Historical data oriented ultrasonic thyroid nodule intelligent detection method
By acquiring patients' historical information and using deep learning algorithms, a set of isotopic detection angles was established, solving the problem of existing ultrasound thyroid nodule detection relying on experience and realizing an automated and efficient detection solution.
Patent Information
- Application Number
- CN202511041128.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-28
- Publication Date
- 2025-11-11
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing ultrasound methods for detecting thyroid nodules rely on operator skills and experience, lacking automation and historical data guidance, resulting in high reproducibility and low efficiency.
By obtaining patients' historical medical information, it is determined whether the patient is an old or new patient, and corresponding detection strategies are adopted. A smart detection model is built using deep learning algorithms, and a set of isotopic detection angles is established to perform ultrasound thyroid nodule detection.
It improves the comprehensiveness and accuracy of detection, saves the running time of the matching procedure, provides a reference for the angle of use of ultrasonic testing instruments, and improves detection efficiency and accuracy.
Smart Images
Figure CN120918703A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of ultrasound thyroid nodule detection technology, specifically to an intelligent ultrasound thyroid nodule detection method based on historical data. Background Technology
[0002] Ultrasound thyroid nodule detection is a non-invasive imaging technique used to assess the presence, size, shape, borders, internal structure, and blood flow signals of nodules within the thyroid gland. Its main purpose is to detect nodules, evaluate their benign or malignant characteristics, and guide further examination and treatment.
[0003] Commonly used ultrasound methods for detecting thyroid nodules include grayscale ultrasound, color Doppler ultrasound, and elastography. These methods all require a high level of operator skill and experience, cannot be automated, and rely on the operator's experience without historical data guidance, significantly increasing repetitiveness and unnecessary testing. To address these needs, this proposal aims to develop an intelligent ultrasound method for detecting thyroid nodules that is data-driven and improves both efficiency and accuracy. Summary of the Invention
[0004] This invention provides a historical data-driven intelligent ultrasound method for detecting thyroid nodules, which helps to solve the problems mentioned in the background section.
[0005] This invention provides the following technical solution: a historical data-guided intelligent ultrasound detection method for thyroid nodules, comprising:
[0006] Obtain the patient's historical medical records;
[0007] Determine whether a patient is a repeat patient based on their past medical history.
[0008] If the patient is not a previously diagnosed patient, the new patient testing strategy will be used to detect thyroid nodules.
[0009] If the patient is a long-term patient, an intelligent detection method will be used to detect thyroid nodules.
[0010] The patient's test results and the testing angle used by the testing instrument during the test are used to create and save a corresponding dataset through a data processing program;
[0011] Based on the size of the thyroid nodules, the examination results of patients with thyroid nodules and the angles used by the instruments during the examination were classified to form a thyroid nodule patient detection dataset. A deep learning algorithm was used to build an intelligent detection model and train it.
[0012] Establish all angles used by ultrasound detection instruments at the same coordinate points, and denot them as the isotopic detection angle set. Combine all the isotopic detection angle sets into the thyroid nodule isotopic detection angle set.
[0013] The test results are displayed on the user interface, and a diagnostic report is generated.
[0014] Optionally, determining whether a patient is a returning patient based on their past medical history specifically includes:
[0015] Set the treatment duration threshold to T;
[0016] Set the threshold for the frequency of medical visits to P;
[0017] Obtain the patient's first and last visit dates;
[0018] Get the number of patient visits. If no patient visits are found, record the number of patient visits as 0.
[0019] Obtain the patient's treatment duration. If no treatment time is obtained, record the patient's treatment duration as 0.
[0020] The patient's total interval between visits = the date of the last visit - the date of the first visit. If the patient's first visit date and last visit date are not available, the patient's total interval between visits is recorded as 1.
[0021] Patient visit frequency = Number of patient visits / Total interval between patient visits;
[0022] If a patient's consecutive consultation time exceeds T or the patient's consultation frequency exceeds P, then this patient is considered a long-term patient.
[0023] Otherwise, this patient will be classified as a new patient.
[0024] Optionally, the use of the new patient detection strategy for thyroid nodule detection specifically includes:
[0025] The patient's medical history was taken to determine the location of the thyroid nodules;
[0026] Obtain the user's Adam's apple and epiglottis, and denote them as the Adam's apple point and epiglottis point, respectively;
[0027] Starting from the laryngeal protuberance, draw a ray through the epiglottis, and denote the direction of the ray as the left direction;
[0028] Place the patient flat on the testing table, and mark the direction the patient's face is facing as "up".
[0029] With the laryngeal protuberance as the origin, draw a ray passing through the origin and the epiglottis. This ray is denoted as the Y-axis, and the leftward direction is denoted as the positive direction of the Y-axis. Draw a horizontal ray perpendicular to the Y-axis and pointing towards the right thyroid gland through the origin, which is denoted as the X-axis, and the direction of the ray is denoted as the positive direction of the X-axis. Draw a vertical ray perpendicular to the Y-axis through the origin, which is denoted as the Z-axis, and the upward direction is denoted as the positive direction of the Z-axis.
[0030] Thyroid nodules are detected in patients using ultrasound equipment;
[0031] The end of the ultrasound instrument that contacts the human body is designated as the lower end, and the other end is designated as the upper end;
[0032] The coordinates of the lower end of the ultrasonic instrument in the three-dimensional coordinate system are obtained and denoted as the lower end coordinates.
[0033] Detection is performed using a co-location detection strategy;
[0034] The patient's ultrasound images and the tilt angle of the ultrasound instrument during the examination are uploaded to the data processing program.
[0035] Optionally, the use of intelligent detection methods to detect thyroid nodules in patients specifically includes:
[0036] Obtain the data set of the ultrasound instrument usage angle during the patient's most recent thyroid nodule examination, and denote it as the patient's ultrasound instrument usage angle data set.
[0037] Obtain a dataset of ultrasound instrument usage angles during examinations of similar patients, and denote it as the dataset of ultrasound instrument usage angles for similar patients.
[0038] First, use the elements in the patient's ultrasound instrument usage angle dataset to perform the initial detection in sequence;
[0039] Obtain elements from the similar patient ultrasound instrument usage angle dataset that are not in the patient ultrasound instrument usage angle dataset, and form a secondary detection ultrasound instrument detection angle dataset.
[0040] The patient was subjected to a secondary examination using a secondary ultrasound instrument to detect angle datasets.
[0041] The patient's ultrasound images and the tilt angle of the ultrasound instrument during the examination are uploaded to the data processing program.
[0042] Optionally, the step of saving the patient's test results and the detection angle used by the detection instrument during the test by establishing a corresponding dataset through a data processing program specifically includes:
[0043] Obtain the patient's ID number;
[0044] Obtain the patient's appointment date;
[0045] Obtain the angle at which the ultrasound instrument is used during the patient's examination;
[0046] Acquire patient ultrasound images;
[0047] Determine if the patient is a returning patient;
[0048] If the patient is not a returning patient, create a root folder named after the patient's ID number and a subfolder named after the patient's visit date. Create a file named "Ultrasound Angle File" to save the ultrasound angles used during the patient's examination. Save the image files with the corresponding ultrasound angle names.
[0049] If it is a returning patient, find the root folder based on the patient's ID number, create a second-level folder named after the patient's visit date, create a file named "Ultrasound Instrument Angle File" to save the ultrasound instrument angle used during the patient's examination, and save the image files with the corresponding ultrasound instrument angle file names.
[0050] Optionally, the thyroid nodule patient examination dataset is categorized based on the size of the thyroid nodules, classifying the patient's examination results and the angle used by the instrument during testing. This dataset specifically includes:
[0051] Obtain the size of the thyroid nodule diagnosed in the patient and record it as the thyroid nodule size;
[0052] The obtained thyroid nodule size is compared with the thyroid nodule patient detection dataset to determine whether the thyroid nodule size category exists in the thyroid nodule patient detection dataset.
[0053] If it does not exist, create a new entry in the thyroid nodule patient detection dataset named after the obtained thyroid nodule size, and save the detection angle used by the ultrasound instrument during the detection of this patient;
[0054] If it exists, determine whether the angle used by the ultrasound instrument during the examination of this patient exists in the corresponding entry in the dataset of thyroid nodule patients. If it does not exist, add this angle data to the corresponding entry.
[0055] The processed entries are used to create a dataset of ultrasound instrument usage angles for patients of the same type.
[0056] Optionally, all angles used by the ultrasonic testing instruments to establish the same coordinate points form a set of corresponding testing angles, specifically including:
[0057] Obtain the coordinates of the lower endpoint of the ultrasonic testing instrument during testing, and record them as the testing contact coordinates;
[0058] Obtain the detection angles used by the ultrasonic testing instrument for detecting contact coordinates, and form a set of detection angles;
[0059] Determine whether the detection contact coordinates exist in the set of detection angles at the same location as the thyroid nodule;
[0060] If an element exists in the set of isotopic detection angles for thyroid nodules, then determine whether all elements in the detection angle set exist in the corresponding isotopic detection angle set. If an element exists but does not exist in the corresponding isotopic detection angle set, then add it to the corresponding isotopic detection angle set.
[0061] If the thyroid nodule does not exist in the isotopic detection angle set, a new isotopic detection angle set is created with the detection contact point coordinates as the name, and all elements in the detection angle set are added to the new isotopic detection angle set.
[0062] Optionally, the step of constructing and training an intelligent detection model using deep learning algorithms specifically includes:
[0063] Obtain data from different entries in the dataset of patients with thyroid nodules;
[0064] The acquired data is preprocessed, including noise removal, image enhancement, and standardization.
[0065] Multiple features are extracted from the preprocessed data, including morphological features, texture features, and grayscale features;
[0066] Using the extracted features, an intelligent detection model is constructed and trained using deep learning algorithms;
[0067] Input the ultrasound image of the thyroid gland to be detected, use the trained model to detect nodules, and output the detection results.
[0068] Optionally, the use of a co-location detection strategy specifically includes:
[0069] Obtain the coordinates of the lower end of the ultrasonic testing instrument and record them as the testing coordinates;
[0070] Determine whether the detection coordinates exist in the set of detection angles corresponding to thyroid nodules;
[0071] If it exists, obtain the maximum and minimum angles in the corresponding set of detection angles, and adjust the ultrasonic detector from the minimum angle to the maximum angle in sequence;
[0072] If not found, obtain the maximum and minimum angles from all isotopic detection angle sets in the thyroid nodule isotopic detection angle set. Compare all the maximum and minimum angles to find the final maximum and minimum angles. Use the final minimum angle to adjust the detection angles sequentially towards the final maximum angle. Select the angles corresponding to the clearest images as the operating angles of the ultrasound detector under this detection coordinate. If none of the acquired images are clear, decrease the final minimum angle and increase the final maximum angle until a clear image is found. Use the angle corresponding to the clear image as the operating angle of the ultrasound detector under this detection coordinate.
[0073] The present invention has the following beneficial effects:
[0074] 1. By verifying a patient's past medical history, we can determine if they are a returning patient and provide different testing options. A data processing program creates a dataset that allows attending physicians to quickly review a patient's history and saves time on the matching process. Categorizing patients by the size of their thyroid nodules allows for testing from different angles during follow-up examinations, improving the comprehensiveness of the testing. Simultaneously, a set of corresponding testing angles provides a reference for the use of ultrasound instruments at the same testing points for new patients.
[0075] 2. By obtaining patients' medical information, including treatment duration, date of visit, and number of visits, the frequency of visits is calculated. The treatment duration and frequency of visits are used to determine whether a patient is a repeat patient, which improves the accuracy of the judgment.
[0076] 3. The left direction was determined by using the Adam's apple and epiglottis as references, and the upward direction was determined by the direction the face was facing during the test. At the same time, a three-dimensional coordinate system was established using the Adam's apple, epiglottis, and right thyroid gland. The establishment of the three-dimensional coordinate system is helpful in determining the position of the lower end of the ultrasound instrument during the test. The use of the same-position detection strategy provides a basis for the angle of use of the ultrasound instrument for the detection position in the test of new patients.
[0077] 4. By first conducting a preliminary examination using the patient's ultrasound instrument usage angle dataset, and then conducting a secondary examination based on the differences between the elements in the similar patient ultrasound instrument usage angle dataset and the patient's ultrasound instrument usage angle dataset, it is beneficial to review the patient's previous examination position. At the same time, using different angles to examine the same examination position is beneficial to investigate the patient's condition from different angles.
[0078] 5. By creating root folders using the patient's ID number, the difference and uniqueness between each root folder are ensured. Creating second-level folders using the date of visit facilitates viewing the patient's visit data for different dates. Using the angle of the ultrasound instrument during the patient's examination as the name of the examination image allows for quick retrieval of the examination image results obtained from different angles, greatly improving the search speed.
[0079] 6. By using thyroid nodule size as a category, a dataset of ultrasound instrument usage angles for patients of the same type can be established. This can provide different detection angles during follow-up examinations of old patients, which is helpful for confirming the patient's condition.
[0080] 7. By establishing and matching the unique coordinates of the lower endpoint during ultrasound detection, the differences in the set of corresponding detection angles are established, and the matching speed is improved. The establishment of the set of corresponding detection angles provides a reference for the use of ultrasound detection instruments when detecting thyroid nodules in new patients. With the reference for detection angles, a systematic operation can be established, which also provides a basis for detection.
[0081] 8. By using elements from the set of corresponding detection angles to detect thyroid nodules in new patients, the efficiency of thyroid nodule detection is improved. By comparing different images to find clear images, the detection angle corresponding to the clear image is used as the detection angle for this detection position of the patient, thus improving the accuracy of the detection angle. Attached Figure Description
[0082] Figure 1 This is a schematic diagram of the detection process of the present invention. Detailed Implementation
[0083] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0084] An example of an intelligent ultrasound detection method for thyroid nodules based on historical data includes:
[0085] Obtain the patient's historical medical records;
[0086] Determining whether a patient is a repeat patient based on their past medical history is beneficial for implementing more effective testing protocols.
[0087] If the patient is not a previously diagnosed patient, the new patient testing strategy will be used to detect thyroid nodules.
[0088] If the patient is a long-term patient, an intelligent detection method will be used to detect thyroid nodules.
[0089] The patient's test results and the testing angle used by the testing instrument during the test are used to create and save a corresponding dataset through a data processing program;
[0090] Based on the size of the thyroid nodules, the examination results of patients with thyroid nodules and the angles used by the instruments during the examination were classified to form a thyroid nodule patient detection dataset. A deep learning algorithm was used to build an intelligent detection model and train it.
[0091] Establish all angles used by ultrasound detection instruments at the same coordinate points, and denot them as the isotopic detection angle set. Combine all the isotopic detection angle sets into the thyroid nodule isotopic detection angle set.
[0092] The test results are displayed on the user interface, and a diagnostic report is generated.
[0093] By verifying a patient's past medical history, we can determine if they are a returning patient and provide different testing options. A data processing program creates a corresponding dataset, which allows attending physicians to quickly review a patient's history and saves time on the matching process. Categorizing patients by the size of their thyroid nodules allows for testing from different angles during follow-up examinations, improving the comprehensiveness of the testing. Simultaneously, a set of corresponding testing angles provides a reference for the use of ultrasound instruments at the same testing points for new patients.
[0094] The method of determining whether a patient is a returning patient based on their past medical history specifically includes:
[0095] Set the treatment duration threshold to T. This threshold is usually set to six months or one year, and can be adjusted according to the actual situation.
[0096] Set the threshold for the frequency of visits to a P value. The smaller the P value, the lower the frequency and the fewer visits per unit time. The larger the P value, the more visits per unit time.
[0097] Obtain the patient's first and last visit dates;
[0098] Get the number of patient visits. If no patient visits are found, record the number of patient visits as 0.
[0099] Obtain the patient's treatment duration. If no treatment time is obtained, record the patient's treatment duration as 0.
[0100] Total patient visit interval = last visit date - first visit date. If the patient's first visit date and last visit date are not available, the total patient visit interval is recorded as 1. Since the total patient visit interval is used as a divisor when calculating the patient's visit frequency, and the divisor cannot be 0 in the basic calculation formula, the total patient visit interval is recorded as 1.
[0101] Patient visit frequency = Number of patient visits / Total interval between patient visits;
[0102] If a patient's consecutive consultation time exceeds T or the patient's consultation frequency exceeds P, then this patient is considered a long-term patient.
[0103] Otherwise, this patient will be classified as a new patient.
[0104] By obtaining patients' medical information, including treatment duration, date of visit, and number of visits, the frequency of visits can be calculated. Using treatment duration and frequency of visits as two metrics to determine whether a patient is a repeat patient improves the accuracy of the assessment.
[0105] The new patient detection strategy for thyroid nodules specifically includes:
[0106] The patient's medical history was taken to determine the location of the thyroid nodules;
[0107] The user's Adam's apple and epiglottis were obtained and denoted as the Adam's apple point and epiglottic point, respectively. The Adam's apple is part of the thyroid cartilage in the larynx. However, after puberty, due to the increase in androgen levels in males, the thyroid cartilage develops faster and is more prominent than in females, so the Adam's apple in males is usually more obvious. Females have relatively lower androgen levels, and the development of the thyroid cartilage is not as significant as in males, so the Adam's apple is less prominent, but it still exists. Therefore, the controversy that females do not have an Adam's apple is ruled out.
[0108] Starting from the laryngeal protuberance, draw a ray through the epiglottis, and denote the direction of the ray as the left direction;
[0109] Place the patient flat on the testing table, and mark the direction the patient's face is facing as "up".
[0110] With the laryngeal protuberance as the origin, draw a ray passing through the origin and the epiglottis. This ray is denoted as the Y-axis, and the leftward direction is denoted as the positive direction of the Y-axis. Draw a horizontal ray perpendicular to the Y-axis and pointing towards the right thyroid gland through the origin, which is denoted as the X-axis, and the direction of the ray is denoted as the positive direction of the X-axis. Draw a vertical ray perpendicular to the Y-axis through the origin, which is denoted as the Z-axis, and the upward direction is denoted as the positive direction of the Z-axis.
[0111] Thyroid nodules are detected in patients using ultrasound equipment;
[0112] The end of the ultrasound instrument that contacts the human body is designated as the lower end, and the other end is designated as the upper end;
[0113] The coordinates of the lower end of the ultrasonic instrument in the three-dimensional coordinate system are obtained and denoted as the lower end coordinates.
[0114] Detection is performed using a co-location detection strategy;
[0115] The patient's ultrasound images and the tilt angle of the ultrasound instrument during the examination are uploaded to the data processing program.
[0116] The left direction was determined by using the Adam's apple and epiglottis as references, and the upward direction was determined by the direction the person's face was facing during the test. At the same time, a three-dimensional coordinate system was established using the Adam's apple, epiglottis, and right thyroid gland. The establishment of the three-dimensional coordinate system is helpful in determining the position of the lower end of the ultrasound instrument during the test. The use of the same-position detection strategy provides a basis for the angle of use of the ultrasound instrument for the detection position in the test of new patients.
[0117] The method of using intelligent detection to detect thyroid nodules in patients specifically includes:
[0118] Obtain the data set of the ultrasound instrument usage angle during the patient's most recent thyroid nodule examination, and denote it as the patient's ultrasound instrument usage angle data set.
[0119] Obtain a dataset of ultrasound instrument usage angles during examinations of similar patients, and denote it as the dataset of ultrasound instrument usage angles for similar patients.
[0120] First, use the elements in the patient's ultrasound instrument usage angle dataset to perform the initial detection in sequence;
[0121] Obtain elements from the similar patient ultrasound instrument usage angle dataset that are not in the patient ultrasound instrument usage angle dataset, and form a secondary detection ultrasound instrument detection angle dataset.
[0122] The patient was subjected to a secondary examination using a secondary ultrasound instrument to detect angle datasets.
[0123] The patient's ultrasound images and the tilt angle of the ultrasound instrument during the examination are uploaded to the data processing program.
[0124] By first conducting a preliminary examination using a dataset of the angles from which the patient's ultrasound instrument was used, and then conducting a secondary examination based on the differences between the elements in the dataset of the angles from which the patient's ultrasound instrument was used and the elements in the dataset of the angles from which the patient's ultrasound instrument was used, it is beneficial to review the patient's previous examination position. At the same time, using different angles to examine the same examination position is beneficial to investigate the patient's condition from different perspectives.
[0125] The process of creating and saving a dataset based on the patient's test results and the testing angle used by the testing instrument during the test through a data processing program includes:
[0126] Obtain the patient's ID number;
[0127] Obtain the patient's appointment date;
[0128] Obtain the angle at which the ultrasound instrument is used during the patient's examination;
[0129] Acquire patient ultrasound images;
[0130] Determine if the patient is a returning patient;
[0131] If the patient is not a returning patient, create a root folder named after the patient's ID number. Since everyone's ID number is unique, using the patient's ID number as the name of the root folder conforms to the principle of uniqueness. Create a second-level folder named after the patient's visit date. Create a file for the angle of ultrasound instrument used during the patient's examination to save the angle of ultrasound instrument used. Save the image files with the corresponding ultrasound instrument angle name.
[0132] If it is a returning patient, find the root folder based on the patient's ID number, create a second-level folder named after the patient's visit date, create a file named "Ultrasound Instrument Angle File" to save the ultrasound instrument angle used during the patient's examination, and save the image files with the corresponding ultrasound instrument angle file names.
[0133] By creating root folders named after patients' ID numbers, the differences and uniqueness between each root folder are ensured. Creating second-level folders using the date of visit facilitates viewing the patient's visit data from different dates. Using the angle of the ultrasound instrument during the patient's examination as the name of the examination image allows for quick retrieval of the examination image results obtained from different angles, greatly improving the search speed.
[0134] The dataset is categorized based on the size of the thyroid nodules, classifying the examination results and the angles used by the instruments during testing to form a dataset of thyroid nodule patient examinations. This dataset specifically includes:
[0135] The size of the diagnosed thyroid nodules in patients is obtained. The size of thyroid nodules is unique. Using the size of thyroid nodules as the data set name for the angle of ultrasound examination of similar patients can effectively distinguish different data sets and quickly find matching data sets, which are denoted as thyroid nodule size.
[0136] The obtained thyroid nodule size is compared with the thyroid nodule patient detection dataset to determine whether the thyroid nodule size category exists in the thyroid nodule patient detection dataset.
[0137] If it does not exist, create a new entry in the thyroid nodule patient detection dataset named after the obtained thyroid nodule size, and save the detection angle used by the ultrasound instrument during the detection of this patient;
[0138] If it exists, determine whether the angle used by the ultrasound instrument during the examination of this patient exists in the corresponding entry in the dataset of thyroid nodule patients. If it does not exist, add this angle data to the corresponding entry.
[0139] The processed entries are used to create a dataset of ultrasound instrument usage angles for patients of the same type.
[0140] By using thyroid nodule size as a category, a dataset of ultrasound instrument usage angles for patients of the same type can be established. This can provide different detection angles during follow-up examinations of old patients, which is helpful for confirming the patient's condition.
[0141] All angles used by the ultrasonic testing instruments to establish the same coordinate points form a set of corresponding testing angles, specifically including:
[0142] Obtain the coordinates of the lower endpoint of the ultrasonic testing instrument during testing, and record them as the testing contact coordinates;
[0143] Obtain the detection angles used by the ultrasonic testing instrument for detecting contact coordinates, and form a set of detection angles;
[0144] Determine whether the detection contact coordinates exist in the set of detection angles at the same location as the thyroid nodule;
[0145] If an element exists in the set of isotopic detection angles for thyroid nodules, then determine whether all elements in the detection angle set exist in the corresponding isotopic detection angle set. If an element exists but does not exist in the corresponding isotopic detection angle set, then add it to the corresponding isotopic detection angle set.
[0146] If the thyroid nodule does not exist in the isotopic detection angle set, a new isotopic detection angle set is created with the detection contact point coordinates as the name, and all elements in the detection angle set are added to the new isotopic detection angle set.
[0147] By establishing and matching the unique coordinates of the lower endpoint during ultrasound detection, the differences in the set of corresponding detection angles are established, and the matching speed is improved. The establishment of the set of corresponding detection angles provides a reference for the use of ultrasound detection instruments when detecting thyroid nodules in new patients. With the reference for detection angles, a systematic operation can be established, which also provides a basis for detection.
[0148] The process of constructing and training an intelligent detection model using deep learning algorithms specifically includes:
[0149] Obtain data from different entries in the dataset of patients with thyroid nodules;
[0150] The acquired data is preprocessed, including noise removal, image enhancement, and standardization.
[0151] Multiple features are extracted from the preprocessed data, including morphological features, texture features, and grayscale features;
[0152] Using the extracted features, an intelligent detection model is constructed and trained using deep learning algorithms;
[0153] Input the ultrasound image of the thyroid gland to be detected, use the trained model to detect nodules, and output the detection results.
[0154] Training intelligent detection models with a large amount of historical patient test data can improve both the accuracy and efficiency of test results.
[0155] The use of the same-position detection strategy specifically includes:
[0156] Obtain the coordinates of the lower end of the ultrasonic testing instrument and record them as the testing coordinates;
[0157] Determine whether the detection coordinates exist in the set of detection angles corresponding to thyroid nodules;
[0158] If it exists, obtain the maximum and minimum angles in the corresponding set of detection angles, and adjust the ultrasonic detector from the minimum angle to the maximum angle in sequence;
[0159] If not found, obtain the maximum and minimum angles from all isotopic detection angle sets in the thyroid nodule isotopic detection angle set. Compare all the maximum and minimum angles to find the final maximum and minimum angles. Use the final minimum angle to adjust the detection angles sequentially towards the final maximum angle. Select the angles corresponding to the clearest images as the operating angles of the ultrasound detector under this detection coordinate. If none of the acquired images are clear, decrease the final minimum angle and increase the final maximum angle until a clear image is found. Use the angle corresponding to the clear image as the operating angle of the ultrasound detector under this detection coordinate.
[0160] By using elements from the set of isotopic detection angles to detect thyroid nodules in new patients, the efficiency of thyroid nodule detection is improved. By comparing different images to find clear images, the detection angle corresponding to the clear image is used as the detection angle for this patient, thus improving the accuracy of the detection angle.
[0161] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0162] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A historical data-driven intelligent ultrasound detection method for thyroid nodules, characterized in that, include: Obtain the patient's historical medical records; Determine whether a patient is a repeat patient based on their past medical history. If the patient is not a previously diagnosed patient, the new patient testing strategy will be used to detect thyroid nodules. If the patient is a long-term patient, an intelligent detection method will be used to detect thyroid nodules. The patient's test results and the testing angle used by the testing instrument during the test are used to create and save a corresponding dataset through a data processing program; Based on the size of the thyroid nodules, the examination results of patients with thyroid nodules and the angles used by the instruments during the examination were classified to form a thyroid nodule patient detection dataset. A deep learning algorithm was used to build an intelligent detection model and train it. Establish all angles used by ultrasound detection instruments at the same coordinate points, and denot them as the isotopic detection angle set. Combine all the isotopic detection angle sets into the thyroid nodule isotopic detection angle set. The test results are displayed on the user interface, and a diagnostic report is generated.
2. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 1, characterized in that, The method of determining whether a patient is a returning patient based on their past medical history specifically includes: Set the treatment duration threshold to T; Set the threshold for the frequency of medical visits to P; Obtain the patient's first and last visit dates; Get the number of patient visits. If no patient visits are found, record the number of patient visits as 0. Obtain the patient's treatment duration. If no treatment time is obtained, record the patient's treatment duration as 0. The patient's total interval between visits = the date of the last visit - the date of the first visit. If the patient's first visit date and last visit date are not available, the patient's total interval between visits is recorded as 1. Patient visit frequency = Number of patient visits / Total interval between patient visits; If a patient's consecutive consultation time exceeds T or the patient's consultation frequency exceeds P, then this patient is considered a long-term patient. Otherwise, this patient will be classified as a new patient.
3. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 1, characterized in that, The new patient detection strategy for thyroid nodules specifically includes: The patient's medical history was taken to determine the location of the thyroid nodules; Obtain the user's Adam's apple and epiglottis, and denote them as the Adam's apple point and epiglottis point, respectively; Starting from the laryngeal protuberance, draw a ray through the epiglottis, and denote the direction of the ray as the left direction; Place the patient flat on the testing table, and mark the direction the patient's face is facing as "up". With the laryngeal protuberance as the origin, draw a ray passing through the origin and the epiglottis. This ray is denoted as the Y-axis, and the leftward direction is denoted as the positive direction of the Y-axis. Draw a horizontal ray perpendicular to the Y-axis and pointing towards the right thyroid gland through the origin, which is denoted as the X-axis, and the direction of the ray is denoted as the positive direction of the X-axis. Draw a vertical ray perpendicular to the Y-axis through the origin, which is denoted as the Z-axis, and the upward direction is denoted as the positive direction of the Z-axis. Thyroid nodules are detected in patients using ultrasound equipment; The end of the ultrasound instrument that contacts the human body is designated as the lower end, and the other end is designated as the upper end; The coordinates of the lower end of the ultrasonic instrument in the three-dimensional coordinate system are obtained and denoted as the lower end coordinates. Detection is performed using a same-position detection strategy; The patient's ultrasound images and the tilt angle of the ultrasound instrument during the examination are uploaded to the data processing program.
4. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 1, characterized in that, The method of using intelligent detection to detect thyroid nodules in patients specifically includes: Obtain the data set of the ultrasound instrument usage angle during the patient's most recent thyroid nodule examination, and denote it as the patient's ultrasound instrument usage angle data set. Obtain a dataset of ultrasound instrument usage angles during examinations of similar patients, and denote it as the dataset of ultrasound instrument usage angles for similar patients. First, use the elements in the patient's ultrasound instrument usage angle dataset to perform the initial detection in sequence; Obtain elements from the similar patient ultrasound instrument usage angle dataset that are not in the patient ultrasound instrument usage angle dataset, and form a secondary detection ultrasound instrument detection angle dataset. The patient was subjected to a secondary examination using a secondary ultrasound instrument to detect angle datasets. The patient's ultrasound images and the tilt angle of the ultrasound instrument during the examination are uploaded to the data processing program.
5. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 1, characterized in that, The process of creating and saving a dataset based on the patient's test results and the testing angle used by the testing instrument during the test through a data processing program includes: Obtain the patient's ID number; Obtain the patient's appointment date; Obtain the angle at which the ultrasound instrument is used during the patient's examination; Acquire patient ultrasound images; Determine if the patient is a returning patient; If the patient is not a returning patient, create a root folder named after the patient's ID number and a subfolder named after the patient's visit date. Create a file named "Ultrasound Angle File" to save the ultrasound angles used during the patient's examination. Save the image files with the corresponding ultrasound angle names. If it is a returning patient, find the root folder based on the patient's ID number, create a second-level folder named after the patient's visit date, create a file named "Ultrasound Instrument Angle File" to save the ultrasound instrument angle used during the patient's examination, and save the image files with the corresponding ultrasound instrument angle file names.
6. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 1, characterized in that, The dataset is categorized based on the size of the thyroid nodules, classifying the examination results and the angles used by the instruments during testing to form a dataset of thyroid nodule patient examinations. This dataset specifically includes: Obtain the size of the thyroid nodule diagnosed in the patient and record it as the thyroid nodule size; The obtained thyroid nodule size is compared with the thyroid nodule patient detection dataset to determine whether the thyroid nodule size category exists in the thyroid nodule patient detection dataset. If it does not exist, create a new entry in the thyroid nodule patient detection dataset named after the obtained thyroid nodule size, and save the detection angle used by the ultrasound instrument during the detection of this patient; If it exists, determine whether the angle used by the ultrasound instrument during the examination of this patient exists in the corresponding entry in the dataset of thyroid nodule patients. If it does not exist, add this angle data to the corresponding entry. The processed entries are used to create a dataset of ultrasound instrument usage angles for patients of the same type.
7. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 1, characterized in that, All angles used by the ultrasonic testing instruments to establish the same coordinate points form a set of corresponding testing angles, specifically including: Obtain the coordinates of the lower endpoint of the ultrasonic testing instrument during testing, and record them as the testing contact coordinates; Obtain the detection angles used by the ultrasonic testing instrument for detecting contact coordinates, and form a set of detection angles; Determine whether the detection contact coordinates exist in the set of detection angles at the same location as the thyroid nodule; If an element exists in the set of isotopic detection angles for thyroid nodules, then determine whether all elements in the detection angle set exist in the corresponding isotopic detection angle set. If an element exists but does not exist in the corresponding isotopic detection angle set, then add it to the corresponding isotopic detection angle set. If the thyroid nodule does not exist in the isotopic detection angle set, a new isotopic detection angle set is created with the detection contact point coordinates as the name, and all elements in the detection angle set are added to the new isotopic detection angle set.
8. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 1, characterized in that, The process of constructing and training an intelligent detection model using deep learning algorithms specifically includes: Obtain data from different entries in the dataset of patients with thyroid nodules; The acquired data is preprocessed, including noise removal, image enhancement, and standardization. Multiple features are extracted from the preprocessed data, including morphological features, texture features, and grayscale features; Using the extracted features, an intelligent detection model is constructed and trained using deep learning algorithms; Input the ultrasound image of the thyroid gland to be detected, use the trained model to detect nodules, and output the detection results.
9. The intelligent ultrasound detection method for thyroid nodules based on historical data as described in claim 3, characterized in that, The use of the same-position detection strategy specifically includes: Obtain the coordinates of the lower end of the ultrasonic testing instrument and record them as the testing coordinates; Determine whether the detection coordinates exist in the set of detection angles corresponding to thyroid nodules; If it exists, obtain the maximum and minimum angles in the corresponding set of detection angles, and adjust the ultrasonic detector from the minimum angle to the maximum angle in sequence; If not found, obtain the maximum and minimum angles from all isotopic detection angle sets in the thyroid nodule isotopic detection angle set. Compare all the maximum and minimum angles to find the final maximum and minimum angles. Use the final minimum angle to adjust the detection angles sequentially towards the final maximum angle. Select the angles corresponding to the clearest images as the operating angles of the ultrasound detector under this detection coordinate. If none of the acquired images are clear, decrease the final minimum angle and increase the final maximum angle until a clear image is found. Use the angle corresponding to the clear image as the operating angle of the ultrasound detector under this detection coordinate.