TWI773534B - Tooth age prediction method and tooth age prediction system - Google Patents
Tooth age prediction method and tooth age prediction system Download PDFInfo
- Publication number
- TWI773534B TWI773534B TW110134790A TW110134790A TWI773534B TW I773534 B TWI773534 B TW I773534B TW 110134790 A TW110134790 A TW 110134790A TW 110134790 A TW110134790 A TW 110134790A TW I773534 B TWI773534 B TW I773534B
- Authority
- TW
- Taiwan
- Prior art keywords
- tooth
- target
- processing unit
- germ development
- image
- Prior art date
Links
Images
Landscapes
- Apparatus For Radiation Diagnosis (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
一種齒齡預測方法，包含：對於多個預定牙位的每一者，一處理單元根據多個參考牙齒影像及多筆參考牙位資料，訓練產生一牙位辨識模型；對於多個牙齒種類的每一者，且對於多個牙胚發育階段的每一者，該處理單元根據該等參考牙齒影像及多筆標籤資料，訓練產生一牙胚發育階段辨識模型；該處理單元根據相關於一目標患者的多個目標牙齒影像，使用該等牙位辨識模型產生多個牙位辨識結果；及對於每一目標牙齒影像，該處理單元根據該牙位辨識結果使用該等牙胚發育階段辨識模型產生多個牙胚發育階段辨識結果，並根據該等牙胚發育階段辨識結果產生一齒齡預測結果。A method for predicting tooth age, comprising: for each of a plurality of predetermined tooth positions, a processing unit is trained to generate a tooth position identification model according to a plurality of reference tooth images and a plurality of reference tooth position data; each, and for each of a plurality of tooth germ development stages, the processing unit trains to generate a tooth germ development stage identification model according to the reference tooth images and the plurality of label data; a plurality of target tooth images of the patient, using the tooth position identification models to generate a plurality of tooth position identification results; and for each target tooth image, the processing unit generates the tooth germ development stage identification models according to the tooth position identification results A plurality of tooth germ development stage identification results are generated, and a tooth age prediction result is generated according to the tooth germ development stage identification results.
Description
本發明是有關於一種預測方法，特別是指一種使用機器學習技術的齒齡預測方法。本發明還有關於一種齒齡預測系統。 The present invention relates to a prediction method, in particular to a tooth age prediction method using machine learning technology. The present invention also relates to a tooth age prediction system.
齒齡(dental age)於臨床上用來推估受測者實際年齡(chronological age)以及受測者齒列發展(dentitional development)的速率。過慢的發展速率可能暗示著全身性的發展遲緩，如呆小症或生長遲緩等；而過快的齒列發展速率，則可能出現在早熟、肥胖的孩童上。在顎骨發育仍未完全的狀態下，過快的齒列發展可能促使或惡化齒列不整的態勢，拉高臨床治療的難度。 Dental age is used clinically to estimate the chronological age of the subject and the rate of the subject's dentitional development. Too slow development rate may suggest systemic developmental delay, such as cretinism or growth retardation, while too fast dentition development rate may appear in precocious, obese children. In the state where the jawbone is still not fully developed, the rapid development of dentition may promote or exacerbate the situation of dentition irregularity, which increases the difficulty of clinical treatment.
一般的齒齡判讀方式，需要由專業醫師來判讀環口X光片(Panoramic film)。然而人為判讀存在不穩定性及主觀性。因此，如何利用人工智慧發展出一種新的齒齡預測方法，改善齒齡判讀的準確性及穩定性，遂成為本發明進一步要探討的主題。 The general way of interpreting dental age requires a professional physician to interpret the Panoramic film. However, human interpretation is unstable and subjective. Therefore, how to use artificial intelligence to develop a new tooth age prediction method to improve the accuracy and stability of tooth age interpretation has become a further subject of the present invention.
因此，本發明的目的，即在提供一種齒齡預測方法。 Therefore, the purpose of the present invention is to provide a method for predicting tooth age.
本發明的另一目的，即在提供一種齒齡預測系統。 Another object of the present invention is to provide a tooth age prediction system.
於是，本發明齒齡預測方法，藉由一齒齡預測系統實施，該齒齡預測系統包含一輸出單元及一處理單元，該方法包含：對於多個預定牙位的每一者，該處理單元根據相關於多位參考患者的多個參考牙齒影像其中包含一屬於該預定牙位的牙齒的影像者，及多筆分別對應於該等參考牙齒影像且指示該預定牙位的參考牙位資料，訓練一第一原始模型而產生一對應於該預定牙位的牙位辨識模型，每一預定牙位屬於多個牙齒種類其中一者，每一牙齒種類相關於多個牙胚發育階段；對於該等牙齒種類的每一者，且對於該牙齒種類所相關的該等牙胚發育階段的每一者，該處理單元根據該等參考牙齒影像其中所包含的該牙齒屬於該牙齒種類者，及多筆指示相似或不相似的標籤資料，以對比式學習，訓練一第二原始模型而產生一對應於該牙齒種類及該牙胚發育階段的牙胚發育階段辨識模型；該處理單元根據相關於該等參考患者的多筆齒齡預測訓練資料，訓練一第三原始模型而產生一齒齡預測模型，每一齒齡預測訓練資料包含多筆分別對應於該等預定牙位且指示該等牙胚發育階段其中一者的參考牙胚發育階段、一參考性別，及一參考齒齡；該處理單元根據相關於一目標患者的多個目標牙齒影像，使用對應於該等預定牙位的該等牙位辨識模型，產生多個分別對應於該等目標牙齒影像的牙位辨識結果，每一牙位辨識結果指示該等預定牙位其 中一者；對於每一目標牙齒影像，該處理單元根據該目標牙齒影像及對應的該牙位辨識結果，使用對應於該牙位辨識結果所屬的該牙齒種類的該等牙胚發育階段辨識模型，產生多個對應於該目標牙齒影像且分別相關於該等牙胚發育階段的牙胚發育階段辨識結果；對於每一目標牙齒影像，該處理單元根據對應的該等牙胚發育階段辨識結果，產生一對應於該目標牙齒影像的目標牙胚發育階段；及該處理單元根據對應於該等目標牙齒影像的該等目標牙胚發育階段及一相關於該目標患者的目標性別，使用該齒齡預測模型，產生一齒齡預測結果，並經由該輸出單元輸出該齒齡預測結果。 Therefore, the tooth age prediction method of the present invention is implemented by a tooth age prediction system, the tooth age prediction system includes an output unit and a processing unit, and the method includes: for each of a plurality of predetermined tooth positions, the processing unit According to a plurality of reference tooth images related to a plurality of reference patients including an image of a tooth belonging to the predetermined tooth position, and a plurality of reference tooth position data respectively corresponding to the reference tooth images and indicating the predetermined tooth position, training a first original model to generate a tooth position identification model corresponding to the predetermined tooth position, each predetermined tooth position belongs to one of a plurality of tooth types, and each tooth type is related to a plurality of tooth germ development stages; each of the tooth types, and for each of the tooth germ developmental stages associated with the tooth type, the tooth included in the processing unit according to the reference tooth images belongs to the tooth type, and more The pen indicates similar or dissimilar label data, and uses comparative learning to train a second original model to generate a tooth germ development stage identification model corresponding to the tooth type and the tooth germ development stage; the processing unit Referring to multiple sets of dental age prediction training data of the patient, a third original model is trained to generate a dental age prediction model, and each tooth age prediction training data includes multiple strokes corresponding to the predetermined tooth positions and indicating the tooth germs a reference tooth germ development stage of one of the developmental stages, a reference gender, and a reference tooth age; the processing unit uses the teeth corresponding to the predetermined tooth positions according to a plurality of target tooth images related to a target patient A position identification model, generating a plurality of tooth position identification results corresponding to the target tooth images, each tooth position identification result indicating the predetermined tooth position For each target tooth image, the processing unit uses the tooth germ development stage identification models corresponding to the tooth type to which the tooth position identification result belongs according to the target tooth image and the corresponding tooth position identification result , generating a plurality of identification results of the tooth germ development stages corresponding to the target tooth image and respectively related to the development stages of the tooth germ; for each target tooth image, the processing unit, according to the identification results of the corresponding development stages of the tooth germ, generating a target tooth germ development stage corresponding to the target tooth image; and the processing unit using the tooth age according to the target tooth germ development stage corresponding to the target tooth image and a target gender associated with the target patient The prediction model generates a tooth age prediction result, and outputs the tooth age prediction result through the output unit.
在一些實施態樣中，該處理單元於產生對應於該目標牙齒影像的該目標牙胚發育階段，是將該等牙胚發育階段辨識結果其中所指示的相似度最高者所相關的該牙胚發育階段作為該目標牙胚發育階段所指示的該牙胚發育階段。 In some embodiments, the processing unit generates the target tooth germ developmental stage corresponding to the target tooth image, and is the tooth germ associated with the tooth germ development stage identification results indicated by the highest similarity The developmental stage is the tooth germ developmental stage indicated by the target tooth germ developmental stage.
在一些實施態樣中，該處理單元是根據包含有該等目標牙齒影像的一目標全口影像，使用該牙位辨識模型，擷取出該等目標牙齒影像並產生該等牙位辨識結果。 In some implementations, the processing unit uses the tooth position recognition model to extract the target tooth images and generate the tooth position recognition results according to a target full-mouth image including the target tooth images.
本發明齒齡預測系統，包含一輸出單元及一處理單元。該處理單元電連接於該輸出單元。 The tooth age prediction system of the present invention includes an output unit and a processing unit. The processing unit is electrically connected to the output unit.
對於多個預定牙位的每一者，該處理單元根據相關於多位參考患者的多個參考牙齒影像其中包含一屬於該預定牙位的牙 齒的影像者，及多筆分別對應於該等參考牙齒影像且指示該預定牙位的參考牙位資料，訓練一第一原始模型而產生一對應於該預定牙位的牙位辨識模型，每一預定牙位屬於多個牙齒種類其中一者，每一牙齒種類相關於多個牙胚發育階段。 For each of a plurality of predetermined tooth positions, the processing unit includes a tooth belonging to the predetermined tooth position according to a plurality of reference tooth images associated with a plurality of reference patients The imager of the teeth, and a plurality of reference tooth position data corresponding to the reference tooth images and indicating the predetermined tooth position respectively, train a first original model to generate a tooth position identification model corresponding to the predetermined tooth position, each A predetermined tooth position belongs to one of a plurality of tooth types, and each tooth type is associated with a plurality of tooth germ development stages.
對於該等牙齒種類的每一者，且對於該牙齒種類所相關的該等牙胚發育階段的每一者，該處理單元根據該等參考牙齒影像其中所包含的該牙齒屬於該牙齒種類者，及多筆指示相似或不相似的標籤資料，以對比式學習，訓練一第二原始模型而產生一對應於該牙齒種類及該牙胚發育階段的牙胚發育階段辨識模型。 for each of the tooth types, and for each of the tooth germ developmental stages to which the tooth type is associated, the tooth included in the processing unit according to the reference tooth images belongs to that tooth type, and a plurality of labels indicating similarity or dissimilarity, through comparative learning, a second original model is trained to generate a tooth germ development stage identification model corresponding to the tooth type and the tooth germ development stage.
該處理單元根據相關於該等參考患者的多筆齒齡預測訓練資料，訓練一第三原始模型而產生一齒齡預測模型，每一齒齡預測訓練資料包含多筆分別對應於該等預定牙位且指示該等牙胚發育階段其中一者的參考牙胚發育階段、一參考性別，及一參考齒齡。 The processing unit trains a third original model to generate a dental age prediction model according to a plurality of dental age prediction training data related to the reference patients, and each tooth age prediction training data includes a plurality of teeth corresponding to the predetermined teeth respectively. and indicates a reference tooth germ development stage, a reference gender, and a reference tooth age for one of the tooth germ development stages.
該處理單元根據相關於一目標患者的多個目標牙齒影像，使用對應於該等預定牙位的該等牙位辨識模型，產生多個分別對應於該等目標牙齒影像的牙位辨識結果，每一牙位辨識結果指示該等預定牙位其中一者。 The processing unit uses the tooth position identification models corresponding to the predetermined tooth positions according to a plurality of target tooth images related to a target patient to generate a plurality of tooth position identification results corresponding to the target tooth images, each A tooth position identification result indicates one of the predetermined tooth positions.
對於每一目標牙齒影像，該處理單元根據該目標牙齒影像及對應的該牙位辨識結果，使用對應於該牙位辨識結果所屬的該牙齒種類的該等牙胚發育階段辨識模型，產生多個對應於該目標牙 齒影像且分別相關於該等牙胚發育階段的牙胚發育階段辨識結果。 For each target tooth image, the processing unit generates a plurality of tooth germ development stage identification models corresponding to the tooth type to which the tooth position identification result belongs according to the target tooth image and the corresponding tooth position identification result. corresponding to the target tooth The tooth images are respectively related to the identification results of the tooth germ development stages of the tooth germ development stages.
對於每一目標牙齒影像，該處理單元根據對應的該等牙胚發育階段辨識結果，產生一對應於該目標牙齒影像的目標牙胚發育階段。 For each target tooth image, the processing unit generates a target tooth germ development stage corresponding to the target tooth image according to the identification results of the corresponding tooth germ development stages.
該處理單元根據對應於該等目標牙齒影像的該等目標牙胚發育階段及一相關於該目標患者的目標性別，使用該齒齡預測模型，產生一齒齡預測結果，並經由該輸出單元輸出該齒齡預測結果。 The processing unit uses the tooth age prediction model according to the target tooth germ development stages corresponding to the target tooth images and a target gender related to the target patient to generate a tooth age prediction result, which is output through the output unit The tooth age prediction results.
本發明齒齡預測方法，藉由一齒齡預測系統實施，該齒齡預測系統包含一輸出單元及一處理單元，該方法包含：對於多個預定牙位的每一者，該處理單元根據相關於多位參考患者的多個參考牙齒影像其中包含一屬於該預定牙位的牙齒的影像者，及多筆分別對應於該等參考牙齒影像且指示該預定牙位的參考牙位資料，訓練一第一原始模型而產生一對應於該預定牙位的牙位辨識模型，每一預定牙位屬於多個牙齒種類其中一者，每一牙齒種類相關於多個牙胚發育階段；對於該等牙齒種類的每一者，且對於該牙齒種類所相關的該等牙胚發育階段的每一者，該處理單元根據該等參考牙齒影像其中所包含的該牙齒屬於該牙齒種類者，及多筆指示相似或不相似的標籤資料，以對比式學習，訓練一第二原始模型而產生一對應於該牙齒種類及該牙胚發育階段的牙胚發育階段辨識模型；該處理單元根據相關於一目標患者的多個目標牙齒影像，使用對應於該 等預定牙位的該等牙位辨識模型，產生多個分別對應於該等目標牙齒影像的牙位辨識結果，每一牙位辨識結果指示該等預定牙位其中一者；對於每一目標牙齒影像，該處理單元根據該目標牙齒影像及對應的該牙位辨識結果，使用對應於該牙位辨識結果所屬的該牙齒種類的該等牙胚發育階段辨識模型，產生多個對應於該目標牙齒影像且分別相關於該等牙胚發育階段的牙胚發育階段辨識結果；對於每一目標牙齒影像，該處理單元根據對應的該等牙胚發育階段辨識結果，產生一對應於該目標牙齒影像的目標牙胚發育階段；該處理單元根據對應於該等目標牙齒影像的該等目標牙胚發育階段，及一牙胚發育階段分數對照表，產生多個分別對應於該等目標牙齒影像的分數；該處理單元將對應於該等目標牙齒影像的該等分數加總產生一總分；及該處理單元根據該總分、一相關於該目標患者的目標性別，及多個總分齒齡對照表其中相關於該目標性別所指示的性別者，產生一齒齡預測結果，並經由該輸出單元輸出該齒齡預測結果。 The tooth age prediction method of the present invention is implemented by a tooth age prediction system, the tooth age prediction system includes an output unit and a processing unit, the method includes: for each of a plurality of predetermined tooth positions, the processing unit according to the relevant In a plurality of reference tooth images of a plurality of reference patients including an image of a tooth belonging to the predetermined tooth position, and a plurality of reference tooth position data corresponding to the reference tooth images and indicating the predetermined tooth position, training a The first original model generates a tooth position identification model corresponding to the predetermined tooth position, each predetermined tooth position belongs to one of a plurality of tooth types, and each tooth type is related to a plurality of tooth germ development stages; for these teeth each of the types, and for each of the tooth germ developmental stages to which the tooth type is associated, the processing unit according to the reference tooth images contained therein to which the tooth belongs to the tooth type, and multiple indications Similar or dissimilar label data, through comparative learning, a second original model is trained to generate a tooth germ development stage identification model corresponding to the tooth type and the tooth germ development stage; the processing unit is related to a target patient. of multiple target tooth images, using images corresponding to the The tooth position identification models of the predetermined tooth positions are generated, and a plurality of tooth position identification results corresponding to the target tooth images are generated, and each tooth position identification result indicates one of the predetermined tooth positions; for each target tooth an image, the processing unit generates a plurality of identification models corresponding to the target tooth using the tooth germ developmental stage identification models corresponding to the tooth type to which the tooth position identification result belongs, according to the target tooth image and the corresponding tooth position identification result. The images are respectively related to the identification results of the developmental stages of the tooth germs; for each target tooth image, the processing unit generates an image corresponding to the target tooth image according to the identification results of the corresponding developmental stages of the tooth germs. target tooth germ development stages; the processing unit generates a plurality of scores corresponding to the target tooth images according to the target tooth germ development stages corresponding to the target tooth images and a tooth germ development stage score comparison table; The processing unit adds up the scores corresponding to the target dental images to generate a total score; and the processing unit generates a total score according to the total score, a target gender related to the target patient, and a plurality of total score dental age comparison tables A tooth age prediction result is generated in relation to the gender indicated by the target gender, and the tooth age prediction result is output through the output unit.
本發明齒齡預測系統，包含一輸出單元及一處理單元。該處理單元電連接於該輸出單元。 The tooth age prediction system of the present invention includes an output unit and a processing unit. The processing unit is electrically connected to the output unit.
對於多個預定牙位的每一者，該處理單元根據相關於多位參考患者的多個參考牙齒影像其中包含一屬於該預定牙位的牙齒的影像者，及多筆分別對應於該等參考牙齒影像且指示該預定牙位的參考牙位資料，訓練一第一原始模型而產生一對應於該預定牙 位的牙位辨識模型，每一預定牙位屬於多個牙齒種類其中一者，每一牙齒種類相關於多個牙胚發育階段。 For each of a plurality of predetermined tooth positions, the processing unit is based on a plurality of reference tooth images associated with a plurality of reference patients including an image of a tooth belonging to the predetermined tooth position, and a plurality of pens corresponding to the reference teeth, respectively Tooth image and reference tooth position data indicating the predetermined tooth position, training a first original model to generate a corresponding tooth position of the predetermined tooth A tooth position identification model of each position, each predetermined tooth position belongs to one of a plurality of tooth types, and each tooth type is associated with a plurality of tooth germ development stages.
對於該等牙齒種類的每一者，且對於該牙齒種類所相關的該等牙胚發育階段的每一者，該處理單元根據該等參考牙齒影像其中所包含的該牙齒屬於該牙齒種類者，及多筆指示相似或不相似的標籤資料，以對比式學習，訓練一第二原始模型而產生一對應於該牙齒種類及該牙胚發育階段的牙胚發育階段辨識模型。 for each of the tooth types, and for each of the tooth germ developmental stages to which the tooth type is associated, the tooth included in the processing unit according to the reference tooth images belongs to that tooth type, and a plurality of labels indicating similarity or dissimilarity, through comparative learning, a second original model is trained to generate a tooth germ development stage identification model corresponding to the tooth type and the tooth germ development stage.
該處理單元根據相關於一目標患者的多個目標牙齒影像，使用對應於該等預定牙位的該等牙位辨識模型，產生多個分別對應於該等目標牙齒影像的牙位辨識結果，每一牙位辨識結果指示該等預定牙位其中一者。 The processing unit uses the tooth position identification models corresponding to the predetermined tooth positions according to a plurality of target tooth images related to a target patient to generate a plurality of tooth position identification results corresponding to the target tooth images, each A tooth position identification result indicates one of the predetermined tooth positions.
對於每一目標牙齒影像，該處理單元根據該目標牙齒影像及對應的該牙位辨識結果，使用對應於該牙位辨識結果所屬的該牙齒種類的該等牙胚發育階段辨識模型，產生多個對應於該目標牙齒影像且分別相關於該等牙胚發育階段的牙胚發育階段辨識結果。 For each target tooth image, the processing unit generates a plurality of tooth germ development stage identification models corresponding to the tooth type to which the tooth position identification result belongs according to the target tooth image and the corresponding tooth position identification result. The identification results of the tooth germ development stages corresponding to the target tooth image and respectively related to the tooth germ development stages.
對於每一目標牙齒影像，該處理單元根據對應的該等牙胚發育階段辨識結果，產生一對應於該目標牙齒影像的目標牙胚發育階段。 For each target tooth image, the processing unit generates a target tooth germ development stage corresponding to the target tooth image according to the identification results of the corresponding tooth germ development stages.
該處理單元根據對應於該等目標牙齒影像的該等目標牙胚發育階段，及一牙胚發育階段分數對照表，產生多個分別對應於 該等目標牙齒影像的分數。 The processing unit generates a plurality of target tooth germ development stages corresponding to the target tooth images and a comparison table of tooth germ development stage scores, respectively corresponding to The score for those target tooth images.
該處理單元將對應於該等目標牙齒影像的該等分數加總產生一總分。 The processing unit adds up the scores corresponding to the target tooth images to generate a total score.
該處理單元根據該總分、一相關於該目標患者的目標性別，及多個總分齒齡對照表其中相關於該目標性別所指示的性別者，產生一齒齡預測結果，並經由該輸出單元輸出該齒齡預測結果。 The processing unit generates a tooth age prediction result according to the total score, a target gender related to the target patient, and a plurality of total score dental age comparison tables related to the gender indicated by the target gender, and outputs the result through the output The unit outputs the tooth age prediction result.
本發明的功效在於：藉由訓練及使用該等牙位辨識模型產生該等牙位辨識結果，並訓練及使用該等牙胚發育階段辨識模型產生該等牙胚發育階段辨識結果，並藉由訓練及使用該齒齡預測模型產生該齒齡預測結果，或者，藉由使用該牙胚發育階段分數對照表及該等總分齒齡對照表產生該齒齡預測結果，從而能改善齒齡判讀的準確性及穩定性。 The effect of the present invention is: by training and using the tooth position identification models to generate the tooth position identification results, and training and using the tooth germ development stage identification models to generate the tooth germ development stage identification results, and by Train and use the tooth age prediction model to generate the tooth age prediction result, or generate the tooth age prediction result by using the tooth germ development stage score comparison table and the total score tooth age comparison table to generate the tooth age prediction result, thereby improving the interpretation of the tooth age accuracy and stability.
100:齒齡預測系統 100: Tooth Age Prediction System
1:輸出單元 1: output unit
2:處理單元 2: Processing unit
S01~S07:步驟 S01~S07: Steps
S11~S18:步驟 S11~S18: Steps
本發明的其他的特徵及功效，將於參照圖式的實施方式中清楚地呈現，其中：圖1是本發明齒齡預測方法的一第一實施例的一硬體連接關係示意圖；及圖2(包含圖2A及圖2B)是該第一實施例的一流程圖；圖3(包含圖3A及圖3B)是本發明齒齡預測方法的一第二實 施例的一流程圖。 Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, wherein: FIG. 1 is a schematic diagram of a hardware connection relationship of a first embodiment of a tooth age prediction method of the present invention; and FIG. 2 (including FIG. 2A and FIG. 2B ) is a flowchart of the first embodiment; FIG. 3 (including FIG. 3A and FIG. 3B ) is a second implementation of the tooth age prediction method of the present invention A flowchart of an embodiment.
在本發明被詳細描述之前，應當注意在以下的說明內容中，類似的元件是以相同的編號來表示。 Before the present invention is described in detail, it should be noted that in the following description, similar elements are designated by the same reference numerals.
參閱圖1與圖2，本發明齒齡預測方法的一第一實施例，藉由一齒齡預測系統100實施，該齒齡預測系統100包含一輸出單元1，及一電連接於該輸出單元1的處理單元2。該輸出單元1例如為一顯示單元，但不以此為限。
Referring to FIGS. 1 and 2 , a first embodiment of the tooth age prediction method of the present invention is implemented by a tooth
該處理單元2可包含(但不限於)一單核處理器、一個多核處理器、一個雙核手機處理器、一微處理器、一微控制器、一數位訊號處理器(DSP)、一現場可程式邏輯閘陣列(FPGA)、一特殊應用積體電路(ASIC)及一射頻積體電路(RFIC)其中至少一者。
The
以下說明該第一實施例的流程。首先，如步驟S01所示，對於多個預定牙位的每一者，該處理單元2根據相關於多位參考患者的多個參考牙齒影像其中包含一屬於該預定牙位的牙齒的影像者，及多筆分別對應於該等參考牙齒影像且指示該預定牙位的參考牙位資料(由專業醫師提供)，訓練一第一原始模型(例如RetinaNet)而產生一對應於該預定牙位的牙位辨識模型，每一預
定牙位屬於多個牙齒種類其中一者，每一牙齒種類相關於多個牙胚發育階段。
The flow of this first embodiment will be described below. First, as shown in step S01, for each of a plurality of predetermined tooth positions, the
該等預定牙位例如是由FDI牙位表示法表示的牙位31~38。該等參考牙齒影像可以是由一般的拍攝局部口腔的X光機拍攝產生，也可以是以全口X光機拍攝產生全口影像再由專業醫師圈選擷取產生，且在訓練前調整影像尺寸為78x78pixels，並灰階處理(Gray scale(0-255))。 The predetermined tooth positions are, for example, tooth positions 31 to 38 represented by the FDI tooth position notation. These reference tooth images can be captured by a general X-ray machine that captures part of the oral cavity, or they can be captured by a full-mouth X-ray machine to produce a full-mouth image and then selected and captured by professional physicians, and the images are adjusted before training. The size is 78x78pixels, and grayscale processing (Gray scale (0-255)).
該等牙齒種類分別為門齒(INCISORS)、犬齒(CANINE)、前臼齒(PREMOLAR)及臼齒(MOLAR)。前臼齒及臼齒相關的牙胚發育階段有牙胚發育階段A~H，門齒及犬齒相關的牙胚發育階段有牙胚發育階段C~H。 These tooth types are incisors (INCISORS), canines (CANINE), premolars (PREMOLAR) and molars (MOLAR). The tooth germ development stages related to premolars and molars include tooth germ development stages A to H, and the tooth germ development stages related to incisors and canines include tooth germ development stages C to H.
在本實施例中，該等參考患者的納入標準包含具有完整齒列及在2歲至20歲之間的男性和女性，排除條件包含先天缺牙、阻生牙、病理性牙齒萌發障礙(如囊腫、齒瘤等)、足以影響骨骼及牙齒萌發的顏面部外傷及齒顎矯正治療中。 In this embodiment, the inclusion criteria of these reference patients include males and females with complete dentition and between the ages of 2 and 20, and the exclusion conditions include congenitally missing teeth, impacted teeth, pathological tooth eruption disorders (such as Cysts, dental tumors, etc.), facial trauma and orthodontic treatment that can affect bone and tooth germination.
接著，如步驟S02所示，對於該等牙齒種類的每一者，且對於該牙齒種類所相關的該等牙胚發育階段的每一者，該處理單元2根據該等參考牙齒影像其中所包含的該牙齒屬於該牙齒種類者，及多筆指示相似或不相似的標籤資料(由專業醫師提供)，以對比式學習，訓練一第二原始模型(例如Siamese network)而產
生一對應於該牙齒種類及該牙胚發育階段的牙胚發育階段辨識模型。
Next, as shown in step S02, for each of the tooth types, and for each of the tooth germ development stages associated with the tooth type, the
接著，如步驟S03所示，該處理單元2根據相關於該等參考患者的多筆齒齡預測訓練資料，訓練一第三原始模型(例如support vector machine)而產生一齒齡預測模型，每一齒齡預測訓練資料包含多筆分別對應於該等預定牙位且指示該等牙胚發育階段其中一者的參考牙胚發育階段、一參考性別，及一參考齒齡。
Next, as shown in step S03, the
接著，如步驟S04所示，該處理單元2根據相關於一目標患者的多個目標牙齒影像，使用對應於該等預定牙位的該等牙位辨識模型，產生多個分別對應於該等目標牙齒影像的牙位辨識結果，每一牙位辨識結果指示該等預定牙位其中一者。
Next, as shown in step S04, the
在本實施例中，該處理單元2是根據包含有該等目標牙齒影像的一目標全口影像，使用該牙位辨識模型，擷取出該等目標牙齒影像並產生該等牙位辨識結果。
In this embodiment, the
接著，如步驟S05所示，對於每一目標牙齒影像，該處理單元2根據該目標牙齒影像及對應的該牙位辨識結果，使用對應於該牙位辨識結果所屬的該牙齒種類的該等牙胚發育階段辨識模型，產生多個對應於該目標牙齒影像且分別相關於該等牙胚發育階段的牙胚發育階段辨識結果。
Next, as shown in step S05, for each target tooth image, the
接著，如步驟S06所示，對於每一目標牙齒影像，該處
理單元2根據對應的該等牙胚發育階段辨識結果，產生一對應於該目標牙齒影像的目標牙胚發育階段。
Next, as shown in step S06, for each target tooth image, the
The
在本實施例中，該處理單元2於產生對應於該目標牙齒影像的該目標牙胚發育階段，是將該等牙胚發育階段辨識結果其中所指示的相似度最高者所相關的該牙胚發育階段作為該目標牙胚發育階段所指示的該牙胚發育階段。
In this embodiment, when the
接著，如步驟S07所示，該處理單元2根據對應於該等目標牙齒影像的該等目標牙胚發育階段及一相關於該目標患者的目標性別，使用該齒齡預測模型，產生一齒齡預測結果，並經由該輸出單元1輸出該齒齡預測結果。
Next, as shown in step S07, the
參閱圖1及圖3，以下說明本發明齒齡預測方法的一第二實施例的步驟。首先，步驟S11、S12、S13、S14、S15分別與該第一實施例的步驟S01、S02、S04、S05、S06相同，在此不予贅述。 Referring to FIG. 1 and FIG. 3 , the steps of a second embodiment of the tooth age prediction method of the present invention will be described below. First, steps S11 , S12 , S13 , S14 , and S15 are respectively the same as steps S01 , S02 , S04 , S05 , and S06 of the first embodiment, and will not be repeated here.
接著，如步驟S16所示，該處理單元2根據對應於該等目標牙齒影像的該等目標牙胚發育階段，及一牙胚發育階段分數對照表，產生多個分別對應於該等目標牙齒影像的分數。
Next, as shown in step S16, the
接著，如步驟S17所示，該處理單元2將對應於該等目標牙齒影像的該等分數加總產生一總分
Next, as shown in step S17, the
接著，如步驟S18所示，該處理單元2根據該總分、一相
關於該目標患者的目標性別，及多個總分齒齡對照表其中相關於該目標性別所指示的性別者，產生一齒齡預測結果，並經由該輸出單元1輸出該齒齡預測結果。步驟S16~S18是使用Demirjian’s method產生該齒齡預測結果。
Next, as shown in step S18, the
綜上所述，本發明齒齡預測方法藉由訓練及使用該等牙位辨識模型產生該等牙位辨識結果，並訓練及使用該等牙胚發育階段辨識模型產生該等牙胚發育階段辨識結果，並藉由訓練及使用該齒齡預測模型產生該齒齡預測結果，或者，藉由使用該牙胚發育階段分數對照表及該等總分齒齡對照表產生該齒齡預測結果，從而能改善齒齡判讀的準確性及穩定性，減少不同牙醫師間因人而異的判讀差異，並減少判讀所耗費的時間，進而能優化牙醫師與技師的工作流程，並且有助於跨世代的研究探討，故確實能達成本發明的目的。 To sum up, the tooth age prediction method of the present invention generates the tooth position identification results by training and using the tooth position identification models, and trains and uses the tooth germ development stage identification models to generate the tooth germ development stage identifications. As a result, the tooth age prediction result is generated by training and using the tooth age prediction model, or by using the tooth germ development stage score comparison table and the total score tooth age comparison table to generate the tooth age prediction result, thereby It can improve the accuracy and stability of dental age interpretation, reduce the differences in interpretation among different dentists, and reduce the time spent on interpretation, which can optimize the workflow of dentists and technicians, and help cross generations. Therefore, the purpose of the present invention can indeed be achieved.
惟以上所述者，僅為本發明的實施例而已，當不能以此限定本發明實施的範圍，凡是依本發明申請專利範圍及專利說明書內容所作的簡單的等效變化與修飾，皆仍屬本發明專利涵蓋的範圍內。 However, the above are only examples of the present invention, and should not limit the scope of implementation of the present invention. Any simple equivalent changes and modifications made according to the scope of the patent application of the present invention and the contents of the patent specification are still included in the scope of the present invention. within the scope of the invention patent.
S01~S07:步驟 S01~S07: Steps
Claims (12)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW110134790A TWI773534B (en) | 2021-09-17 | 2021-09-17 | Tooth age prediction method and tooth age prediction system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW110134790A TWI773534B (en) | 2021-09-17 | 2021-09-17 | Tooth age prediction method and tooth age prediction system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TWI773534B true TWI773534B (en) | 2022-08-01 |
| TW202312933A TW202312933A (en) | 2023-04-01 |
Family
ID=83806902
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW110134790A TWI773534B (en) | 2021-09-17 | 2021-09-17 | Tooth age prediction method and tooth age prediction system |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TWI773534B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105913424A (en) * | 2016-04-08 | 2016-08-31 | 北京大学口腔医院 | Tooth-based age estimation method and device |
| JP2020103449A (en) * | 2018-12-26 | 2020-07-09 | 株式会社Ｄｏｃｔｏｒｂｏｏｋ | Dental state determination device and dental state determination method |
-
2021
- 2021-09-17 TW TW110134790A patent/TWI773534B/en active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105913424A (en) * | 2016-04-08 | 2016-08-31 | 北京大学口腔医院 | Tooth-based age estimation method and device |
| JP2020103449A (en) * | 2018-12-26 | 2020-07-09 | 株式会社Ｄｏｃｔｏｒｂｏｏｋ | Dental state determination device and dental state determination method |
Also Published As
| Publication number | Publication date |
|---|---|
| TW202312933A (en) | 2023-04-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210118132A1 (en) | Artificial Intelligence System For Orthodontic Measurement, Treatment Planning, And Risk Assessment | |
| US9808326B2 (en) | 3D dentofacial system and method | |
| Hwang et al. | Three-dimensional soft tissue analysis for the evaluation of facial asymmetry in normal occlusion individuals | |
| Nguyen et al. | Alveolar bone segmentation in intraoral ultrasonographs with machine learning | |
| CN105232163A (en) | System and method for determining an orthodontic diagnostic analysis of a patient | |
| KR20100016180A (en) | Method for deriving shape information | |
| Nagpal et al. | Reliability of different facial measurements for determination of vertical dimension of occlusion in edentulous using accepted facial dimensions recorded from dentulous subjects | |
| CN114302690A (en) | Method and system for orthodontic treatment planning | |
| Niemiec | Feline dental radiography and radiology: a primer | |
| Tanikawa et al. | Clinical applicability of automated cephalometric landmark identification: Part I—patient‐related identification errors | |
| US20220218448A1 (en) | Method for aligning digital representations of a patient's jaws | |
| Shahen et al. | Palatal volume and area assessment on digital casts generated from cone-beam computed tomography scans | |
| Moufti et al. | Developing an artificial intelligence solution to autosegment the edentulous mandibular bone for implant planning | |
| Tanikawa et al. | Objective classification of nose–lip–chin profiles and their relation to dentoskeletal traits | |
| Zhao et al. | Evaluation of the stability of the palatal rugae using the three-dimensional superimposition technique following orthodontic treatment | |
| Paranhos et al. | Is there any association between facial type and mandibular dental arch form in subjects with normal occlusion? | |
| TWI773534B (en) | Tooth age prediction method and tooth age prediction system | |
| US10535203B2 (en) | Digital dental mesh segmentation method and digital dental mesh segmentation device | |
| US20230122558A1 (en) | Methods and systems for orthodontic treatment planning with virtual jaw articulator | |
| US20230121899A1 (en) | Automatic registration of intraoral surface and volumetric scan data for orthodontic treatment planning | |
| Lamba et al. | Artificial intelligence in modern dentistry | |
| Ozeki et al. | Change in clinical crown height in adult patients treated by means of the multi-bracket appliance with extraction of four first premolars | |
| Tekale et al. | A three-dimensional finite element analysis of mini-implant supported K-loop for maxillary molar distalization | |
| Nassif et al. | In vitro Macro-qualitative Comparison of Three Enamel Stripping Procedures: What is the Best Shape We can get? | |
| KR102474299B1 (en) | Method for fabricating digital denture and apparatus thereof |