding to the input exosome SERS signal.

[0060] The second learning unit 130 also trains the colon cancer TOO discrimination algorithm, the liver cancer TOO discrimination algorithm, the pancreatic cancer TOO discrimination algorithm, and the stomach cancer TOO discrimination algorithm in the same manner.

[0061] When the learning of the algorithm is completed using steps S210 to S230, the multiple cancer simultaneous diagnosis system 100 diagnoses multiple cancers in the measurement subject.

[0062] First, the signal acquiring unit 140 acquires an exosome SERS signal extracted from the plasma of a subject (step S240).

[0063] FIG. 7 is an example diagram illustrating step S240 shown in FIG.

[0064] As shown in FIG. 7, the user collects plasma from the subject to be measured and applies chromatography to the collected plasma to separate exosomes.

[0065] The user then drops the solution in which the exosomes are dissolved onto the Au nanoparticle assembly array chip and allows it to dry.

[0066] Next, Raman spectroscopy is performed on the Au nanoparticle assembled array chip to obtain n*m (e.g., 100) exosome SERS signals corresponding to the dot array.

[0067] When step S240 is completed, the cancer diagnosis unit 150 inputs the n*m ​​exosome SERS signals into a cancer classification algorithm to determine whether the subject has cancer (step S250).

[0068] In more detail, the cancer diagnosis unit 150 inputs the n*m ​​exosome SERS signals to the cancer classification algorithm, which then outputs a signal value of 0 or 1 corresponding to each of the n*m ​​exosome SERS signals.

[0069] If the average of the output signal values ​​is close to 0, the cancer diagnosis unit 150 diagnoses the measurement subject as a normal subject, and if the average of the output signal values ​​is close to 1, the cancer diagnosis unit 150 diagnoses the measurement subject as a cancer patient.

[0070] In step S250, if the measurement subject is diagnosed as a cancer patient, the cancer information providing unit 160 inputs the n*m ​​exosome SERS signals into a plurality of trained TOO discrimination algorithms to classify the cancer of the measurement subject (step S260).

[0071] In more detail, the cancer information providing unit 160 inputs n*m exosome SERS signals to a lung cancer TOO discrimination algorithm, a breast cancer TOO discrimination algorithm, a colon cancer TOO discrimination algorithm, a liver cancer TOO discrimination algorithm, a pancreatic cancer TOO discrimination algorithm, and a gastric cancer TOO discrimination algorithm.

[0072] The lung cancer TOO discrimination algorithm then outputs a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compares the average of the output signal values ​​with the lung cancer classification standard value to determine whether the subject has lung cancer.

[0073] The breast cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compares the average of the output signal values ​​with the breast cancer classification standard value to determine whether the subject has breast cancer.

[0074] The colon cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compares the average of the output signal values ​​with the colon cancer classification standard value to determine whether the subject has colon cancer.

[0075] The liver cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compares the average of the output signal values ​​with the liver cancer classification standard value to determine whether the subject has liver cancer.

[0076] The pancreatic cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compares the average of the output signal values ​​with the pancreatic cancer classification standard value to determine whether the subject has pancreatic cancer.

[0077] Finally, the gastric cancer TOO discrimination algorithm outputs a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compares the average of the output signal values ​​with the gastric cancer classification standard value to determine whether the subject has gastric cancer.

[0078] Next, the cancer information providing unit 160 provides a prediction result of carcinoma using the results output from the six TOO discrimination algorithms.

[0079] Hereinafter, the results of simultaneously diagnosing multiple cancers using the cancer classification algorithm and the TOO discrimination algorithm according to an embodiment of the present invention will be described in more detail with reference to FIGS.

[0080] FIG. 8 is an exemplary diagram showing the accuracy of cancer diagnosis according to an embodiment of the present invention.

[0081] As shown in Figure 8, when all exosome SERS signals obtained from cancer patients are labeled as 1, the exosome SERS signals obtained in areas where cancer exosomes do not exist are different from the actual correct answer. In other words, the prediction accuracy for individual exosome SERS signals may be somewhat lower.

[0082] However, when diagnosis is made using the average signal value output by a cancer classification algorithm for multiple exosome SERS signals obtained from cancer patients, the diagnostic accuracy is approximately 99%.

[0083] FIG. 9 is an example diagram showing the degree of false positive diagnosis when a cancer patient is diagnosed according to an embodiment of the present invention and then a carcinoma is classified using the TOO discrimination algorithm.

[0084] As shown in FIG. 9, comparing the conventional One vs. Rest approach with the approach of classifying carcinomas after cancer diagnosis according to an embodiment of the present invention, it was confirmed that the One vs. Rest approach had a misdiagnosis rate of approximately 21.8%, while the approach of classifying carcinomas after cancer diagnosis had a misdiagnosis rate of approximately 4.5%.

[0085] Therefore, the simultaneous multiple cancer diagnosis system according to the embodiment of the present invention has the effect of significantly reducing the probability of false positive diagnosis.

[0086] FIG. 10 is an example diagram showing an ROC curve for an artificial intelligence-based simultaneous diagnosis method for multiple cancers using exosome SERS signals according to an embodiment of the present invention, and FIG. 11 is an example diagram showing the accuracy of a multiple cancer diagnosis result using a TOO discrimination algorithm according to an embodiment of the present invention.

[0087] Specifically, Fig. 10 shows receiver operation characteristic (ROC) curves for verifying the effectiveness of the cancer classification algorithm and TOO discrimination algorithm according to an embodiment of the present invention. The area under the ROC curve (AUC) closer to 1 indicates greater usefulness, and the cancer classification algorithm according to an embodiment of the present invention shows a value of approximately 0.983, and multiple TOO discrimination algorithms also show excellent performance.

[0088] In addition, as shown in FIG. 11, a high concordance rate was observed when comparing actual diagnosis results with prediction results according to an embodiment of the present invention, and it was confirmed that relatively accurate prediction is possible even for patients with early-stage cancer.

[0089] In this way, the simultaneous multiple cancer diagnosis system according to the present invention inputs an exosome SERS signal map including multiple exosome SERS signals into a cancer classification algorithm to first diagnose the presence or absence of cancer, thereby reducing false positive diagnoses, and inputs exosome SERS signal maps diagnosed as cancer into multiple TOO discrimination algorithms and reanalyzes them to classify carcinomas, thereby simultaneously diagnosing multiple cancers and being useful for early cancer detection.

[0090] In addition, the multiple cancer simultaneous diagnosis system according to the present invention overcomes the conventional limitation of signal heterogeneity by analyzing the complex SERS signal of exosomes through artificial intelligence. This enables non-invasive liquid biopsy of cancer using exosomes. In addition, cancer diagnosis is possible using only blood, without the risk of radiation exposure as with X-ray or CT, or invasive tissue biopsy, and it can be used not only for cancer diagnosis but also for patient treatment monitoring.

[0091] The present invention has been described with reference to the embodiments shown in the drawings, but these are merely illustrative, and those skilled in the art will understand that various modifications and equivalent embodiments are possible. Therefore, the true technical scope of the present invention should be determined by the technical spirit of the claims. [Explanation of symbols]

[0092] 100: Multiple cancer simultaneous diagnosis system 110: SERS signal collection unit 120: First Study Section 130: Second Learning Section 140: Signal acquisition section 150: Cancer Diagnosis Department 160: Cancer Information Department

Claims

1. In an artificial intelligence-based simultaneous multiple cancer diagnosis system using exosome SERS signals, A signal acquisition unit that drops exosomes acquired from a subject onto a chip and then acquires a plurality of exosome SERS signals from the chip; A cancer diagnosis unit that inputs the acquired exosome SERS signals into a trained cancer classification algorithm, acquires a signal value of 0 or 1 for each of the acquired exosome SERS signals, and diagnoses cancer or normal using the average of the acquired signal values; and In the case of a cancer diagnosis, the plurality of exosome SERS signals are input to a plurality of TOO discrimination algorithms to obtain a signal value of 0 or 1 for each of the plurality of exosome SERS signals, and a cancer information providing unit is provided to predict carcinoma using an average of the obtained signal values, and to provide information on the predicted carcinoma; The cancer classification algorithm comprises: The system is trained to classify a first plurality of exosome SERS signals obtained from exosomes collected from plasma of a normal individual and a second plurality of exosome SERS signals obtained from exosomes collected from plasma of a cancer patient, the cancer patient including a plurality of carcinomas; The plurality of TOO discrimination algorithms include: A simultaneous multiple cancer diagnosis system that classifies second exosome SERS signals obtained from cancer patients who have a specific cancer among multiple cancers and second exosome SERS signals obtained from the remaining cancer patients excluding the specific cancer, and learns to determine whether or not a specific cancer among the multiple cancers is present.

2. A SERS signal collection unit that acquires a plurality of first exosome SERS signals and a plurality of second exosome SERS signals, labels the first exosome SERS signals as 0, and labels the second exosome SERS signals as 1; and A first learning unit that inputs the labeled first exosome SERS signal and the labeled second exosome SERS signal into a cancer classification algorithm and trains the cancer classification algorithm to classify each of the input exosome SERS signals into 0 or 1; The system for simultaneous diagnosis of multiple cancers according to claim 1 , further comprising:

3. The system for simultaneous diagnosis of multiple cancers according to claim 2, further comprising a second learning unit that inputs a second exosome SERS signal obtained from a cancer patient corresponding to a specific carcinoma among the multiple carcinomas and a second exosome SERS signal obtained from the remaining cancer patients excluding the specific carcinoma into a plurality of TOO discrimination algorithms, and trains each TOO discrimination algorithm to determine whether the second exosome SESR signal corresponds to the specific carcinoma.

4. The signal acquisition unit includes: The system for simultaneous diagnosis of multiple cancers according to claim 1, wherein an exosome SERS signal map including n*m exosome SERS signals is acquired from the chip including n*m (where n and m are the same or different natural numbers) dot arrays.

5. The cancer diagnosis section includes: Inputting the n*m ​​exosome SERS signals into the cancer classification algorithm, and outputting a signal value of 0 or 1 corresponding to each of the n*m ​​exosome SERS signals; 5. The system for simultaneous diagnosis of multiple cancers according to claim 4, wherein if the average of the output signal values ​​is close to 0, the cancer is classified as normal, and if the average of the output signal values ​​is close to 1, the cancer is classified and diagnosed.

6. The cancer information providing unit comprises: The multiple cancer simultaneous diagnosis system according to claim 5, wherein the n*m ​​exosome SERS signals are input to the multiple TOO discrimination algorithms, and the multiple TOO discrimination algorithms output a signal value of 0 or 1 for each of the input n*m exosome SERS signals, and compare the average of the output signal values ​​with a classification standard value for the carcinoma to determine whether or not each of the carcinomas corresponds to the corresponding carcinoma.

Images