數(shù)字閃爍探測器
發(fā)布時間：2018-07-20 08:55
【摘要】：PET通過檢測攝入活體內(nèi)放射性示蹤劑的分布狀態(tài)對機體內(nèi)各器官的代謝水平、生化反應(yīng)、功能活動和灌注進行定量、動態(tài)地評估。按信號流順序,PET被分為探測器系統(tǒng)、數(shù)據(jù)獲得系統(tǒng)、圖像重建系統(tǒng)以及圖像顯示與分析系統(tǒng)。探測器系統(tǒng)通常由多個閃爍探測器排布為環(huán)形結(jié)構(gòu)的方式構(gòu)成,用于捕獲放射性示蹤劑衰變后產(chǎn)生的Gamma粒子并形成相應(yīng)的閃爍脈沖。數(shù)據(jù)獲得系統(tǒng)通過對閃爍脈沖的分析、處理獲取粒子的能量沉積信息并形成符合數(shù)據(jù)。圖像重建系統(tǒng)則利用符合數(shù)據(jù)反演出放射性示蹤劑在機體內(nèi)的分布狀態(tài)并通過后續(xù)的圖像顯示與分析系統(tǒng)展現(xiàn)出來�，F(xiàn)代PET系統(tǒng)中,數(shù)據(jù)獲得系統(tǒng)通常采用模數(shù)混合的架構(gòu)。模擬電路被用來對閃爍脈沖進行處理,粒子的能量沉積大小、時間以及位置等信息轉(zhuǎn)化為相應(yīng)的模擬電壓量或觸發(fā)信號量后再由后續(xù)的電路進行數(shù)字化獲取以及符合甄別。眾所周知,模擬電路對閃爍脈沖的幅值大小、脈沖持續(xù)時間、頻率范圍等特性較為敏感,數(shù)據(jù)獲得系統(tǒng)需要依據(jù)閃爍探測器的特性進行針對性設(shè)計;模擬電路難以實現(xiàn)復(fù)雜的脈沖分析、處理方法,粒子能量信息多由一些簡單的線性方法及電路提取,性能優(yōu)異但算法復(fù)雜的脈沖處理方法無法應(yīng)用；模擬電路系統(tǒng)難以應(yīng)對閃爍探測器輸出脈沖的基線漂移、事件堆疊等情況,限制了系統(tǒng)的穩(wěn)定性和性能的進一步提升。近年來,隨著數(shù)字信號處理技術(shù)和方法的發(fā)展,將閃爍脈沖直接數(shù)字化,利用軟件算法替代傳統(tǒng)模擬電路提取粒子能量沉積信息的方式極具吸引力。數(shù)字化后的閃爍脈沖在傳輸、處理等過程中不再因模擬電路的信號帶寬、噪聲干擾等因素影響信號質(zhì)量、惡化粒子能量沉積信息提取精度,這有利于大型、復(fù)雜系統(tǒng)的設(shè)計和開發(fā)：粒子能量沉積信息提取的準確性也將因高性能的數(shù)字信號處理技術(shù)和方法的應(yīng)用得以提高：此外,數(shù)字閃爍脈沖下基線漂移和事件堆疊等情況可在信號處理和分析的過程中得到補償和校正,系統(tǒng)的性能和穩(wěn)定性將因此而進一步提高。 將數(shù)字化電路與閃爍探測器相結(jié)合,直接輸出數(shù)字閃爍脈沖的數(shù)字閃爍探測器極具發(fā)展前景。數(shù)字閃爍探測器的實現(xiàn)將簡化PET系統(tǒng)的結(jié)構(gòu),降低系統(tǒng)的工程開發(fā)難度；快速發(fā)展的數(shù)字信號處理技術(shù)與方法也將得以應(yīng)用于閃爍脈沖的分析與處理,加速PET系統(tǒng)的發(fā)展。閃爍脈沖的準確數(shù)字化是數(shù)字閃爍探測器實現(xiàn)的關(guān)鍵。受采樣率和功耗的限制,一直以來利用ADC直接獲取數(shù)字閃爍脈沖的方式存在工程實現(xiàn)困難、硬件成本昂貴、數(shù)字脈沖欠采樣等諸多問題。在對粒子能量沉積時間信息提取精度有較高要求的TOF PET等系統(tǒng)中,ADC難以直接應(yīng)用。由我課題組提出的多閾值電壓采樣(Multi-Voltage Threshold,以下簡稱MVT)方法則通過對閃爍脈沖過閾值電壓的時間點進行采樣的方式完成閃爍脈沖的數(shù)字化。相應(yīng)的采樣電路僅需少量比較器和時間數(shù)字轉(zhuǎn)換器便可實現(xiàn),具有工程開發(fā)難度小、硬件成本低等特點。在此背景下,本文以MVT方法為基礎(chǔ)圍繞數(shù)字閃爍探測器的關(guān)鍵技術(shù)、系統(tǒng)架構(gòu)以及應(yīng)用展開研究工作。 首先,通過研究指出閃爍脈沖下降沿中不可避免的噪聲影響現(xiàn)有MVT采樣方法獲取的數(shù)字閃爍脈沖精度和后續(xù)的粒子能量沉積大小信息的提取精度。這一噪聲雖然可以通過低通濾波器濾除,但粒子能量沉積時間信息的提取精度卻因此而惡化。針對這一問題,論文提出了精確MVT采樣方法并與現(xiàn)有MVT采樣方法下得到的數(shù)字閃爍脈沖提取的粒子能量沉積精度進行了對比研究。精確MVT采樣方法實現(xiàn)了噪聲干擾下數(shù)字閃爍脈沖的精確數(shù)字化,在不惡化粒子能量沉積時間信息提取精度的前提下有效提高了能量沉積大信息的提取精度。對應(yīng)的能量分辨率由原先的16.9%@511keV優(yōu)化到13.0%@511keV。 其次,設(shè)計并實現(xiàn)了精確MVT采樣電路。解決了傳統(tǒng)TDC因死時間而無法應(yīng)用于精確MVT采樣電路的問題。提出了采用LVDS接受器實現(xiàn)閾值比較器的方法,不僅提高了MVT采樣電路的集成度、降低了系統(tǒng)功耗,還提高了脈沖過閾值電壓的時間信息獲取精度。提出了精確MVT采樣電路校正方法,解決了MVT采樣電路中不同通道間的閾值電壓和時間響應(yīng)不一致帶來的數(shù)字化精度下降的問題。將實現(xiàn)的精確MVT采樣電路應(yīng)用于一對LYSO/SiPM探測器輸出脈沖的數(shù)字化,最終獲得了13.9%@511keV的能量分辨率和438ps的時間分辨率。 再次,提出了數(shù)字閃爍探測器架構(gòu),按信號流的順序劃分為探測器單元、數(shù)字化單元以及接口單元。數(shù)字閃爍脈沖通過接口單元傳輸出來,粒子能量沉積信息則由后續(xù)數(shù)字脈沖分析、處理平臺通過分析數(shù)字閃爍脈沖的方式提取。利用該架構(gòu)下的數(shù)字閃爍探測器進行PET等系統(tǒng)的開發(fā),僅需在計算機等系統(tǒng)實現(xiàn)的數(shù)字脈沖分析、處理平臺中開發(fā)相應(yīng)的算法和程序即可完成。數(shù)字閃爍探測器的實現(xiàn),降低了PET等系統(tǒng)的開發(fā)難度。論文以該架構(gòu)和精確MVT采樣電路為基礎(chǔ)完成了數(shù)字閃爍探測器的實現(xiàn),時間分辨率為525ps、能量分辨率為15.1%@511keV。 最后,提出了數(shù)字PET系統(tǒng)架構(gòu),該架構(gòu)下的數(shù)字PET成像系統(tǒng)具有系統(tǒng)架構(gòu)簡潔,工程實現(xiàn)簡單,系統(tǒng)成像性能優(yōu)異等特點。為針對應(yīng)用,適應(yīng)性的構(gòu)建PET成像系統(tǒng)提供了技術(shù)基礎(chǔ)。采用該系統(tǒng)架構(gòu)和數(shù)字閃爍探測器,本文針對具體應(yīng)用需求完成了三種具有不同成像視野的數(shù)字PET系統(tǒng)的設(shè)計與實現(xiàn)。對其中針對臨床腦部成像的PET系統(tǒng)進行了初步的性能評估和假體成像。該系統(tǒng)的空間分辨率為2.5mmm,時間、能量分辨率分別為543ps和16.3%@511keV。與單個數(shù)字PET探測器的性能相比,系統(tǒng)性能未發(fā)生顯著惡化。這一些結(jié)果初步揭示了數(shù)字PET探測器及數(shù)字化架構(gòu)下PET成像系統(tǒng)的優(yōu)勢。
[Abstract]:PET quantified the metabolic levels, biochemical reactions, functional activities and perfusion of various organs in the body by detecting the distribution of radioactive tracers in the living body. According to the sequence of signal flow, PET was divided into detector system, data acquisition system, image reconstruction system and image display and analysis system. It is often made up of multiple scintillation detectors as a ring structure. It is used to capture Gamma particles produced by radioactive tracer decay and form corresponding scintillation pulses. The data acquisition system can obtain the energy deposition information of the particles by analysis of the scintillation pulse and form the consistent data. The distribution of radioactive tracers in the body is shown by the data. In the modern PET system, the data acquisition system usually uses a modular mixed structure. Analog circuits are used to process scintillation pulses, the size, time and position of the particle size, time and position. It is well known that the analog circuit is more sensitive to the magnitude of the scintillation pulse, the duration of the pulse, the frequency range and so on. The data acquisition system needs to be designed according to the characteristics of the scintillation detector. It is difficult to realize complex pulse analysis, processing method, the processing method, the particle energy information is extracted by some simple linear methods and circuits. The performance is excellent, but the algorithm complex pulse processing method can not be applied. The analog circuit system is difficult to cope with the base line drift of the output pulse of the scintillation detector, the event stacking and so on, which limits the system. In recent years, with the development of digital signal processing technology and methods, it is very attractive to digitize scintillation pulses directly and use software algorithms instead of traditional analog circuits to extract particle energy deposition information. The digital scintillation pulse is no longer due to analog electricity in the process of transmission and processing. The signal bandwidth, noise interference and other factors affect the quality of the signal and deteriorate the precision of particle energy deposition information extraction. This is beneficial to the design and development of large and complex systems. The accuracy of particle energy deposition information extraction will also be improved by the application of high performance digital signal processing and square methods: in addition, digital scintillation pulses will be improved. The conditions of lower baseline drift and event stacking can be compensated and corrected in the process of signal processing and analysis, and the performance and stability of the system will be further improved.
Digital scintillation detectors, which combine digital circuits with scintillation detectors, are very promising for digital scintillation detectors with direct output of digital scintillation pulses. The realization of digital scintillation detectors will simplify the structure of the PET system and reduce the difficulty of the development of the system. The rapid development of digital signal processing techniques and methods will also be applied to the scintillation pulse. Analysis and processing to accelerate the development of the PET system. The key to the realization of the digital scintillation detector is the accurate digitization of the scintillation pulse. Under the restriction of the sampling rate and the power consumption, there are many problems such as the difficulty of engineering realization, the expensive hardware cost, the undersampling of the digital pulse and so on. In the case of the particle energy, the method of obtaining the digital scintillation pulse by the sampling rate and the power consumption is limited. In the system of TOF PET, such as high precision of extracting time information, ADC is difficult to apply directly. The method of multi threshold voltage sampling (Multi-Voltage Threshold, hereinafter referred to as hereinafter referred to as MVT) proposed by our group is to digitize the scintillation pulse by sampling the time point of the threshold voltage of the scintillation pulse. The sampling circuit only needs a small amount of comparator and time digital converter. It has the characteristics of little difficulty in engineering development and low cost of hardware. Under this background, this paper is based on the key technology, system architecture and application of digital scintillation detector based on MVT method.
First, the precision of the digital scintillation pulse obtained by the existing MVT sampling method and the extraction precision of the subsequent particle energy deposition information obtained by the unavoidable noise in the falling edge of the scintillation pulse are studied. Although the noise can be filtered through a low pass filter, the extraction precision of the particle energy deposition time information is therefore the result of this. In order to solve this problem, an accurate MVT sampling method is proposed and compared with the particle energy deposition precision obtained from the digital scintillation pulse obtained under the existing MVT sampling method. The precise MVT sampling method realizes the exact digital character of the digital scintillation pulse under the noise interference, and does not deteriorate the energy deposition time letter of the particle. Under the premise of accuracy of extraction, the extraction accuracy of large energy information is effectively improved. The corresponding energy resolution is optimized from original 16.9%@511keV to 13.0%@511keV..
Secondly, an accurate MVT sampling circuit is designed and implemented. The problem that the traditional TDC can not be applied to the precise MVT sampling circuit because of the dead time is solved. A method of using the LVDS receiver to realize the threshold comparator is proposed, which not only improves the integration degree of the MVT sampling circuit, reduces the power consumption of the system, but also improves the time information of the pulse over threshold voltage. The accurate MVT sampling circuit correction method is proposed to solve the problem of the digital precision decline caused by different threshold voltage and time response between different channels in the MVT sampling circuit. The accurate MVT sampling circuit is applied to the digitization of the output pulse of a pair of LYSO/SiPM detectors, and the 13.9%@511keV is finally obtained. The resolution of energy and the time resolution of 438ps.
Thirdly, the architecture of digital scintillation detector is proposed, which is divided into detector unit, digital unit and interface unit according to the sequence of signal flow. The digital scintillation pulse is transmitted through the interface unit. The particle energy deposition information is analyzed by the subsequent digital pulse, and the processing platform is extracted through the analysis of the digital scintillation pulse. The development of the digital scintillation detector for PET, such as the digital pulse analysis and the development of the corresponding algorithms and programs in the processing platform. The realization of the digital scintillation detector reduces the difficulty of the development of PET and other systems. This paper is based on the architecture and the accurate MVT sampling circuit. The implementation of digital scintillation detector has a time resolution of 525ps and an energy resolution of 15.1%@511keV..
Finally, the architecture of digital PET system is proposed. The digital PET imaging system under this architecture has the characteristics of simple system architecture, simple engineering implementation and excellent imaging performance. It provides a technical basis for the application of the adaptive construction of PET imaging system. The system architecture and digital scintillation detectors are adopted. This paper is aimed at the specific application requirements. The design and implementation of three digital PET systems with different imaging fields are completed. A preliminary performance evaluation and prosthesis imaging of the PET system for clinical brain imaging are performed. The spatial resolution of the system is 2.5mmm, time, and energy resolution is compared to the performance of 543ps and 16.3%@ 511keV., respectively, with a single digital PET detector. The performance of the system has not significantly deteriorated. These results preliminarily reveal the advantages of the digital PET detector and the PET imaging system under the digital architecture.
【學(xué)位授予單位】：華中科技大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2015
【分類號】：R817
本文編號：2132974
[Abstract]:PET quantified the metabolic levels, biochemical reactions, functional activities and perfusion of various organs in the body by detecting the distribution of radioactive tracers in the living body. According to the sequence of signal flow, PET was divided into detector system, data acquisition system, image reconstruction system and image display and analysis system. It is often made up of multiple scintillation detectors as a ring structure. It is used to capture Gamma particles produced by radioactive tracer decay and form corresponding scintillation pulses. The data acquisition system can obtain the energy deposition information of the particles by analysis of the scintillation pulse and form the consistent data. The distribution of radioactive tracers in the body is shown by the data. In the modern PET system, the data acquisition system usually uses a modular mixed structure. Analog circuits are used to process scintillation pulses, the size, time and position of the particle size, time and position. It is well known that the analog circuit is more sensitive to the magnitude of the scintillation pulse, the duration of the pulse, the frequency range and so on. The data acquisition system needs to be designed according to the characteristics of the scintillation detector. It is difficult to realize complex pulse analysis, processing method, the processing method, the particle energy information is extracted by some simple linear methods and circuits. The performance is excellent, but the algorithm complex pulse processing method can not be applied. The analog circuit system is difficult to cope with the base line drift of the output pulse of the scintillation detector, the event stacking and so on, which limits the system. In recent years, with the development of digital signal processing technology and methods, it is very attractive to digitize scintillation pulses directly and use software algorithms instead of traditional analog circuits to extract particle energy deposition information. The digital scintillation pulse is no longer due to analog electricity in the process of transmission and processing. The signal bandwidth, noise interference and other factors affect the quality of the signal and deteriorate the precision of particle energy deposition information extraction. This is beneficial to the design and development of large and complex systems. The accuracy of particle energy deposition information extraction will also be improved by the application of high performance digital signal processing and square methods: in addition, digital scintillation pulses will be improved. The conditions of lower baseline drift and event stacking can be compensated and corrected in the process of signal processing and analysis, and the performance and stability of the system will be further improved.
Digital scintillation detectors, which combine digital circuits with scintillation detectors, are very promising for digital scintillation detectors with direct output of digital scintillation pulses. The realization of digital scintillation detectors will simplify the structure of the PET system and reduce the difficulty of the development of the system. The rapid development of digital signal processing techniques and methods will also be applied to the scintillation pulse. Analysis and processing to accelerate the development of the PET system. The key to the realization of the digital scintillation detector is the accurate digitization of the scintillation pulse. Under the restriction of the sampling rate and the power consumption, there are many problems such as the difficulty of engineering realization, the expensive hardware cost, the undersampling of the digital pulse and so on. In the case of the particle energy, the method of obtaining the digital scintillation pulse by the sampling rate and the power consumption is limited. In the system of TOF PET, such as high precision of extracting time information, ADC is difficult to apply directly. The method of multi threshold voltage sampling (Multi-Voltage Threshold, hereinafter referred to as hereinafter referred to as MVT) proposed by our group is to digitize the scintillation pulse by sampling the time point of the threshold voltage of the scintillation pulse. The sampling circuit only needs a small amount of comparator and time digital converter. It has the characteristics of little difficulty in engineering development and low cost of hardware. Under this background, this paper is based on the key technology, system architecture and application of digital scintillation detector based on MVT method.
First, the precision of the digital scintillation pulse obtained by the existing MVT sampling method and the extraction precision of the subsequent particle energy deposition information obtained by the unavoidable noise in the falling edge of the scintillation pulse are studied. Although the noise can be filtered through a low pass filter, the extraction precision of the particle energy deposition time information is therefore the result of this. In order to solve this problem, an accurate MVT sampling method is proposed and compared with the particle energy deposition precision obtained from the digital scintillation pulse obtained under the existing MVT sampling method. The precise MVT sampling method realizes the exact digital character of the digital scintillation pulse under the noise interference, and does not deteriorate the energy deposition time letter of the particle. Under the premise of accuracy of extraction, the extraction accuracy of large energy information is effectively improved. The corresponding energy resolution is optimized from original 16.9%@511keV to 13.0%@511keV..
Secondly, an accurate MVT sampling circuit is designed and implemented. The problem that the traditional TDC can not be applied to the precise MVT sampling circuit because of the dead time is solved. A method of using the LVDS receiver to realize the threshold comparator is proposed, which not only improves the integration degree of the MVT sampling circuit, reduces the power consumption of the system, but also improves the time information of the pulse over threshold voltage. The accurate MVT sampling circuit correction method is proposed to solve the problem of the digital precision decline caused by different threshold voltage and time response between different channels in the MVT sampling circuit. The accurate MVT sampling circuit is applied to the digitization of the output pulse of a pair of LYSO/SiPM detectors, and the 13.9%@511keV is finally obtained. The resolution of energy and the time resolution of 438ps.
Thirdly, the architecture of digital scintillation detector is proposed, which is divided into detector unit, digital unit and interface unit according to the sequence of signal flow. The digital scintillation pulse is transmitted through the interface unit. The particle energy deposition information is analyzed by the subsequent digital pulse, and the processing platform is extracted through the analysis of the digital scintillation pulse. The development of the digital scintillation detector for PET, such as the digital pulse analysis and the development of the corresponding algorithms and programs in the processing platform. The realization of the digital scintillation detector reduces the difficulty of the development of PET and other systems. This paper is based on the architecture and the accurate MVT sampling circuit. The implementation of digital scintillation detector has a time resolution of 525ps and an energy resolution of 15.1%@511keV..
Finally, the architecture of digital PET system is proposed. The digital PET imaging system under this architecture has the characteristics of simple system architecture, simple engineering implementation and excellent imaging performance. It provides a technical basis for the application of the adaptive construction of PET imaging system. The system architecture and digital scintillation detectors are adopted. This paper is aimed at the specific application requirements. The design and implementation of three digital PET systems with different imaging fields are completed. A preliminary performance evaluation and prosthesis imaging of the PET system for clinical brain imaging are performed. The spatial resolution of the system is 2.5mmm, time, and energy resolution is compared to the performance of 543ps and 16.3%@ 511keV., respectively, with a single digital PET detector. The performance of the system has not significantly deteriorated. These results preliminarily reveal the advantages of the digital PET detector and the PET imaging system under the digital architecture.
【學(xué)位授予單位】：華中科技大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2015
【分類號】：R817
【相似文獻】
相關(guān)期刊論文 前1條
1 王震濤;張建國;楊翊方;王海軍;;無機閃爍探測器的發(fā)展和應(yīng)用[J];中國輻射衛(wèi)生;2008年04期
相關(guān)會議論文 前10條
1 范銳鋒;馬景芳;艾杰;戴紅躍;;雙通道閃爍探測器探測脈沖中子技術(shù)[A];中國工程物理研究院科技年報(2008年版)[C];2009年
2 陳家斌;唐道源;丁永坤;王紅斌;馮杰;騰浩;張海鷹;;超快猝滅塑料閃爍探測器性能標定[A];中國工程物理研究院科技年報（1998）[C];1998年
3 何景棠;陳端保;朱國義;毛裕芳;董小黎;李祖豪;;新型雪崩硅光二極管作為閃爍探測器讀出元件的應(yīng)用研究[A];第8屆全國核電子學(xué)與核探測技術(shù)學(xué)術(shù)年會論文集（一）[C];1996年
4 張承模;馬宇倩;王煥玉;吳枚;沈培若;李毅;徐玉朋;趙敏;白新華;;宇宙γ暴閃爍探測器研究[A];第8屆全國核電子學(xué)與核探測技術(shù)學(xué)術(shù)年會論文集（一）[C];1996年
5 范銳鋒;馬景芳;艾杰;文延偉;戴紅躍;;雙通道閃爍探測器探測中子技術(shù)[A];第十二屆全國核電子學(xué)與核探測技術(shù)學(xué)術(shù)年會論文集[C];2004年
6 楊其京;;巧用γ閃爍探測器的“坪”特性[A];中國計量協(xié)會冶金分會2008年會論文集[C];2008年
7 楊其京;;巧用γ閃爍探測器的“坪”特性[A];2008全國第十三屆自動化應(yīng)用技術(shù)學(xué)術(shù)交流會論文集[C];2008年
8 孟丹;鄧長明;程昶;任熠;宋稱心;劉蕓;王志剛;;兩種大面積塑料閃爍探測器的性能對比及工藝設(shè)計[A];第十四屆全國核電子學(xué)與核探測技術(shù)學(xué)術(shù)年會論文集（上冊）[C];2008年
9 孟丹;鄧長明;程昶;任熠;宋稱心;劉蕓;王志剛;;兩種大面積塑料閃爍探測器的性能對比及工藝設(shè)計[A];第十四屆全國核電子學(xué)與核探測技術(shù)學(xué)術(shù)年會論文集（1）[C];2008年
10 顏素娟;姚歷農(nóng);;一個新型的閃爍探測器活度測量系統(tǒng)[A];第8屆全國核電子學(xué)與核探測技術(shù)學(xué)術(shù)年會論文集（二）[C];1996年
相關(guān)博士學(xué)位論文 前1條
1 奚道明;數(shù)字閃爍探測器[D];華中科技大學(xué);2015年
相關(guān)碩士學(xué)位論文 前7條
1 王立宗;載~6Li塑料閃爍探測器性能研究[D];中國工程物理研究院;2004年
2 張偉杰;快定時閃爍探測器的研究及應(yīng)用[D];蘭州大學(xué);2013年
3 羅飛;低水平放射性氙探測技術(shù)研究[D];中國工程物理研究院;2014年
4 胡強;雙維位置靈敏CsI（Tl）閃爍探測的性能研究[D];西南大學(xué);2008年
5 賈牧霖;基于硅PIN光電二極管袖珍式γ輻射計的設(shè)計與實現(xiàn)[D];成都理工大學(xué);2009年
6 肖明;輕便型自動巖芯伽瑪掃描儀的研制[D];成都理工大學(xué);2011年
7 李星;基于能譜修正環(huán)境和防護水平γ劑量率儀的研制[D];成都理工大學(xué);2012年
，本文編號：2132974
本文鏈接：http://sikaile.net/yixuelunwen/swyx/2132974.html
最近更新
教材專著
