{"title":"利用竞争神经网络研究人类乳腺疾病的显微图像","authors":"R. Allan, W. Kinsner","doi":"10.1109/CCECE.2001.933698","DOIUrl":null,"url":null,"abstract":"Competitive neural networks offer a unique opportunity to extract features from medical images objectively. An advantage of this approach is that medical image analysis could be automated or semi-automated. This automation could lead to improved precision and accuracy of diagnostic interpretation, while semi-automation could achieve much the same goal and would serve as a natural stepping-stone to full automation. This paper shows that all types of competitive neural networks can extract general features from images obtained through a microscope of four types of human breast disease, two benign and two malignant. Assessed qualitatively, the features broadly encompass thresholding and edge detection. These features are extracted regardless of supervision or lack of supervision. To visual inspection, there are no obvious sharp distinctions between benign and malignant diagnoses, the most important distinction in tissue diagnosis.","PeriodicalId":184523,"journal":{"name":"Canadian Conference on Electrical and Computer Engineering 2001. Conference Proceedings (Cat. No.01TH8555)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2001-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"10","resultStr":"{\"title\":\"A study of microscopic images of human breast disease using competitive neural networks\",\"authors\":\"R. Allan, W. Kinsner\",\"doi\":\"10.1109/CCECE.2001.933698\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Competitive neural networks offer a unique opportunity to extract features from medical images objectively. An advantage of this approach is that medical image analysis could be automated or semi-automated. This automation could lead to improved precision and accuracy of diagnostic interpretation, while semi-automation could achieve much the same goal and would serve as a natural stepping-stone to full automation. This paper shows that all types of competitive neural networks can extract general features from images obtained through a microscope of four types of human breast disease, two benign and two malignant. Assessed qualitatively, the features broadly encompass thresholding and edge detection. These features are extracted regardless of supervision or lack of supervision. To visual inspection, there are no obvious sharp distinctions between benign and malignant diagnoses, the most important distinction in tissue diagnosis.\",\"PeriodicalId\":184523,\"journal\":{\"name\":\"Canadian Conference on Electrical and Computer Engineering 2001. Conference Proceedings (Cat. No.01TH8555)\",\"volume\":\"17 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2001-05-13\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"10\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Canadian Conference on Electrical and Computer Engineering 2001. Conference Proceedings (Cat. No.01TH8555)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/CCECE.2001.933698\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Canadian Conference on Electrical and Computer Engineering 2001. Conference Proceedings (Cat. No.01TH8555)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CCECE.2001.933698","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
A study of microscopic images of human breast disease using competitive neural networks
Competitive neural networks offer a unique opportunity to extract features from medical images objectively. An advantage of this approach is that medical image analysis could be automated or semi-automated. This automation could lead to improved precision and accuracy of diagnostic interpretation, while semi-automation could achieve much the same goal and would serve as a natural stepping-stone to full automation. This paper shows that all types of competitive neural networks can extract general features from images obtained through a microscope of four types of human breast disease, two benign and two malignant. Assessed qualitatively, the features broadly encompass thresholding and edge detection. These features are extracted regardless of supervision or lack of supervision. To visual inspection, there are no obvious sharp distinctions between benign and malignant diagnoses, the most important distinction in tissue diagnosis.