METHOD OF EVALUATING LESION IN MEDICAL IMAGE

• Main Authors: KIM, Hyeongsub, KIM, Kyungdoc, LEE, Hongseok, KIM, Yeong Won
• Format: Patent
• Language: English; French; Korean
• Published: 04.08.2022
• Subjects: BEER; BIOCHEMISTRY; CALCULATING; CHEMISTRY; COMPOSITIONS OR TEST PAPERS THEREFOR; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL ORENZYMOLOGICAL PROCESSES; COUNTING; ENZYMOLOGY; HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATIONTECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING ORPROCESSING OF MEDICAL OR HEALTHCARE DATA; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEICACIDS OR MICROORGANISMS; METALLURGY; MICROBIOLOGY; MUTATION OR GENETIC ENGINEERING; PHYSICS; PROCESSES OF PREPARING SUCH COMPOSITIONS; SPIRITS; VINEGAR; WINE

According to one embodiment of the present disclosure, a medical image-based lesion analysis method performed by a computing device is disclosed. The method may comprise the steps of: extracting a first feature of each of a plurality of tiles included in an input image by using a pre-trained artificial neural network; and by using the pre-trained artificial neural network, extracting a second feature for the entirety of the input image on the basis of information on whether a lesion is present for each tile and information on a pattern of a lesion in each tile, generated on the basis of first features of the plurality of tiles. Selon un mode de réalisation de la présente divulgation, est divulguée une méthode d'analyse de lésion fondée sur une image médicale mise en œuvre par un dispositif informatique. La méthode peut consister : à extraire une première caractéristique de chaque pavé d'une pluralité de pavés compris dans une image d'entrée à l'aide d'un réseau neuronal artificiel pré-entraîné ; et à l'aide du réseau neuronal artificiel pré-entraîné, à extraire une seconde caractéristique pour la totalité de l'image d'entrée en fonction d'informations indiquant si une lésion est présente pour chaque pavé et d'informations sur un motif d'une lésion dans chaque pavé, générée en fonction de premières caractéristiques de la pluralité de pavés. 본 개시의 일 실시예에 따라 컴퓨팅 장치에 의해 수행되는 의료 영상 기반의 병변 분석 방법이 개시된다. 상기 방법은, 사전 학습된 인공 신경망을 사용하여 입력 영상에 포함된 복수의 타일들 각각의 제 1 특징을 추출하는 단계; 및 상기 사전 학습된 인공 신경망을 사용하여, 상기 복수의 타일들의 제 1 특징들을 기초로 생성된 타일 별 병변의 존재 여부에 관한 정보 및 타일 별 병변의 패턴에 관한 정보를 기초로, 상기 입력 영상 전체에 대한 제 2 특징을 추출하는 단계를 포함할 수 있다.