TWI775563B

TWI775563B - Full-range fourier domain optical coherence tomography system

Info

Publication number: TWI775563B
Application number: TW110129680A
Authority: TW
Inventors: 鄭旭志; 黃俊銘; 漢龍陳; 廖柏閎; 邱原霆
Original assignee: 國立虎尾科技大學
Priority date: 2021-08-11
Filing date: 2021-08-11
Publication date: 2022-08-21
Also published as: TW202306544A

Abstract

Disclosed is a full-range Fourier domain optical coherence tomography system, which is for producing noise-free Fourier domain optical coherence tomography information based on original Fourier domain optical coherence tomography information, comprising a Fourier domain optical coherence tomography device for performing Fourier domain optical coherence tomography to generate the original Fourier domain optical coherence tomography information; a training noise-free information generating device for generating training noise-free Fourier domain optical coherence tomography information; an artificial intelligence training device for performing noise-removal processing on the training original Fourier domain optical coherence tomography information to repeatedly generate transitional Fourier domain optical coherence tomography information until a comparison-distinction result of a information-difference-comparison processing thereof is indistinguishable, and thereby being trained to get a noise-free information generating device.

Description

全域式頻域光學同調斷層掃描系統Global frequency domain optical coherence tomography system

本發明相關於一種頻域光學同調斷層掃描系統，特別是相關於一種基於人工智慧技術之全域式頻域光學同調斷層掃描系統。 The invention relates to a frequency domain optical coherence tomography scanning system, in particular to an all-domain frequency domain optical coherence tomography scanning system based on artificial intelligence technology.

光學同調斷層掃描技術(Optical Coherence Tomography,OCT)是一種高分辨率、非接觸、非破壞的生物醫學成像新方法，通過高分辨的探測樣品不同深度層面背向散射回來的光強度，可以得到樣品的深度資訊。此技術能對活體組織的斷層結構和生理功能進行觀察，它具備組織病理分析所需的高分辨力，能發揮病變早期診斷、過程監視和手術引導等臨床功能。 Optical Coherence Tomography (OCT) is a high-resolution, non-contact, non-destructive new biomedical imaging method. By detecting the backscattered light intensity at different depths of the sample with high resolution, the sample can be obtained. in-depth information. This technology can observe the tomographic structure and physiological function of living tissue. It has the high resolution required for histopathological analysis, and can play clinical functions such as early diagnosis of lesions, process monitoring and surgical guidance.

光學同調斷層掃描技術中的頻域光學同調斷層掃描(Fourier Domain Optical Coherence Tomography,FDOCT)之優點為成像速度快。然而，傳統的頻域光學同調斷層掃描於光譜儀所量測頻域上干涉訊號具有餘弦(cosine)函數時，在做反傅利葉轉換所得到時域上的干涉訊號，會讓訊號產生自相關雜訊(Autocorrelation Noise)與鏡像雜訊(Mirror Noise or Conjugate Artifacts)的問題，進而使得量測的樣品訊號模糊不清，對成像品質造成重大的影響而導致使用者難以判讀樣品的真實結構，或是容易產生誤判。在習知技術中，若欲解決前述之問題，則需要藉由複雜的光學系統、艱難的演算法，甚至需要多次量測來累積樣品的影像資訊，方能有效消除此等雜訊，而這對於使用者而言不僅成本高，複雜程度也高。 The advantage of frequency domain optical coherence tomography (Fourier Domain Optical Coherence Tomography, FDOCT) in optical coherence tomography is fast imaging speed. However, when the interference signal in the frequency domain measured by the spectrometer has a cosine function in the conventional frequency domain optical coherence tomography, the interference signal in the time domain obtained by inverse Fourier transform will cause the signal to produce autocorrelation noise. (Autocorrelation Noise) and mirror noise (Mirror Noise or Conjugate Artifacts), which makes the measured sample signal ambiguous, which has a significant impact on the image quality and makes it difficult for users to interpret the real structure of the sample, or it is easy to misjudgment occurs. in conventional technology In order to solve the aforementioned problems, it is necessary to use complex optical systems, difficult algorithms, and even multiple measurements to accumulate image information of the sample, so as to effectively eliminate such noise, which is very important for users. Not only is the cost high, but the complexity is also high.

因此，習知技術的頻域光學同調斷層掃描仍有其改良之必要。 Therefore, the conventional frequency domain optical coherence tomography still needs to be improved.

因此，本發明的目的即在提供一種全域式頻域光學同調斷層掃描系統，能夠以簡單、省時、成本相對低廉的方式消除傳統頻域光學同調斷層掃描系統所產生的雜訊。 Therefore, the purpose of the present invention is to provide a global frequency domain optical coherence tomography scanning system, which can eliminate the noise generated by the traditional frequency domain optical coherence tomography scanning system in a simple, time-saving and relatively low cost manner.

本發明為解決習知技術之問題所採用之技術手段係提供一種全域式頻域光學同調斷層掃描系統，係用以根據原始頻域光學同調斷層掃描資訊產生無雜訊頻域光學同調斷層掃描資訊，該全域式頻域光學同調斷層掃描系統包含：頻域光學同調斷層掃描裝置，經配置而用於進行頻域光學同調斷層掃描，以產生該原始頻域光學同調斷層掃描資訊；訓練用無雜訊資訊產生裝置，連接於該頻域光學同調斷層掃描裝置，接收該頻域光學同調斷層掃描裝置對於複數個訓練用樣本進行頻域光學同調斷層掃描所產生的訓練用原始頻域光學同調斷層掃描資訊，並據以產生相對應的訓練用無雜訊頻域光學同調斷層掃描資訊；人工智慧訓練裝置，係連接於該頻域光學同調斷層掃描裝置以及該訓練用無雜訊資訊產生裝置，該人工智慧訓練裝置包括生成單元以及判別單元，該生成單元係用以接收該訓練用原始頻域光學同調斷層掃描資訊，並對該訓練用原始頻域光學同調斷層掃描資訊進行去除雜訊處理而產生過渡頻域光學同調斷層掃描資訊，進而將該過渡頻域光學同調斷層掃描資訊傳輸至該判別單元，該判別單元係用以接收該訓練用無雜訊頻域光學同調斷層掃描資訊以及該過渡頻域光學同調斷層掃描資訊，並對該訓練用無雜訊頻域光學同調斷層掃描資訊以及該過渡頻域光學同調斷層掃描資訊進行資訊差異比對處理而產生比對分辨結果，並使該生成單元對該訓練用原始頻域光學同調斷層掃描資訊根據該比對分辨結果重新進行另一去除雜訊處理，進而產生次世代的該過渡頻域光學同調斷層掃描資訊，直至所產生的該過渡頻域光學同調斷層掃描資訊與該訓練用無雜訊頻域光學同調斷層掃描資訊的該比對分辨結果為該判別單元所無法分辨，該生成單元再各別對其餘複數個該訓練用樣本的該訓練用原始頻域光學同調斷層掃描資訊進行該去除雜訊處理，直至全部所產生之其餘複數個該過渡頻域光學同調斷層掃描資訊與其餘複數個該訓練用無雜訊頻域光學同調斷層掃描資訊的該比對分辨結果為該判別單元所無法分辨，而藉此將該生成單元訓練成為無雜訊資訊生成裝置；以及該無雜訊資訊生成裝置，連接於該頻域光學同調斷層掃描裝置，係接收該頻域光學同調斷層掃描裝置對於複數個待測檢體進行頻域光學同調斷層掃描所產生的該原始頻域光學同調斷層掃描資訊，並根據該原始頻域光學同調斷層掃描資訊而對應生成該無雜訊頻域光學同調斷層掃描資訊。 The technical means adopted by the present invention to solve the problems of the prior art is to provide a global frequency-domain optical coherence tomography system for generating noise-free frequency-domain optical coherence tomography information according to the original frequency-domain optical coherence tomography information , the global frequency-domain optical coherence tomography system comprises: a frequency-domain optical coherence tomography device configured to perform frequency-domain optical coherence tomography to generate the original frequency-domain optical coherence tomography information; The information generation device is connected to the frequency-domain optical coherence tomography device, and receives the original frequency-domain optical coherence tomography scan for training generated by the frequency-domain optical coherence tomography scan for a plurality of training samples. information, and generate corresponding training-use noise-free frequency-domain optical coherence tomography information; an artificial intelligence training device is connected to the frequency-domain optical coherence tomography device and the training-use noise-free information generation device, the The artificial intelligence training device includes a generating unit and a judging unit, the generating unit is used for receiving the original frequency-domain optical coherence tomography information for training, and performing noise removal processing on the original frequency-domain optical coherence tomography information for training to generate Transition frequency domain optical coherence tomography information, and then transmit the transition frequency domain optical coherence tomography information to the discriminating unit, the discriminating unit The element system is used for receiving the training noise-free frequency-domain optical coherence tomography information and the transition frequency-domain optical coherence tomography information, and for the training noise-free frequency-domain optical coherence tomography information and the transition frequency-domain optical coherence tomography The coherence tomography information is subjected to information difference comparison processing to generate a comparison and discrimination result, and the generating unit performs another noise removal process on the original frequency-domain optical coherence tomography information for training according to the comparison and discrimination result, and further Generating the next generation of the transition frequency domain optical coherence tomography information, until the comparison and discrimination result between the generated transition frequency domain optical coherence tomography information and the training noise-free frequency domain optical coherence tomography information is the discrimination The generating unit further performs the denoising process on the original frequency-domain optical coherence tomography information for the training of the remaining plurality of training samples, until all the remaining plurality of the transition frequency domains are generated. The comparison and discrimination result of the optical coherence tomography information and the other plurality of noise-free frequency-domain optical coherence tomography information for training cannot be distinguished by the discriminating unit, thereby training the generating unit to generate noise-free information device; and the noise-free information generating device, connected to the frequency domain optical coherence tomography device, for receiving the frequency domain optical coherence tomography scanning device for a plurality of objects to be tested. The original frequency-domain optical coherence tomography information is generated, and the noise-free frequency-domain optical coherence tomography information is correspondingly generated according to the original frequency-domain optical coherence tomography information.

在本發明的一實施例中係提供一種全域式頻域光學同調斷層掃描系統，該頻域光學同調斷層掃描裝置所產生之該訓練用原始頻域光學同調斷層掃描資訊以及該訓練用無雜訊資訊產生裝置所產生之該訓練用無雜訊頻域光學同調斷層掃描資訊皆屬於一維圖片資訊。 In an embodiment of the present invention, a global frequency-domain optical coherence tomography scanning system is provided, the training-use original frequency-domain optical coherence tomography scanning information generated by the frequency-domain optical coherence tomography scanning device and the training-use noise-free The training-use noise-free frequency-domain optical coherence tomography information generated by the information generating device belongs to one-dimensional image information.

在本發明的一實施例中係提供一種全域式頻域光學同調斷層掃描系統，該全域式頻域光學同調斷層掃描系統更包括維度轉換裝置，該維度轉換裝置係與該頻域光學同調斷層掃描裝置、該訓練用無雜訊資訊產生裝置以及該人工智慧訓練裝置相互資訊連接，用以接收該訓練用原始頻域光學同調斷層掃描資訊以及該訓練用無雜訊頻域光學同調斷層掃描資訊，並將該訓練用原始頻域光學同調斷層掃描資訊以及該訓練用無雜訊頻域光學同調斷層掃描資訊之維度予以轉換後而轉為二維圖片資訊，再將維度轉為二維圖片資訊之該訓練用原始頻域光學同調斷層掃描資訊傳輸至該生成單元，以及將維度轉為二維圖片資訊之該訓練用無雜訊頻域光學同調斷層掃描資訊傳輸至該判別單元。 In an embodiment of the present invention, a global frequency domain optical coherence tomography scanning system is provided. The global frequency domain optical coherence tomography scanning system further includes a dimension conversion device, and the dimension conversion device is connected to the frequency domain optical coherence tomography scanning system. device, the noise-free information generating device for training, and the The artificial intelligence training devices are connected to each other for information to receive the original frequency-domain optical coherence tomography information for training and the noise-free frequency-domain optical coherence tomography information for training, and transmit the original frequency-domain optical coherence tomography information for training And the dimension of the training-use noise-free frequency-domain optical coherence tomography information is converted into two-dimensional image information, and then the original frequency-domain optical coherence tomography information used for training is transmitted to the two-dimensional image information. The generating unit, and the training noise-free frequency-domain optical coherence tomography information for converting dimensions into two-dimensional image information are transmitted to the discriminating unit.

在本發明的一實施例中係提供一種全域式頻域光學同調斷層掃描系統，該無雜訊資訊生成裝置所生成之該無雜訊頻域光學同調斷層掃描資訊係屬於二維圖片資訊，該無雜訊頻域光學同調斷層掃描資訊遂藉由該維度轉換裝置而將該無雜訊頻域光學同調斷層掃描資訊之維度轉為一維圖片資訊。 In an embodiment of the present invention, a global frequency-domain optical coherence tomography scanning system is provided. The noise-free frequency-domain optical coherence tomography scanning information generated by the noise-free information generating device belongs to two-dimensional image information. The noise-free frequency-domain optical coherence tomography information is then converted from the dimension of the noise-free frequency-domain optical coherence tomography information into one-dimensional picture information by the dimension conversion device.

在本發明的一實施例中係提供一種全域式頻域光學同調斷層掃描系統，該比對分辨結果係為該訓練用無雜訊頻域光學同調斷層掃描資訊以及該過渡頻域光學同調斷層掃描資訊之資訊差異度。 In an embodiment of the present invention, a global frequency-domain optical coherence tomography system is provided, and the comparison and resolution result is the noise-free frequency-domain optical coherence tomography scan information for training and the transition frequency-domain optical coherence tomography scan The degree of information discrepancy of information.

在本發明的一實施例中係提供一種全域式頻域光學同調斷層掃描系統，該訓練用無雜訊資訊產生裝置更包括鏡像雜訊去除單元，用於消除該訓練用原始頻域光學同調斷層掃描資訊中的自相關雜訊以及鏡像雜訊，使該訓練用無雜訊資訊產生裝置產生相對應於該訓練用原始頻域光學同調斷層掃描資訊且不包含自相關雜訊以及鏡像雜訊之該訓練用無雜訊頻域光學同調斷層掃描資訊。 In an embodiment of the present invention, a global frequency-domain optical coherence tomography scanning system is provided. The noise-free information generating device for training further includes an image noise removing unit for eliminating the original frequency-domain optical coherence tomography for training. The autocorrelation noise and the image noise in the scanning information make the noise-free information generating device for training generate corresponding to the original frequency domain optical coherence tomography scanning information for training and do not contain the autocorrelation noise and the image noise. The training uses noise-free frequency-domain optical coherence tomography information.

經由本發明的該全域式頻域光學同調斷層掃描系統所採用之技術手段，使得使用者得以藉由人工智慧技術而將進行頻域光學同調斷層掃描後所得出的一維圖片資訊中之自相關雜訊以及鏡像雜訊完整的消除。本發明的該全域式頻域光學同調斷層掃描系統不僅能夠節省許多光學系統的成本以及整個光學同調斷層掃描技術的建置成本，亦能夠降低光學系統的複雜度、有效減少產生無雜訊頻域光學同調斷層掃描資訊的所需時間。 Through the technical means adopted by the global frequency domain optical coherence tomography system of the present invention, the user can use artificial intelligence technology to perform autocorrelation in the one-dimensional image information obtained after frequency domain optical coherence tomography scanning Noise and image noise are completely eliminated. The whole of the present invention The frequency domain optical coherence tomography system can not only save the cost of many optical systems and the construction cost of the entire optical coherence tomography technology, but also reduce the complexity of the optical system and effectively reduce the generation of noise-free frequency domain optical coherence tomography. time required for information.

100:全域式頻域光學同調斷層掃描系統 100: Global Frequency Domain Optical Coherence Tomography System

100a:全域式頻域光學同調斷層掃描系統 100a: Global Frequency Domain Optical Coherence Tomography System

1:頻域光學同調斷層掃描裝置 1: Frequency domain optical coherence tomography device

2:訓練用無雜訊資訊產生裝置 2: Noise-free information generation device for training

21:鏡像雜訊去除單元 21: Image noise removal unit

3:人工智慧訓練裝置 3: artificial intelligence training device

31:生成單元 31: Generate unit

31a:無雜訊資訊生成裝置 31a: Noise-free information generation device

32:判別單元 32: discriminant unit

4:維度轉換裝置 4: Dimension conversion device

CP:資訊差異比對處理 CP: information difference comparison processing

CR:比對分辨結果 CR: Comparison and Resolution Results

DT:預設差異度閾值 DT: preset difference threshold

NI:無雜訊頻域光學同調斷層掃描資訊 NI: Noise-Free Frequency Domain Optical Coherence Tomography Information

NP:去除雜訊處理 NP: Noise Removal Processing

OI:原始頻域光學同調斷層掃描資訊 OI: Raw Frequency Domain Optical Coherence Tomography Information

S:待測檢體 S: object to be tested

TI:過渡頻域光學同調斷層掃描資訊 TI: Transition Frequency Domain Optical Coherence Tomography Information

TS:訓練用樣本 TS: training samples

TNI:訓練用無雜訊頻域光學同調斷層掃描資訊 TNI: Noise-Free Frequency Domain Optical Coherence Tomography Information for Training

TOI:訓練用原始頻域光學同調斷層掃描資訊 TOI: Raw Frequency Domain Optical Coherence Tomography Information for Training

FDOCT:頻域光學同調斷層掃描 FDOCT: Frequency Domain Optical Coherence Tomography

第1a圖為顯示根據本發明的一實施例的全域式頻域光學同調斷層掃描系統的方塊示意圖；第1b圖為顯示根據本發明的實施例的全域式頻域光學同調斷層掃描系統的運作流程示意圖；第2a圖為顯示根據本發明的另一實施例的全域式頻域光學同調斷層掃描系統的方塊示意圖；第2b圖為顯示根據本發明的實施例的全域式頻域光學同調斷層掃描系統的運作流程示意圖；第3a圖為顯示根據本發明的實施例的全域式頻域光學同調斷層掃描系統的訓練用原始頻域光學同調斷層掃描資訊之一維圖片資訊示意圖；第3b圖為顯示根據本發明的實施例的全域式頻域光學同調斷層掃描系統的訓練用原始頻域光學同調斷層掃描資訊之二維圖片資訊示意圖；第4a圖為顯示根據本發明的實施例的全域式頻域光學同調斷層掃描系統的訓練用無雜訊頻域光學同調斷層掃描資訊之一維圖片資訊示意圖；第4b圖為顯示根據本發明的實施例的全域式頻域光學同調斷層掃描系統的訓練用無雜訊頻域光學同調斷層掃描資訊之二維圖片資訊示意圖。 FIG. 1a is a schematic block diagram showing a global frequency-domain optical coherence tomography scanning system according to an embodiment of the present invention; FIG. 1b is a schematic diagram showing the operation flow of the global frequency-domain optical coherence tomography scanning system according to an embodiment of the present invention Schematic diagram; Fig. 2a is a block diagram showing a global frequency domain optical coherence tomography system according to another embodiment of the present invention; Fig. 2b is a schematic diagram showing a global frequency domain optical coherence tomography scanning system according to an embodiment of the present invention Figure 3a is a schematic diagram showing the one-dimensional picture information of the original frequency domain optical coherence tomography scanning information for training of the global frequency domain optical coherence tomography system according to an embodiment of the present invention; A schematic diagram of two-dimensional picture information of the original frequency-domain optical coherence tomography scanning information for training of the global frequency-domain optical coherence tomography system according to the embodiment of the present invention; FIG. A schematic diagram of one-dimensional picture information of noise-free frequency-domain optical coherence tomography information for training of the coherence tomography system; Figure 4b shows the noise-free frequency-domain optical coherence tomography system for training according to an embodiment of the present invention. Schematic diagram of 2D image information of frequency domain optical coherence tomography information.

以下根據第1a圖至第4b圖，而說明本發明的實施方式。該說明並非為限制本發明的實施方式，而為本發明之實施例的一種。 Embodiments of the present invention will be described below with reference to FIGS. 1a to 4b. This description is not intended to limit the embodiments of the present invention, but is an example of the present invention.

如第1a圖至第1b圖所示，依據本發明的一實施例的一種全域式頻域光學同調斷層掃描系統100，係用以根據原始頻域光學同調斷層掃描資訊OI產生無雜訊頻域光學同調斷層掃描資訊NI，該全域式頻域光學同調斷層掃描系統100包含：頻域光學同調斷層掃描裝置1，經配置而用於進行頻域光學同調斷層掃描FDOCT，以產生該原始頻域光學同調斷層掃描資訊OI；訓練用無雜訊資訊產生裝置2，連接於該頻域光學同調斷層掃描裝置1，接收該頻域光學同調斷層掃描裝置1對於複數個訓練用樣本TS進行頻域光學同調斷層掃描FDOCT所產生的訓練用原始頻域光學同調斷層掃描資訊TOI，並據以產生相對應的訓練用無雜訊頻域光學同調斷層掃描資訊TNI；人工智慧訓練裝置3，係連接於該頻域光學同調斷層掃描裝置1以及該訓練用無雜訊資訊產生裝置2，該人工智慧訓練裝置3包括生成單元31以及判別單元32，該生成單元31係用以接收該訓練用原始頻域光學同調斷層掃描資訊TOI，並對該訓練用原始頻域光學同調斷層掃描資訊TOI進行去除雜訊處理NP而產生過渡頻域光學同調斷層掃描資訊TI，進而將該過渡頻域光學同調斷層掃描資訊TI傳輸至該判別單元32，該判別單元32係用以接收該訓練用無雜訊頻域光學同調斷層掃描資訊TNI以及該過渡頻域光學同調斷層掃描資訊TI，並對該訓練用無雜訊頻域光學同調斷層掃描資訊TNI以及該過渡頻域光學同調斷層掃描資訊TI進行資訊差異比對處理CP而產生比對分辨結果CR，並使該生成單元31對該訓練用原始頻域光學同調斷層掃描資訊TOI根據該比對分辨結果CR重新進行另一去除雜訊處理NP，進而產生次世代的該過渡頻域光學同調斷層掃描資訊TI，直至所產生的該過渡頻域光學同調斷層掃描資訊TI與該訓練用無雜訊頻域光學同調斷層掃描資訊TNI的該比對分辨結果CR為該判別單元32所無法分辨，該生成單元31再各別對其餘複數個該訓練用樣本TS的該訓練用原始頻域光學同調斷層掃描資訊TOI進行該去除雜訊處理NP，直至全部所產生之其餘複數個該過渡頻域光學同調斷層掃描資訊TI與其餘複數個該訓練用無雜訊頻域光學同調斷層掃描資訊TNI的該比對分辨結果CR為該判別單元32所無法分辨，而藉此將該生成單元31訓練成為無雜訊資訊生成裝置31a。 As shown in FIGS. 1a to 1b, a global frequency-domain optical coherence tomography system 100 according to an embodiment of the present invention is used to generate a noise-free frequency domain according to the original frequency-domain optical coherence tomography information OI Optical coherence tomography information NI, the global frequency-domain optical coherence tomography system 100 includes: a frequency-domain optical coherence tomography device 1 configured to perform a frequency-domain optical coherence tomography FDOCT to generate the original frequency-domain optical coherence tomography The coherence tomography information OI; the noise-free information generation device 2 for training is connected to the frequency-domain optical coherence tomography device 1, and receives the frequency-domain optical coherence tomography device 1 to perform frequency-domain optical coherence on a plurality of training samples TS The original frequency-domain optical coherence tomography information TOI for training generated by the tomographic FDOCT is used to generate the corresponding training-use noise-free frequency-domain optical coherence tomography information TNI; the artificial intelligence training device 3 is connected to the frequency domain optical coherence tomography information TNI. Domain optical coherence tomography device 1 and the training-use noise-free information generating device 2, the artificial intelligence training device 3 comprises a generating unit 31 and a discriminating unit 32, the generating unit 31 is used for receiving the training original frequency domain optical coherence tomography information TOI, and the original frequency domain optical coherence tomography information TOI for training is subjected to noise removal processing NP to generate transition frequency domain optical coherence tomography information TI, and then the transition frequency domain optical coherence tomography information TI is transmitted To the discriminating unit 32, the discriminating unit 32 is used for receiving the training noise-free frequency-domain optical coherence tomography information TNI and the transition frequency-domain optical coherence tomography information TI, and for training the noise-free frequency domain optical coherence tomography information TNI The optical coherence tomography information TNI and the transition frequency domain optical coherence tomography information TI perform information difference comparison processing CP to generate a comparison and resolution result CR, and make the generating unit 31 perform the original frequency domain optical coherence tomography information for training. TOI distinguishes according to this alignment As a result, CR performs another noise removal process NP again, and then generates the next generation of the transition frequency domain optical coherence tomography information TI, until the generated transition frequency domain optical coherence tomography information TI and the training noise-free frequency domain. The comparison and discrimination result CR of the domain optical coherence tomography information TNI cannot be distinguished by the discriminating unit 32, and the generating unit 31 further scans the original frequency domain optical coherence tomography scans for the training of the remaining plurality of the training samples TS respectively. The information TOI performs the denoising processing NP until all the remaining plural pieces of the transition frequency-domain optical coherence tomography information TI are compared with the remaining plural pieces of the training noise-free frequency-domain optical coherence tomography information TNI. The discrimination result CR cannot be discriminated by the discriminating unit 32 , thereby training the generating unit 31 to become the noise-free information generating apparatus 31 a.

如第1b圖所示，根據本發明的實施例的該全域式頻域光學同調斷層掃描系統100，欲將該生成單元31訓練成為該無雜訊資訊生成裝置31a(如本發明的另一實施例以及第2a圖所示)，首先必須要將大量不同的該訓練用樣本TS各別輸入至該頻域光學同調斷層掃描裝置1。詳細而言，尚未完成針對該生成單元31之訓練的該全域式頻域光學同調斷層掃描系統100，係在將大量彼此之間相異的該訓練用樣本TS中的一個該訓練用樣本TS輸入至該頻域光學同調斷層掃描裝置1，且所產生的該過渡頻域光學同調斷層掃描資訊TI與該訓練用無雜訊頻域光學同調斷層掃描資訊TNI的該比對分辨結果CR為該判別單元32所無法分辨時，該全域式頻域光學同調斷層掃描系統100遂仍需要再將大量彼此之間相異的該訓練用樣本TS中的另一個該訓練用樣本TS輸入至該頻域光學同調斷層掃描裝置1，直至所產生的該過渡頻域光學同調斷層掃描資訊TI與該訓練用無雜訊頻域光學同調斷層掃描資訊TNI的該比對分辨結果CR為該判別單元32所無法分辨。如此重複上述之內容而將大量彼此之間相異的該訓練用樣本TS各別全數輸入至該頻域光學同調斷層掃描裝置1並使該比對分辨結果CR為該判別單元32所無法分辨後，此時之該生成單元31即訓練成為該無雜訊資訊生成裝置31a。 As shown in FIG. 1b, the global frequency-domain optical coherence tomography system 100 according to an embodiment of the present invention intends to train the generating unit 31 to become the noise-free information generating device 31a (as another implementation of the present invention). 2a), a large number of different training samples TS must be input into the frequency domain optical coherence tomography apparatus 1, respectively. In detail, the global frequency-domain optical coherence tomography system 100 that has not completed the training for the generating unit 31 is inputting one of the training samples TS among a large number of different training samples TS to the frequency domain optical coherence tomography device 1, and the comparison and resolution result CR of the transition frequency domain optical coherence tomography information TI and the training noise-free frequency domain optical coherence tomography information TNI is the judgment When the unit 32 cannot distinguish, the global frequency-domain optical coherence tomography system 100 still needs to input another training sample TS among a large number of different training samples TS into the frequency-domain optical coherence tomography system 100. The coherence tomography device 1, until the comparison and resolution result CR of the generated transition frequency domain optical coherence tomography information TI and the training noise-free frequency domain optical coherence tomography information TNI cannot be distinguished by the discriminating unit 32 . Repeating the above content in this way, a large number of the training samples TS that are different from each other are all input into the After the frequency domain optical coherence tomography apparatus 1 makes the comparison and discrimination result CR indistinguishable by the discrimination unit 32 , the generation unit 31 is then trained to become the noise-free information generation means 31 a.

如第1b圖所示，將大量彼此之間相異的該訓練用樣本TS中的一個該訓練用樣本TS輸入至該頻域光學同調斷層掃描裝置1後，該頻域光學同調斷層掃描裝置1遂針對所輸入之該訓練用樣本TS進行頻域光學同調斷層掃描FDOCT而產生該訓練用原始頻域光學同調斷層掃描資訊TOI。該訓練用無雜訊資訊產生裝置2遂根據該訓練用原始頻域光學同調斷層掃描資訊TOI而產生相對應的訓練用無雜訊頻域光學同調斷層掃描資訊TNI。該生成單元31再根據該訓練用原始頻域光學同調斷層掃描資訊TOI進行該去除雜訊處理NP而產生該過渡頻域光學同調斷層掃描資訊TI。該判別單元32係預先設定好一預設差異度閾值DT，該比對分辨結果CR係為該訓練用無雜訊頻域光學同調斷層掃描資訊TNI以及該過渡頻域光學同調斷層掃描資訊TI之資訊差異度，當該判別單元32對該訓練用無雜訊頻域光學同調斷層掃描資訊TNI以及該過渡頻域光學同調斷層掃描資訊TI進行該資訊差異比對處理CP而產生該比對分辨結果CR時，該判別單元32係依據該預設差異度閾值DT而決定下一個執行動作。當該比對分辨結果CR高於該預設差異度閾值DT時，該判別單元32係將高於該預設差異度閾值DT之該比對分辨結果CR輸出、回饋至該生成單元31，以使該生成單元31依據該比對分辨結果CR而調整生成該過渡頻域光學同調斷層掃描資訊TI之生成參數，該生成單元31遂而重新對該訓練用原始頻域光學同調斷層掃描資訊TOI進行另一去除雜訊處理NP。當該比對分辨結果CR係低於或等於該預設差異度閾值DT時，即，該判別單元32無法分辨該過渡頻域光學同調斷層掃描資訊TI與該訓練用無雜訊頻域光學同調斷層掃描資訊TNI之間的資訊差異度時，將通知該全域式頻域光學同調斷層掃描系統100，以將大量彼此之間相異的該訓練用樣本TS中的另一個該訓練用樣本TS輸入至該頻域光學同調斷層掃描裝置1，重覆上述之循環流程進而將該生成單元31成功訓練成為該無雜訊資訊生成裝置31a。 As shown in Fig. 1b, after inputting one of the training samples TS which are different from each other into the frequency domain optical coherence tomography apparatus 1, the frequency domain optical coherence tomography apparatus 1 Then, the frequency domain optical coherence tomography FDOCT is performed on the input training sample TS to generate the training original frequency domain optical coherence tomography information TOI. The training-use noise-free information generating device 2 then generates corresponding training noise-free frequency-domain optical coherence tomography information TNI according to the training original frequency-domain optical coherence tomography information TOI. The generating unit 31 then performs the noise removal process NP according to the original frequency-domain optical coherence tomography information TOI for training to generate the transition frequency-domain optical coherence tomography information TI. The discriminating unit 32 is preset with a predetermined difference threshold DT, and the comparison and discrimination result CR is the difference between the training noise-free frequency-domain optical coherence tomography information TNI and the transition frequency-domain optical coherence tomography information TI The degree of information difference, when the discriminating unit 32 performs the information difference comparison processing CP on the training noise-free frequency-domain optical coherence tomography information TNI and the transition frequency-domain optical coherence tomography information TI to generate the comparison and discrimination result In the case of CR, the determining unit 32 determines the next execution action according to the preset difference degree threshold DT. When the comparison and discrimination result CR is higher than the preset difference degree threshold DT, the determination unit 32 outputs and feeds back the comparison and discrimination result CR higher than the preset difference degree threshold DT to the generating unit 31 to The generating unit 31 adjusts the generating parameters for generating the transition frequency-domain optical coherence tomography information TI according to the comparison and discrimination result CR, and the generating unit 31 then re-runs the original frequency-domain optical coherence tomography information TOI for training. Another denoising process NP. When the comparison and discrimination result CR is lower than or equal to the preset difference threshold DT, that is, the discriminating unit 32 cannot discriminate the transition frequency domain optical coherence tomography information TI and the training noise-free frequency domain optical coherence When the information difference between the tomographic information TNIs, the global frequency-domain optical coherence tomography will be notified The system 100 is to input another training sample TS among a large number of the training samples TS that are different from each other to the frequency domain optical coherence tomography apparatus 1, repeat the above-mentioned cyclic process and then the generating unit 31 The successful training becomes the noise-free information generating device 31a.

如第2a圖至第2b圖所示，根據本發明的另一實施例的該全域式頻域光學同調斷層掃描系統100a所示，該無雜訊資訊生成裝置31a係連接於該頻域光學同調斷層掃描裝置1，係接收該頻域光學同調斷層掃描裝置1對於複數個待測檢體S進行頻域光學同調斷層掃描FDOCT所產生的該原始頻域光學同調斷層掃描資訊OI，並根據該原始頻域光學同調斷層掃描資訊OI而對應生成該無雜訊頻域光學同調斷層掃描資訊NI。 As shown in FIG. 2a to FIG. 2b, according to the global frequency-domain optical coherence tomography system 100a according to another embodiment of the present invention, the noise-free information generating device 31a is connected to the frequency-domain optical coherence tomography system 100a. The tomography device 1 receives the original frequency domain optical coherence tomography scan information OI generated by the frequency domain optical coherence tomography scan FDOCT performed by the frequency domain optical coherence tomography device 1 on a plurality of objects S to be tested, and according to the original frequency domain optical coherence tomography scan information OI The frequency-domain optical coherence tomography information OI is correspondingly generated to generate the noise-free frequency-domain optical coherence tomography information NI.

詳細而言，與該無雜訊資訊生成裝置31a相連接之該頻域光學同調斷層掃描裝置1，係為前述用以訓練該生成單元31成為該無雜訊資訊生成裝置31a之該頻域光學同調斷層掃描裝置1。 Specifically, the frequency-domain optical coherence tomography device 1 connected to the noise-free information generating device 31a is the aforementioned frequency-domain optical coherence tomography device for training the generating unit 31 to become the noise-free information generating device 31a. Coherence tomography apparatus 1 .

如第2b圖所示，本發明之實施例的該全域式頻域光學同調斷層掃描系統100a的該生成單元31經由前述之流程，已然成功被訓練成為該無雜訊資訊生成裝置31a。因此，使用者只需將該待測檢體S輸入至該頻域光學同調斷層掃描裝置1後，該頻域光學同調斷層掃描裝置1遂針對所輸入之該待測檢體S進行頻域光學同調斷層掃描FDOCT而產生該原始頻域光學同調斷層掃描資訊OI。該原始頻域光學同調斷層掃描資訊OI隨後被傳輸至該無雜訊資訊生成裝置31a，而經由該無雜訊資訊生成裝置31a產生該無雜訊頻域光學同調斷層掃描資訊NI。 As shown in FIG. 2b, the generating unit 31 of the global frequency-domain optical coherence tomography system 100a according to the embodiment of the present invention has been successfully trained to become the noise-free information generating device 31a through the aforementioned process. Therefore, the user only needs to input the test object S to the frequency domain optical coherence tomography apparatus 1 , and the frequency domain optical coherence tomography scanning apparatus 1 performs frequency domain optical coherence on the input test object S. The coherence tomography FDOCT produces the raw frequency domain optical coherence tomography information OI. The original frequency-domain optical coherence tomography information OI is then transmitted to the noise-free information generating device 31a, and the noise-free frequency-domain optical coherence tomography information NI is generated by the noise-free information generating device 31a.

如第2a圖所示，由於該生成單元31已被訓練成為該無雜訊資訊生成裝置31a，則原先用於幫助訊練該生成單元31成為該無雜訊資訊生成裝置31a之該訓練用無雜訊資訊產生裝置2以及該判別單元32在此實施例中係為可忽略。因此，在第2a圖中之該訓練用無雜訊資訊產生裝置2以及該判別單元32係以虛線表示。 As shown in FIG. 2a, since the generating unit 31 has been trained to become the noise-free information generating device 31a, the training-useless data originally used to help the generating unit 31 become the noise-free information generating device 31a The noise information generating device 2 and the judging unit 32 are negligible in this embodiment. because Here, the noise-free information generating device 2 for training and the discriminating unit 32 in Fig. 2a are represented by dotted lines.

如第3a圖以及第4a圖所示，依據本發明的實施例的該全域式頻域光學同調斷層掃描系統100，其中該頻域光學同調斷層掃描裝置1所產生之該訓練用原始頻域光學同調斷層掃描資訊TOI(如第3a圖所示，該圖僅為該訓練用原始頻域光學同調斷層掃描資訊TOI之示意圖，並不代表本案實際產生之該訓練用原始頻域光學同調斷層掃描資訊TOI)以及該訓練用無雜訊資訊產生裝置2所產生之該訓練用無雜訊頻域光學同調斷層掃描資訊TNI(如第4a圖所示，該圖僅為該訓練用無雜訊頻域光學同調斷層掃描資訊TNI之示意圖，並不代表本案實際產生之該訓練用無雜訊頻域光學同調斷層掃描資訊TNI)皆屬於一維圖片資訊。 As shown in Fig. 3a and Fig. 4a, the global frequency-domain optical coherence tomography system 100 according to the embodiment of the present invention, wherein the training original frequency-domain optical coherence tomography generated by the frequency-domain optical coherence tomography device 1 is used for training. Coherence tomography information TOI (as shown in Figure 3a, this figure is only a schematic diagram of the original frequency-domain optical coherence tomography information TOI for training, and does not represent the original frequency-domain optical coherence tomography information for training actually generated in this case. TOI) and the training noise-free frequency-domain optical coherence tomography information TNI generated by the training-use noise-free information generating device 2 (as shown in Figure 4a, which is only the training-use noise-free frequency domain The schematic diagram of the optical coherence tomography information (TNI) does not mean that the noise-free frequency domain optical coherence tomography information (TNI) actually generated in this case belongs to one-dimensional image information.

如第1a圖、第3b圖以及第4b圖所示，依據本發明的實施例的該全域式頻域光學同調斷層掃描系統100，更包括維度轉換裝置4，該維度轉換裝置4係與該頻域光學同調斷層掃描裝置1、該訓練用無雜訊資訊產生裝置2以及該人工智慧訓練裝置3相互資訊連接，用以接收該訓練用原始頻域光學同調斷層掃描資訊TOI以及該訓練用無雜訊頻域光學同調斷層掃描資訊TNI，並將該訓練用原始頻域光學同調斷層掃描資訊TOI以及該訓練用無雜訊頻域光學同調斷層掃描資訊TNI之維度予以轉換後而轉為二維圖片資訊(如第3b圖以及第4b圖所示)，再將維度轉為二維圖片資訊之該訓練用原始頻域光學同調斷層掃描資訊TOI(如第3b圖所示，第3b圖僅為該訓練用原始頻域光學同調斷層掃描資訊TOI之二維圖片資訊示意圖，並不代表本案實際產生之該訓練用原始頻域光學同調斷層掃描資訊TOI之二維圖片資訊)傳輸至該生成單元31，以及將維度轉為二維圖片資訊之該訓練用無雜訊頻域光學同調斷層掃描資訊TNI(如第4b圖所示，第4b圖僅為該訓練用無雜訊頻域光學同調斷層掃描資訊TNI之二維圖片資訊示意圖，並不代表本案實際產生之該訓練用無雜訊頻域光學同調斷層掃描資訊TNI之二維圖片資訊)傳輸至該判別單元32。 As shown in Fig. 1a, Fig. 3b and Fig. 4b, the global frequency domain optical coherence tomography system 100 according to the embodiment of the present invention further includes a dimension conversion device 4, and the dimension conversion device 4 is connected to the frequency domain. The domain optical coherence tomography device 1, the training-use noise-free information generating device 2, and the artificial intelligence training device 3 are mutually informationally connected to receive the training original frequency-domain optical coherence tomography information TOI and the training noise-free information TOI frequency domain optical coherence tomography information TNI, and convert the dimensions of the original frequency domain optical coherence tomography information TOI for training and the noise-free frequency domain optical coherence tomography information TNI for training into a two-dimensional image information (as shown in Figure 3b and Figure 4b), and then convert the dimensions into 2D image information for the original frequency domain optical coherence tomography information TOI for training (as shown in Figure 3b, Figure 3b is only the The schematic diagram of the two-dimensional image information of the original frequency-domain optical coherence tomography information TOI for training does not represent that the two-dimensional image information of the original frequency-domain optical coherence tomography information TOI for training actually generated in this case) is transmitted to the generating unit 31, and the training noise-free frequency-domain optical coherence tomography information TNI (as shown in Figure 4b, which is only the The schematic diagram of the two-dimensional image information of the noise-free frequency-domain optical coherence tomography information TNI for training does not represent that the two-dimensional image information of the training-use noise-free frequency-domain optical coherence tomography information TNI) actually generated in this case is transmitted to the discriminating unit 32 .

詳細而言，本發明的實施例的該人工智慧訓練裝置3係使用pix2pix生成對抗網路技術(Generative Adversarial Network,GAN)，由於此種技術係適用於處理二維圖片資訊，而該頻域光學同調斷層掃描裝置1所產生之該訓練用原始頻域光學同調斷層掃描資訊TOI以及該訓練用無雜訊資訊產生裝置2所產生之該訓練用無雜訊頻域光學同調斷層掃描資訊TNI皆屬於一維圖片資訊，因此，該維度轉換裝置4係依據該全域式頻域光學同調斷層掃描系統100運行過程中之需求而將該訓練用原始頻域光學同調斷層掃描資訊TOI以及該訓練用無雜訊頻域光學同調斷層掃描資訊TNI之維度轉為二維圖片資訊，以使該人工智慧訓練裝置3得以順利進行隨後之運作流程。 In detail, the artificial intelligence training device 3 of the embodiment of the present invention uses the pix2pix generative adversarial network technology (Generative Adversarial Network, GAN), because this technology is suitable for processing two-dimensional image information, and the frequency domain optical The training-use original frequency-domain optical coherence tomography information TOI generated by the coherence tomography device 1 and the training-use noise-free frequency-domain optical coherence tomography information TNI generated by the training-use noise-free information generation device 2 belong to One-dimensional image information, therefore, the dimensional conversion device 4 uses the original frequency-domain optical coherence tomography information TOI for training and the non-cluttered data for training according to the requirements during the operation of the global frequency-domain optical coherence tomography system 100 The dimension of the frequency domain optical coherence tomography information TNI is converted into two-dimensional image information, so that the artificial intelligence training device 3 can smoothly perform the subsequent operation process.

詳細而言，如第1a圖所示，該維度轉換裝置4係與該頻域光學同調斷層掃描裝置1、該訓練用無雜訊資訊產生裝置2以及該人工智慧訓練裝置3相互資訊連接。該頻域光學同調斷層掃描裝置1對於複數個訓練用樣本TS進行頻域光學同調斷層掃描FDOCT所產生的該訓練用原始頻域光學同調斷層掃描資訊TOI，以及該訓練用無雜訊資訊產生裝置2所產生的該訓練用無雜訊頻域光學同調斷層掃描資訊TNI皆傳輸至該維度轉換裝置4處理。接收到該訓練用原始頻域光學同調斷層掃描資訊TOI以及該訓練用無雜訊頻域光學同調斷層掃描資訊TNI的該維度轉換裝置4遂開始將該訓練用原始頻域光學同調斷層掃描資訊TOI以及該訓練用無雜訊頻域光學同調斷層掃描資訊TNI之維度予以轉換為二維圖片資訊，再將維度轉為二維圖片資訊之該訓練用原始頻域光學同調斷層掃描資訊TOI 以及該訓練用無雜訊頻域光學同調斷層掃描資訊TNI傳輸至該人工智慧訓練裝置3，以使該人工智慧訓練裝置3得以順利進行隨後之運作流程。 Specifically, as shown in FIG. 1a , the dimension conversion device 4 is connected with the frequency domain optical coherence tomography device 1 , the training noise-free information generating device 2 and the artificial intelligence training device 3 with each other. The frequency domain optical coherence tomography device 1 performs the frequency domain optical coherence tomography FDOCT on a plurality of training samples TS to generate the training original frequency domain optical coherence tomography information TOI, and the training noise-free information generating device 2. The noise-free frequency domain optical coherence tomography information TNI generated for training is transmitted to the dimensional conversion device 4 for processing. The dimension conversion device 4 having received the original frequency-domain optical coherence tomography information TOI for training and the noise-free frequency-domain optical coherence tomography information TNI for training then starts the original frequency-domain optical coherence tomography information TOI for training And the dimension of the noise-free frequency-domain optical coherence tomography information TNI for training is converted into two-dimensional image information, and then the dimension is converted into two-dimensional image information, the original frequency-domain optical coherence tomography information TOI used for training And the noise-free frequency domain optical coherence tomography information TNI for training is transmitted to the artificial intelligence training device 3 , so that the artificial intelligence training device 3 can smoothly perform the subsequent operation process.

如第2a圖所示，依據本發明的實施例的該全域式頻域光學同調斷層掃描系統100a，其中該無雜訊資訊生成裝置31a所生成之該無雜訊頻域光學同調斷層掃描資訊NI係屬於二維圖片資訊，該無雜訊頻域光學同調斷層掃描資訊NI遂藉由該維度轉換裝置4而將該無雜訊頻域光學同調斷層掃描資訊NI之維度轉為一維圖片資訊。 As shown in FIG. 2a, in the global frequency-domain optical coherence tomography scanning system 100a according to an embodiment of the present invention, the noise-free frequency-domain optical coherence tomography scanning information NI generated by the noise-free information generating device 31a Belonging to two-dimensional image information, the noise-free frequency domain optical coherence tomography information NI is converted into one-dimensional image information by the dimension conversion device 4 .

詳細而言，如第2a圖所示，該頻域光學同調斷層掃描裝置1針對所輸入之該待測檢體S進行頻域光學同調斷層掃描FDOCT而產生該原始頻域光學同調斷層掃描資訊OI後，由於該原始頻域光學同調斷層掃描資訊OI係屬於一維圖片資訊，因此，需要將該原始頻域光學同調斷層掃描資訊OI傳輸至該維度轉換裝置4進行維度轉換而轉為二維圖片資訊，該維度轉換裝置4再將轉為二維圖片資訊之該原始頻域光學同調斷層掃描資訊OI傳輸至該無雜訊資訊生成裝置31a，方能使該無雜訊資訊生成裝置31a順利產生該無雜訊頻域光學同調斷層掃描資訊NI。由於該無雜訊資訊生成裝置31a所生成之該無雜訊頻域光學同調斷層掃描資訊NI係屬於二維圖片資訊，而使用者所欲取得之無雜訊頻域光學同調斷層掃描資訊NI應為一維圖片資訊，因此，需要將該無雜訊資訊生成裝置31a剛生成之該無雜訊頻域光學同調斷層掃描資訊NI傳輸至該維度轉換裝置4進行維度轉換而轉為一維圖片資訊。 In detail, as shown in Fig. 2a, the frequency-domain optical coherence tomography apparatus 1 performs frequency-domain optical coherence tomography (FDOCT) on the input sample S to generate the original frequency-domain optical coherence tomography information OI Then, since the original frequency-domain optical coherence tomography information OI belongs to one-dimensional image information, it is necessary to transmit the original frequency-domain optical coherence tomography information OI to the dimension conversion device 4 for dimension conversion to convert it into a two-dimensional image information, the dimension conversion device 4 then transmits the original frequency domain optical coherence tomography information OI converted into two-dimensional image information to the noise-free information generating device 31a, so that the noise-free information generating device 31a can be successfully generated The noise-free frequency domain optical coherence tomography information NI. Since the noise-free frequency-domain optical coherence tomography information NI generated by the noise-free information generating device 31a is two-dimensional image information, the noise-free frequency-domain optical coherence tomography information NI desired by the user should be It is one-dimensional image information. Therefore, the noise-free frequency-domain optical coherence tomography information NI just generated by the noise-free information generating device 31a needs to be transmitted to the dimensional conversion device 4 for dimensional conversion and converted into one-dimensional image information. .

如第1a圖所示，依據本發明的實施例的該全域式頻域光學同調斷層掃描系統100，其中該訓練用無雜訊資訊產生裝置2更包括鏡像雜訊去除單元21，用於消除該訓練用原始頻域光學同調斷層掃描資訊TOI中的自相關雜訊以及鏡像雜訊，使該訓練用無雜訊資訊產生裝置2產生相對應於該訓練用原始頻域光學同調斷層掃描資訊TOI且不包含自相關雜訊以及鏡像雜訊之該訓練用無雜訊頻域光學同調斷層掃描資訊TNI。 As shown in FIG. 1a , in the global frequency-domain optical coherence tomography scanning system 100 according to an embodiment of the present invention, the noise-free information generating device 2 for training further includes an image noise removing unit 21 for removing the image noise. Autocorrelation noise in TOI of raw frequency-domain optical coherence tomography information for training and Image noise, so that the training-use noise-free information generating device 2 generates the training-use noise-free frequency corresponding to the original frequency-domain optical coherence tomography information TOI used for training and does not contain autocorrelation noise and image noise Domain Optical Coherence Tomography Information TNI.

詳細而言，該鏡像雜訊去除單元21係為本發明所屬技術領域中具有通常知識者所熟知之用於去除鏡像雜訊以及自相關雜訊之技術、裝置或是方法。 Specifically, the image noise removal unit 21 is a technology, device or method for removing image noise and autocorrelation noise well known to those skilled in the art to which the present invention pertains.

如第2a圖至第2b圖所示，依據本發明的實施例的該全域式頻域光學同調斷層掃描系統100a，其中該無雜訊頻域光學同調斷層掃描資訊NI，係為該無雜訊資訊生成裝置31a依據該原始頻域光學同調斷層掃描資訊OI進行該去除雜訊處理NP，將該原始頻域光學同調斷層掃描資訊OI之自相關雜訊以及鏡像雜訊消除後所生成。 As shown in FIGS. 2a to 2b, in the global frequency-domain optical coherence tomography scanning system 100a according to an embodiment of the present invention, the noise-free frequency-domain optical coherence tomography scanning information NI is the noise-free frequency-domain optical coherence tomography scanning information NI The information generating device 31a performs the noise removal process NP according to the original frequency-domain optical coherence tomography information OI, and generates after eliminating the autocorrelation noise and the image noise of the original frequency-domain optical coherence tomography information OI.

藉由本發明所採用之技術手段，使得使用者得以藉由人工智慧技術而將進行傳統的頻域光學同調斷層掃描FDOCT後所得出的一維圖片資訊中之自相關雜訊以及鏡像雜訊完整的消除。本發明的該全域式頻域光學同調斷層掃描系統100以及該全域式頻域光學同調斷層掃描系統100a不僅能夠節省許多光學系統的成本以及整個光學同調斷層掃描技術的建置成本，亦能夠降低光學系統的複雜度、有效減少產生無雜訊頻域光學同調斷層掃描資訊的所需時間，以及幫助使用者能夠更精準的判讀該待測檢體S的真實結構。 With the technical means adopted in the present invention, the user can use artificial intelligence technology to complete the autocorrelation noise and image noise in the one-dimensional image information obtained by traditional frequency domain optical coherence tomography (FDOCT) scanning. eliminate. The global frequency domain optical coherence tomography scanning system 100 and the global frequency domain optical coherence tomography scanning system 100a of the present invention can not only save the cost of many optical systems and the construction cost of the entire optical coherence tomography technology, but also can reduce the cost of optical coherence tomography. The complexity of the system effectively reduces the time required for generating noise-free frequency domain optical coherence tomography information, and helps the user to more accurately interpret the real structure of the test object S.

以上之敘述以及說明僅為本發明之較佳實施例之說明，對於此項技術具有通常知識者當可依據以下所界定申請專利範圍以及上述之說明而作其他之修改，惟此些修改仍應是為本發明之發明精神而在本發明之權利範圍中。 The above descriptions and descriptions are only descriptions of preferred embodiments of the present invention. Those with ordinary knowledge in the art can make other modifications according to the scope of the patent application defined below and the above descriptions, but these modifications should still be It is within the scope of the right of the present invention for the inventive spirit of the present invention.

21:鏡像雜訊去除單元 21: Image noise removal unit

3:人工智慧訓練裝置 3: artificial intelligence training device

31a:無雜訊資訊生成裝置 31a: Noise-free information generation device

32:判別單元 32: discriminant unit

4:維度轉換裝置 4: Dimension conversion device

Claims

一種全域式頻域光學同調斷層掃描系統，係用以根據原始頻域光學同調斷層掃描資訊產生無雜訊頻域光學同調斷層掃描資訊，該全域式頻域光學同調斷層掃描系統包含：頻域光學同調斷層掃描裝置，經配置而用於進行頻域光學同調斷層掃描，以產生該原始頻域光學同調斷層掃描資訊；訓練用無雜訊資訊產生裝置，連接於該頻域光學同調斷層掃描裝置，接收該頻域光學同調斷層掃描裝置對於複數個訓練用樣本進行頻域光學同調斷層掃描所產生的訓練用原始頻域光學同調斷層掃描資訊，並據以產生相對應的訓練用無雜訊頻域光學同調斷層掃描資訊；人工智慧訓練裝置，係連接於該頻域光學同調斷層掃描裝置以及該訓練用無雜訊資訊產生裝置，該人工智慧訓練裝置包括生成單元以及判別單元，該生成單元係用以接收該訓練用原始頻域光學同調斷層掃描資訊，並對該訓練用原始頻域光學同調斷層掃描資訊進行去除雜訊處理而產生過渡頻域光學同調斷層掃描資訊，進而將該過渡頻域光學同調斷層掃描資訊傳輸至該判別單元，該判別單元係用以接收該訓練用無雜訊頻域光學同調斷層掃描資訊以及該過渡頻域光學同調斷層掃描資訊，並對該訓練用無雜訊頻域光學同調斷層掃描資訊以及該過渡頻域光學同調斷層掃描資訊進行資訊差異比對處理而產生比對分辨結果，並使該生成單元對該訓練用原始頻域光學同調斷層掃描資訊根據該比對分辨結果重新進行另一去除雜訊處理，進而產生次世代的該過渡頻域光學同調斷層掃描資訊，直至所產生的該過渡頻域光學同調斷層掃描資訊與該訓練用無雜訊頻域光學同調斷層掃描資訊的該比對分辨結果為該判別單元所無法分辨，該生成單元再各別對其餘複數個該訓練用樣本的該訓練用原始頻域光學同調斷層掃描資訊進行該去除雜訊處理，直至全部所產生之其餘複數個該過渡頻域光學同調斷層掃描資訊與其餘複數個該訓練用無雜訊頻域光學同調斷層掃描資訊的該比對分辨結果為該判別單元所無法分辨，而藉此將該生成單元訓練成為無雜訊資訊生成裝置；以及該無雜訊資訊生成裝置，連接於該頻域光學同調斷層掃描裝置，係接收該頻域光學同調斷層掃描裝置對於複數個待測檢體進行頻域光學同調斷層掃描所產生的該原始頻域光學同調斷層掃描資訊，並根據該原始頻域光學同調斷層掃描資訊而對應生成該無雜訊頻域光學同調斷層掃描資訊。 A global frequency domain optical coherence tomography scanning system is used for generating noise-free frequency domain optical coherence tomography scanning information according to original frequency domain optical coherence tomography scanning information. The global frequency domain optical coherence tomography scanning system comprises: a frequency domain optical coherence tomography device configured for frequency domain optical coherence tomography to generate the raw frequency domain optical coherence tomography information; The noise-free information generating device for training is connected to the frequency domain optical coherence tomography device, and receives the original frequency domain for training generated by the frequency domain optical coherence tomography scanning device for a plurality of training samples. Optical coherence tomography information, and generate corresponding training noise-free frequency-domain optical coherence tomography information accordingly; An artificial intelligence training device is connected to the frequency domain optical coherence tomography device and the training-use noise-free information generating device, the artificial intelligence training device includes a generating unit and a judging unit, and the generating unit is used for receiving the training raw material frequency domain optical coherence tomography information, and the original frequency domain optical coherence tomography information for training is subjected to noise removal processing to generate transition frequency domain optical coherence tomography information, and then the transition frequency domain optical coherence tomography information is transmitted to the judging unit, which is used for receiving the noise-free frequency-domain optical coherence tomography information for training and the transition frequency-domain optical coherence tomography information, and for training the noise-free frequency-domain optical coherence tomography information and the transition frequency-domain optical coherence tomography information is processed for information difference comparison to generate a comparison and resolution result, and the generating unit is made to perform another comparison and resolution on the original frequency-domain optical coherence tomography information for training according to the comparison and resolution result. Noise-removing processing to generate next-generation transition frequency domain optical coherence tomography information until the ratio of the generated transition frequency domain optical coherence tomography information to the noise-free frequency domain optical coherence tomography information for training If the discrimination result cannot be discriminated by the discriminating unit, the generating unit further performs the denoising process on the original frequency-domain optical coherence tomography scanning information for the training of the remaining plurality of training samples, until all the remaining samples are generated. The comparison and discrimination result of the plurality of transition frequency-domain optical coherence tomography information and the rest of the plurality of noise-free frequency-domain optical coherence tomography information for training cannot be distinguished by the discriminating unit, thereby training the generating unit be a noise-free information generating device; and The noise-free information generating device is connected to the frequency domain optical coherence tomography scanning device, and receives the original frequency domain generated by the frequency domain optical coherence tomography scanning device for a plurality of samples to be tested. Optical coherence tomography information, and correspondingly generate the noise-free frequency domain optical coherence tomography information according to the original frequency domain optical coherence tomography information.

如請求項1所述之該全域式頻域光學同調斷層掃描系統，其中該頻域光學同調斷層掃描裝置所產生之該訓練用原始頻域光學同調斷層掃描資訊以及該訓練用無雜訊資訊產生裝置所產生之該訓練用無雜訊頻域光學同調斷層掃描資訊皆屬於一維圖片資訊。The global frequency domain optical coherence tomography system as claimed in claim 1, wherein the training original frequency domain optical coherence tomography information and the training noise-free information generated by the frequency domain optical coherence tomography device are generated The noise-free frequency domain optical coherence tomography information for training generated by the device is all one-dimensional image information.

如請求項1所述之該全域式頻域光學同調斷層掃描系統，其中該全域式頻域光學同調斷層掃描系統更包括維度轉換裝置，該維度轉換裝置係與該頻域光學同調斷層掃描裝置、該訓練用無雜訊資訊產生裝置以及該人工智慧訓練裝置相互資訊連接，用以接收該訓練用原始頻域光學同調斷層掃描資訊以及該訓練用無雜訊頻域光學同調斷層掃描資訊，並將該訓練用原始頻域光學同調斷層掃描資訊以及該訓練用無雜訊頻域光學同調斷層掃描資訊之維度予以轉換後而轉為二維圖片資訊，再將維度轉為二維圖片資訊之該訓練用原始頻域光學同調斷層掃描資訊傳輸至該生成單元，以及將維度轉為二維圖片資訊之該訓練用無雜訊頻域光學同調斷層掃描資訊傳輸至該判別單元。The global frequency-domain optical coherence tomography system as claimed in claim 1, wherein the global frequency-domain optical coherence tomography system further comprises a dimension conversion device, and the dimension conversion device is combined with the frequency-domain optical coherence tomography device, The noise-free information generating device for training and the artificial intelligence training device are connected to each other for information to receive the original frequency-domain optical coherence tomography information for training and the noise-free frequency-domain optical coherence tomography information for training, and send The training uses the original frequency domain optical coherence tomography information and the training uses the dimension of the noise-free frequency domain optical coherence tomography information to be converted into two-dimensional image information, and then the dimension is converted to two-dimensional image information. The original frequency-domain optical coherence tomography information is transmitted to the generating unit, and the training noise-free frequency-domain optical coherence tomography information, which converts dimensions into two-dimensional image information, is transmitted to the discriminating unit.

如請求項1或3所述之該全域式頻域光學同調斷層掃描系統，其中該無雜訊資訊生成裝置所生成之該無雜訊頻域光學同調斷層掃描資訊係屬於二維圖片資訊，該無雜訊頻域光學同調斷層掃描資訊遂藉由該維度轉換裝置而將該無雜訊頻域光學同調斷層掃描資訊之維度轉為一維圖片資訊。The global frequency-domain optical coherence tomography system according to claim 1 or 3, wherein the noise-free frequency-domain optical coherence tomography information generated by the noise-free information generating device belongs to two-dimensional image information, and the The noise-free frequency-domain optical coherence tomography information is then converted from the dimension of the noise-free frequency-domain optical coherence tomography information into one-dimensional picture information by the dimension conversion device.

如請求項1所述之該全域式頻域光學同調斷層掃描系統，其中該比對分辨結果係為該訓練用無雜訊頻域光學同調斷層掃描資訊以及該過渡頻域光學同調斷層掃描資訊之資訊差異度。The global frequency-domain optical coherence tomography system as claimed in claim 1, wherein the comparison and resolution result is the difference between the noise-free frequency-domain optical coherence tomography information for training and the transition frequency-domain optical coherence tomography information Information disparity.

如請求項1所述之該全域式頻域光學同調斷層掃描系統，其中該訓練用無雜訊資訊產生裝置更包括鏡像雜訊去除單元，用於消除該訓練用原始頻域光學同調斷層掃描資訊中的自相關雜訊以及鏡像雜訊，使該訓練用無雜訊資訊產生裝置產生相對應於該訓練用原始頻域光學同調斷層掃描資訊且不包含自相關雜訊以及鏡像雜訊之該訓練用無雜訊頻域光學同調斷層掃描資訊。The global frequency-domain optical coherence tomography scanning system according to claim 1, wherein the noise-free information generating device for training further comprises an image noise removing unit for eliminating the original frequency-domain optical coherence tomography scanning information for training autocorrelation noise and image noise in the training noise-free information generating device to generate the training corresponding to the original frequency domain optical coherence tomography information for training and do not contain autocorrelation noise and image noise Scanning information with noise-free frequency-domain optical coherence tomography.