WO2019124731A1 - Appareil et procédé de traitement de signaux avec multiplexage utilisant un regroupement et un apprentissage profond - Google Patents

Appareil et procédé de traitement de signaux avec multiplexage utilisant un regroupement et un apprentissage profond Download PDF

Info

Publication number
WO2019124731A1
WO2019124731A1 PCT/KR2018/013495 KR2018013495W WO2019124731A1 WO 2019124731 A1 WO2019124731 A1 WO 2019124731A1 KR 2018013495 W KR2018013495 W KR 2018013495W WO 2019124731 A1 WO2019124731 A1 WO 2019124731A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
radiation
clustering
optical sensor
electrical
Prior art date
Application number
PCT/KR2018/013495
Other languages
English (en)
Korean (ko)
Inventor
최용
정지웅
김대운
Original Assignee
서강대학교산학협력단
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 서강대학교산학협력단 filed Critical 서강대학교산학협력단
Publication of WO2019124731A1 publication Critical patent/WO2019124731A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • the present invention relates to a multiplexing signal processing apparatus and method that can be used in a radiation imaging apparatus, and more particularly, to a multiplexing signal processing apparatus for determining a position of a radiation reaction using a result of learning radiation reaction position data, ≪ / RTI >
  • radiological imaging devices such as Positron Emission Tomography (PET), Gamma camera, and Single Photon Emission Computed Tomography (SPECT) And it is widely used in the medical field because it can record the movement route or the distribution of the substance.
  • PET Positron Emission Tomography
  • SPECT Single Photon Emission Computed Tomography
  • Cylindrical positron emission tomography is a device that can detect two extinction beams emitted at the same time. When the image is reconstructed using the detected radiation, a 3D tomographic image of how much radioactive medicines are gathered in the body can be shown.
  • a positron tomography (PET) system comprises a plurality of scintillators, photodetectors, signal processing units, and image processing units arranged in a ring or polygon to detect gamma (?) - rays.
  • the photodetector has a structure in which a quadrangle-shaped scintillator is stacked in a single stage or a multi-stage structure. In the lower part of the scintillator, an optical sensor for processing a detected signal and other signal processors are included.
  • Korean Patent Laid-Open No. 10-2014-0064524 (entitled “Method and Apparatus for Estimating the Distribution of Positions Where Radiation Exits)” includes a step of acquiring a time difference of the radiation detected in the pair of detectors, Discloses a technique for generating a probability distribution function that shows, as a probability, a position where radiation is emitted in consideration of a time difference and a time resolution of a pair of detectors.
  • a radiation imaging apparatus such as a positron tomography (PET) uses a multiplexing circuit to reduce the number of output signals of a photodetector.
  • PET positron tomography
  • Deep learing algorithms have been applied to artificial intelligence algorithms applied to radiation imaging technology.
  • artificial intelligence algorithms such as deep learning are supervised learning algorithms, so users have to acquire and learn learning data for the purpose. Accordingly, in order to acquire learning data for learning the neural network, it is necessary to make a point type radiation with a sneeze, and a mechanical control device for controlling the position of the point of sight of the point radiation is separately required for each of a plurality of positions. There is a problem that the time required to acquire all the learning data for each position is long and the physical setting is required each time the environment of the detector is changed.
  • One embodiment of the present invention provides a multiplexing signal processing device capable of simplifying a deep learning learning data acquisition procedure using a depth learning (Deep Learning) to determine a location of a radiation reaction, and using an unsupervised clustering algorithm, And a signal processing method therefor.
  • a multiplexing signal processing device capable of simplifying a deep learning learning data acquisition procedure using a depth learning (Deep Learning) to determine a location of a radiation reaction, and using an unsupervised clustering algorithm, And a signal processing method therefor.
  • a multiplexing signal processing apparatus comprising a plurality of optical sensors for converting a scintillation signal output from a scintillator for converting radiation into a scintillation signal into an electrical signal
  • a signal detector including an optical sensor array
  • a signal processor for classifying a signal output from the signal detector through an artificial neural network to determine a position of a radiation reaction.
  • the signal processor obtains an electrical signal detected through the optical sensor array while uniformly radiating the radiation to the scintillator through a point source, and outputs the electrical signal obtained at the arbitrary position
  • a plurality of times of collecting the electrical signals at a certain position and clustering the electrical signals at a predetermined position after the radiation is uniformly radiated Maps the positions of the sets and the pixels when the number of pixels matches the number of pixels, and uses the clustered sets of electrical signals as learning data of the classifier.
  • a multiplexing signal processing method for irradiating a scintillation signal output from a scintillator to a scintillator by uniformly radiating the scintillation radiation through a point source, Converting into an electrical signal through an array; Obtaining an electrical signal output through the optical sensor array; Clustering the obtained electrical signal by position; A plurality of times of collecting the electrical signals after uniformly emitting the radiation, and repeating a series of processes of collecting the electrical signals a plurality of times, and when the number of sets of electrical signals by the clustering is equal to the number of pixels corresponding to the optical sensors, Mapping positions between the pixels and the pixels; And generating a learned classifier through Deep Learning using the set of clustering electrical signals as learning data.
  • the pixel region is divided using the clustering algorithm, and the position of the reaction (gamma ray) reaction is discriminated on the basis of the artificial neural network, As a result, the division of the pixel region is clarified and the distortion is improved, so that a radiation detection image having a high resolution can be obtained.
  • an apparatus and method as set forth in the appended claims there is provided an artificial neural network
  • the physical / temporal cost is greatly reduced in acquiring the learning data.
  • the present invention can acquire new learning data in a short period of time and construct an artificial neural network, so that it can be quickly applied to the radiation reaction location determination processing. Based on the artificial neural network using the learning data By classifying the position of each pixel through the deep running classifier and determining the position of the obtained radiation, it is possible to improve the resolution and sensitivity of the image by improving the nonlinear response position discrimination performance of the conventional multiplexing circuit.
  • FIG. 1 is a block diagram of a multiplexing signal processing apparatus using clustering and depth learning according to an embodiment of the present invention.
  • FIG. 2 is a diagram for explaining a configuration of a signal detector according to an embodiment of the present invention.
  • FIG 3 is an exemplary view for explaining a uniform radiation pattern using a point source according to an embodiment of the present invention.
  • FIG. 4 is a flowchart illustrating an artificial neural network learning data collection process using clustering according to an embodiment of the present invention.
  • FIG. 5 is a view for explaining a process of determining a position of a radiation reaction according to an embodiment of the present invention.
  • FIG. 6 is an example of a 2D plane image according to a conventional radiation image processing method for comparison with a result of determining the position of a radiation reaction according to an embodiment of the present invention.
  • FIG. 7 is an example of a 2D plane image according to a determination result of a radiation reaction position according to an embodiment of the present invention.
  • FIG. 8 is a view for comparing the results of the determination of the radiation reaction positions shown in FIGS. 6 and 7.
  • FIG. 8 is a view for comparing the results of the determination of the radiation reaction positions shown in FIGS. 6 and 7.
  • FIG. 9 is a flowchart for explaining a multiplexing signal processing method using clustering and deep learning according to an embodiment of the present invention.
  • multiplexed signal processing apparatus detects a radiation reaction position
  • the multiplexing signal processing apparatus can be applied to various radiation imaging apparatuses, and the type of radiation to be detected is not limited.
  • FIG. 1 is a block diagram of a multiplexing signal processing apparatus using clustering and depth learning according to an embodiment of the present invention.
  • FIG. 2 is a diagram for explaining a configuration of a signal detector according to an embodiment of the present invention.
  • the multiplexing signal processing apparatus 100 is utilized in a radiation image processing apparatus such as a PET or a gamma camera and may be included as a component in the apparatus.
  • the multiplexing signal processing apparatus 100 includes a signal detector 1000 and a signal processor 1200.
  • the signal processor 1200 includes a radiation reaction positioner 1210 and a radiation reaction position detector (1220).
  • the signal detector 1000 includes a photodetector 1010 in which a plurality of optical sensors for converting a scintillation signal output from a scintillator for converting radiation into a scintillation signal are disposed.
  • a photodiode may be used as the optical sensor.
  • the signal detector 1000 further includes a scintillator 10 for outputting a scintillation signal according to a radiation signal, and the scintillator 10 is provided to correspond to one surface (for example, the lower surface) of the scintillator 10 And the photodetector 1010 is disposed at the position shown in FIG.
  • the photodetector 1010 includes a photosensor array in which photosensors each forming a plurality of pixels are arranged in a lattice pattern, and the electric signal through the photosensor array 1010 And a multiplexing circuit for outputting the multiplexed signal.
  • each optical sensor P30 has implemented 16 (i.e., 4 * 4) pixels.
  • 2B shows a photosensor array (i.e., a multiplexing circuit) in which a plurality of photosensors are arranged in a 3 * 3 structure to have a 12 * 12 pixel configuration.
  • a plurality of channels are implemented through the optical sensor array, and signals passing through each channel are transmitted to the signal processor 1200.
  • the multiplexing circuit according to an embodiment of the present invention is a multiplexing circuit having a resistance-based charge distribution characteristic.
  • the photodetector 1010 shown in FIG. 2B distributes a current for each signal generation position, and the divided current is divided into a specific ratio (hereinafter referred to as " , "Signal feature").
  • the type of the multiplexing circuit constituting the photodetector 1010 according to an exemplary embodiment of the present invention is not limited, and a multiplexing circuit using a DPC (Discretized Positioning Circuit) and a SCD (Symmetric Charge Division) It is possible. That is, the photodetector 1010 may be implemented as a multiplexing circuit in which the distributed voltage has a unique ratio for each signal generation position in the output channels A, B, C,
  • the signal processor 1200 identifies which optical sensor is output based on the characteristics of the signal output from the signal detector 1000, and multiplexes the radiation image signal therethrough.
  • the signal processor 1200 classifies the electrical signals output from the signal detector 1000 through a deep learning-based classifier to determine the position of the radiation reaction.
  • the radiation reaction location learning unit 1210 of the signal processor 1200 collects training data using the signal detector 1000 in the uniform radiation state using a point source and transmits the training data through the artificial neural network And generates a classifier.
  • the radiation reaction position determiner 1220 of the signal processor 1200 receives the electric signal detected from the signal detector 1000 for an arbitrary radiation emitting object. Then, the radiation reaction location determiner 1220 classifies the received electrical signal through the classifier generated in advance through the radiation reaction location learning unit 1210, and outputs a result of determining the radiation reaction location.
  • FIG 3 is an exemplary view for explaining a uniform radiation pattern using a point source according to an embodiment of the present invention.
  • a photodetector 1010 having a 12 * 12 pixel structure is shown as an example, as shown in FIG.
  • a scintillation signal is output to the entire position of the photosensor array (that is, the photodetector 1010) through the scintillator 10 at one time.
  • the radiation reaction position learning unit 1210 of the signal processor 1200 acquires an electric signal output from the photodetector 1010, and clusters the obtained electric signal by position. Then, the radiation reaction location learning unit 1210 generates a deep learning classifier by using the electrical signal sets obtained using the result of the clustering as learning data of the artificial neural network.
  • the multiplexing signal processing apparatus 100 uses an electric signal output at one time by the uniformly radiated radiation, so that a separate sighting device is not necessary,
  • the reference data for determining the position of the radiation reaction can be generated without the control process. That is, the multiplexing signal processing apparatus 100 according to an embodiment of the present invention has an effect of greatly reducing the physical / temporal cost as compared with the conventional radiation image processing method.
  • FIG. 4 is a flowchart illustrating an artificial neural network learning data collection process using clustering according to an embodiment of the present invention.
  • the radiation reaction location learning unit 1210 uniformly radiates radiation to the scintillator 10 using a point source as shown in FIG. In this state, the data for the equivalent source (i. E., The electrical signal output through the signal detector 1000 is obtained (S410).
  • the radiation reaction position learning unit 1210 groups the acquired data according to the position (S420).
  • the radiation reaction location learning unit 1210 may apply a non-mapping clustering algorithm such as a Watershed Algorithm or a K Means Algorithm as an algorithm for processing the clustering.
  • a non-mapping clustering algorithm such as a Watershed Algorithm or a K Means Algorithm as an algorithm for processing the clustering.
  • the radiation response location learning unit 1210 may process the clustering using algorithm 1 below.
  • the radiation reaction position learning unit 1210 judges whether or not the number of pixels corresponding to the optical sensor array of the photodetector 1010 matches the number of sets of data generated through clustering (i.e., obtained signals) (S430).
  • step S430 If it is determined in step S430 that the number of detector pixels is not equal to the number of clustered data sets (i.e., the number of data sets is less than the number of detector pixels), the radiation reaction location learning unit 1210 proceeds to step S410 And the procedure from step S430 is repeated. That is, a series of processes of uniformly emitting radiation and then collecting and outputting the output electrical signals regardless of positions are repeated a plurality of times.
  • the radiation response position learning unit 1210 maps the physical position of the pixels of the actual detector to each clustered data set (S440).
  • the radiation reaction location learning unit 1210 learns the artificial neural network using each data set in which the mapping of step S440 is completed as learning data (S450). Through this, an artificial neural network-based classifier can be generated as reference data for determining the position of a radiation reaction to an arbitrary object.
  • the signal processor 1200 identifies a radiation reaction location through a deep learning classifier generated using clustering.
  • FIG. 5 is a view for explaining a process of determining a position of a radiation reaction according to an embodiment of the present invention.
  • the signal processor 1200 detects an electrical signal by radiation emitted from an arbitrary radiation emitting object through the signal detector 1000 as shown in FIG. 5 (a).
  • FIG. 5 (a) detection of an electrical signal through the photodetector 1010 having a 12 * 12 pixel structure is shown as an example in FIGS. 2 and 3.
  • the radiation reaction location determiner 1220 of the signal processor 1200 extracts signal characteristics of the electric signal detected through the signal detector 1000. [ At this time, the radiation reaction position determiner 1220 extracts the signal characteristics of the detected electric signal based on the inherent current or voltage ratio according to the position on the optical sensor array for each optical sensor of the optical detector 1010.
  • the radiation reaction position determiner 1220 inputs the extracted signal characteristics to the deep learning classifier generated in advance through the radiation reaction position learning unit 1220, And outputs the classification result of the reaction position.
  • the artificial neural network is also learned on the basis of extracted signal characteristics in the same manner as the method of extracting the signal characteristics of the detected electric signals.
  • FIG. 6 is an example of a 2D plane image according to a conventional radiation image processing method for comparing with a result of determination of a radiation reaction position according to an embodiment of the present invention.
  • FIG. And is an example of a 2D plane image according to the result.
  • FIG. 8 is a view for comparing the results of the determination of the radiation reaction positions shown in FIGS. 6 and 7.
  • 6 (a) and 6 (b) show a result of performing pixel region division in a 2D plane image generated by a gamma ray detector using a conventional radiographic imaging technique. That is, as shown in FIG. 6 (b), image distortion occurs due to nonlinearity of the signal, and the performance of dividing the pixel region is largely deteriorated.
  • FIG. 7A and 7B illustrate a result of performing pixel region division on a 2D plane image obtained according to a multiplexing signal processing method using clustering and depth learning according to an embodiment of the present invention .
  • FIG. 7B and FIG. 6B it can be seen that the image of FIG. 7B is clearly distinguished from the adjacent pixel region, and the performance of the pixel region segmentation is greatly improved.
  • FIG. 8 compares the peak-to-valley ratios as indices for determining the clarity of segmentation in each 2D plane image of FIGS. 6 and 7.
  • FIG. 8A is a graph showing signal values of coordinates on the x-axis and y-axis in FIG. 6, and
  • FIG. 8B is a graph showing signal values of coordinates on the x- to be.
  • the peak-to-valley ratio is 4.6 in the x-axis and 3.7 in the y-axis in the result of detecting the gamma ray reaction position without using the artificial neural network, as shown in FIG. 6,
  • the peak-to-valley ratio was 102.9 on the x-axis and 77.9 on the y-axis, and it was confirmed that the peak-to-valley ratio increased more than 20 times on each axis. That is, it can be seen that the resolution and pixel resolution of the radiological image are greatly improved by using the clustering according to the embodiment of the present invention and the radiation reaction location detecting method using the artificial neural network, compared to the conventional radiation reaction position detecting system.
  • FIG. 9 is a flowchart illustrating a method of processing a multiplexing signal using clustering and artificial neural networks according to an embodiment of the present invention.
  • a classifier applying clustering and deep learning to a signal detected from the uniform source is generated (S910).
  • a multiplexing signal processing method converts a scintillation signal output from a scintillator into an electrical signal through an optical sensor array in which a plurality of optical sensors are arranged in a state in which the radiation is uniformly radiated to the scintillator through the point radiation source.
  • the multiplexing signal processing method then obtains the electrical signal output through the optical sensor array and clusters the obtained electrical signal by position.
  • a series of processes of randomly acquiring and clustering the electric signals after uniformly emitting radiation are repeated a plurality of times, and when the number of sets of electric signals by clustering is equal to the number of pixels corresponding to the optical sensors, And the position between the pixels.
  • the multiplexing signal processing method generates learned classifiers through deep learning using sets of clustering electrical signals as learning data.
  • the multiplexing signal processing method detects a signal (i.e., an electrical signal corresponding to the emitted radiation) from any object to which the actual radiation reaction position is to be determined through the signal detector (S920).
  • a signal i.e., an electrical signal corresponding to the emitted radiation
  • a gamma ray emitted from an arbitrary radiation emitting object converts a scintillation signal outputted through a scintillator to an electric signal through an optical sensor array and outputs the electric signal.
  • the detected signal is classified through the classifier generated in advance and the gamma ray reaction position is discriminated (S930), and the discriminated result of the radiation reaction position is outputted (S940).
  • the final output value may be a position coordinate value in the optical sensor array, and based on the position coordinate value, the multiplexing signal processing method may further process the step of generating a 2D / 3D radiographic image.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Medical Informatics (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Pathology (AREA)
  • Public Health (AREA)
  • Optics & Photonics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Veterinary Medicine (AREA)
  • Radiology & Medical Imaging (AREA)
  • Heart & Thoracic Surgery (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Nuclear Medicine (AREA)
  • Measurement Of Radiation (AREA)

Abstract

La présente invention concerne un appareil et un procédé de traitement de signaux avec multiplexage utilisant un regroupement et un apprentissage profond. L'appareil de traitement de signaux avec multiplexage comporte: un détecteur de signaux comprenant un réseau de capteurs optiques constitué par une pluralité de capteurs optiques servant à convertir un signal de scintillation, qui est délivré à partir d'un scintillateur pour convertir un rayonnement en un signal de scintillation, en un signal électrique; et un processeur de signaux servant à classifier le signal électrique délivré à partir du détecteur de signaux par l'intermédiaire d'un classificateur généré précédemment basé sur un apprentissage profond, de façon à identifier une position de réaction au rayonnement, le processeur de signaux: acquérant un signal électrique détecté par l'intermédiaire du réseau de capteurs optiques dans un état où un rayonnement est émis uniformément vers le scintillateur par l'intermédiaire d'une source ponctuelle de rayonnement; effectuant un regroupement du signal électrique acquis pour chaque position; lorsque le nombre d'ensembles des signaux électriques regroupés, qui est obtenu en répétant à de multiples reprises une série de processus d'émission uniforme de rayonnement puis en acquérant et en regroupant des signaux électriques, associant le nombre de pixels correspondant à chaque capteur optique, faisant correspondre des positions entre les ensembles et les pixels; et utilisant les ensembles des signaux électriques regroupés comme données d'apprentissage du classificateur.
PCT/KR2018/013495 2017-12-20 2018-11-08 Appareil et procédé de traitement de signaux avec multiplexage utilisant un regroupement et un apprentissage profond WO2019124731A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR1020170176455A KR102114334B1 (ko) 2017-12-20 2017-12-20 군집화와 딥러닝을 이용한 멀티플렉싱 신호 처리 장치 및 방법
KR10-2017-0176455 2017-12-20

Publications (1)

Publication Number Publication Date
WO2019124731A1 true WO2019124731A1 (fr) 2019-06-27

Family

ID=66992982

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2018/013495 WO2019124731A1 (fr) 2017-12-20 2018-11-08 Appareil et procédé de traitement de signaux avec multiplexage utilisant un regroupement et un apprentissage profond

Country Status (2)

Country Link
KR (1) KR102114334B1 (fr)
WO (1) WO2019124731A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110226943A (zh) * 2019-07-05 2019-09-13 上海联影医疗科技有限公司 光子到达探测器的参数计算方法、装置和计算机设备

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102273576B1 (ko) * 2019-11-27 2021-07-06 정재필 기계학습 기반 3차원 공간상 선원의 분포 및 이동 경로 예측 시스템과 그 방법
KR102539695B1 (ko) * 2019-12-10 2023-06-02 연세대학교 원주산학협력단 핵시설 무인감시 및 인공지능 기반의 자동경보 시스템

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2009038726A (ja) * 2007-08-03 2009-02-19 Panasonic Corp 物理量検知装置およびその駆動方法
KR20100023468A (ko) * 2008-08-22 2010-03-04 중앙대학교 산학협력단 다채널 신호의 데이터 획득 시스템
KR101330117B1 (ko) * 2012-07-24 2013-11-20 서울대학교산학협력단 다채널 광전소자를 이용하는 양전자방출 단층촬영장치
US20150276953A1 (en) * 2012-10-26 2015-10-01 Universiteit Gent Calibration of monolithic crystal-based detectors
KR20160050686A (ko) * 2014-10-30 2016-05-11 서강대학교산학협력단 다중 문턱전압을 이용한 의료 영상기기의 신호처리 시스템 및 방법

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101912715B1 (ko) * 2012-11-20 2018-10-29 삼성전자주식회사 방사선이 방출된 위치의 분포를 추정하는 방법 및 장치

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2009038726A (ja) * 2007-08-03 2009-02-19 Panasonic Corp 物理量検知装置およびその駆動方法
KR20100023468A (ko) * 2008-08-22 2010-03-04 중앙대학교 산학협력단 다채널 신호의 데이터 획득 시스템
KR101330117B1 (ko) * 2012-07-24 2013-11-20 서울대학교산학협력단 다채널 광전소자를 이용하는 양전자방출 단층촬영장치
US20150276953A1 (en) * 2012-10-26 2015-10-01 Universiteit Gent Calibration of monolithic crystal-based detectors
KR20160050686A (ko) * 2014-10-30 2016-05-11 서강대학교산학협력단 다중 문턱전압을 이용한 의료 영상기기의 신호처리 시스템 및 방법

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110226943A (zh) * 2019-07-05 2019-09-13 上海联影医疗科技有限公司 光子到达探测器的参数计算方法、装置和计算机设备

Also Published As

Publication number Publication date
KR102114334B1 (ko) 2020-05-22
KR20190074817A (ko) 2019-06-28

Similar Documents

Publication Publication Date Title
WO2019124731A1 (fr) Appareil et procédé de traitement de signaux avec multiplexage utilisant un regroupement et un apprentissage profond
WO2019132589A1 (fr) Dispositif de traitement d'images et procédé de détection d'objets multiples
CN105452902B (zh) 针对小间距闪烁晶体阵列的像素识别
WO2021071288A1 (fr) Procédé et dispositif de formation de modèle de diagnostic de fracture
EP0270761A3 (fr) Système image de rayonnement
WO2016204402A1 (fr) Procédé d'inspection de défaut de composant, et appareil associé
WO2019132590A1 (fr) Procédé et dispositif de transformation d'image
WO2015182904A1 (fr) Appareil d'étude de zone d'intérêt et procédé de détection d'objet d'intérêt
WO2019132592A1 (fr) Dispositif et procédé de traitement d'image
WO2021020866A1 (fr) Système et procédé d'analyse d'images pour surveillance à distance
WO2022010065A1 (fr) Appareil de fusion d'images à deux particules sur la base d'une ouverture codée
WO2024210406A1 (fr) Appareil et système de détection d'objet caché sur la base d'un modèle d'apprentissage d'intelligence artificielle à l'aide d'une image de balayage thz
US20030076992A1 (en) Neural network based element, image pre-processor, and method of pre-processing using a neural network
WO2017052203A1 (fr) Dispositif d'imagerie utilisant une pluralité de lentilles
WO2021071258A1 (fr) Dispositif et procédé d'apprentissage d'image de sécurité mobile basés sur l'intelligence artificielle
WO2012008635A1 (fr) Détecteur de rayons x et procédé de détection
WO2022131469A1 (fr) Procédé et dispositif permettant de séparer une empreinte digitale d'une image d'empreinte digitale en chevauchement en utilisant un algorithme d'apprentissage profond
US20230063565A1 (en) A computer-implemented method for identifying and localizing radiation events and a pixilated radiation detector for carrying out the method
WO2021125539A1 (fr) Dispositif, procédé et programme informatique de classification d'objets présents dans une image
WO2018088872A1 (fr) Procédé de traitement de signaux de multiplexage utilisant des impulsions bipolaires
Furnée et al. Precision 3-D motion analysis system for real-time applications
WO2017086741A1 (fr) Dispositif de détection de position de signal utilisant un détecteur à réseau permettant de réduire les erreurs
WO2018159868A1 (fr) Procédé de segmentation de région d'image médicale et dispositif s'y rapportant
WO2023277644A1 (fr) Procédé de fourniture d'informations requises pour le diagnostic d'une maladie buccale, et appareil pour sa mise en œuvre
KR20220056604A (ko) 멀티플렉싱 신호 처리 시스템

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18892412

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18892412

Country of ref document: EP

Kind code of ref document: A1