{"id":6167,"date":"2019-05-18T11:55:58","date_gmt":"2019-05-18T02:55:58","guid":{"rendered":"http:\/\/www.is.doshisha.ac.jp\/news\/?p=6167"},"modified":"2021-03-09T11:09:42","modified_gmt":"2021-03-09T02:09:42","slug":"digestive-disease-week-2019","status":"publish","type":"post","link":"https:\/\/is.doshisha.ac.jp\/news\/?p=6167","title":{"rendered":"Digestive Disease Week 2019"},"content":{"rendered":"<p><!--\u3000\u2193\u2193\u2193\u3000\u6bb5\u843d\u533a\u5207\u308a\u306f (\u6587\u7ae0) \u3067\u56f2\u3080\u3000\u2193\u2193\u2193\u3000--><br \/>\n2019\u5e745\u670818\u65e5\uff5e5\u670821\u65e5\u306b\u304b\u3051\u3066\u30a2\u30e1\u30ea\u30ab \u30b5\u30f3\u30c7\u30a3\u30a8\u30b4\u306b\u3066\u958b\u50ac\u3055\u308c\u307e\u3057\u305fDigestive Disease Week2019\u306b\u53c2\u52a0\u3044\u305f\u3057\u307e\u3057\u305f\uff0e\u672c\u5b66\u4f1a\u306f\uff0c\u80c3\u8178\u75c5\u5b66\uff0c\u809d\u81d3\u75c5\u5b66\uff0c\u5185\u8996\u93e1\u691c\u67fb\uff0c\u80c3\u8178\u5916\u79d1\u306e\u5206\u91ce\u306b\u304a\u3051\u308b\u533b\u5e2b\uff0c\u7814\u7a76\u8005\u304a\u3088\u3073\u305d\u306e\u696d\u754c\u306e\u6700\u5927\u7d44\u7e54\u3067\u3042\u308a\uff0c\u4e00\u6d41\u306e\u7814\u7a76\u306b\u5bfe\u3059\u308b\u30a2\u30d7\u30ed\u30fc\u30c1\u3092\u5b66\u3073\u305d\u306e\u5206\u91ce\u306e\u5f15\u5c0e\u8005\u3068\u30b3\u30df\u30e5\u30cb\u30b1\u30fc\u30b7\u30e7\u30f3\u3092\u56f3\u308b\u3053\u3068\u3067\uff0c\u81ea\u5206\u306e\u7814\u7a76\u306b\u5bfe\u3059\u308b\u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u3092\u5f97\u308b\u3053\u3068\u3092\u76ee\u7684\u306b\u958b\u50ac\u3055\u308c\u3066\u3044\u307e\u3059\uff0e\u672c\u7814\u7a76\u5ba4\u304b\u3089\u306f\u5ee3\u5b89\u5148\u751f\uff0c\u5965\u6751\u99ff\u4ecb(M2)\uff0c\u6e05\u91ce\u5141\u8cb4(M1)\u306e3\u540d\u304c\u53c2\u52a0\u3057\u307e\u3057\u305f\uff0e\u767a\u8868\u5f62\u5f0f\u306f\u5965\u6751(M2)\uff0c\u6e05\u91ce(M1)\u304c\u30dd\u30b9\u30bf\u30fc\u767a\u8868\u3092\u884c\u3044\u307e\u3057\u305f\uff0e\u767a\u8868\u984c\u76ee\u306f\u4ee5\u4e0b\u306e\u901a\u308a\u3067\u3059\uff0e<\/p>\n<div style=\"border: 1.5px solid #CCC; padding: 7px; border-radius: 7px;\">\n<ul><\/p>\n<li>&#8220;UNSUPERVISED MACHINE LEARNING BASED AUTOMATIC DEMARCATION LINE DRAWING SYSTEM ON NBI IMAGES OF EARLY GASTRIC CANCER&#8221;<br \/>\u3000S.OKUMURA, T.YASUDA, H.ICHIKAWA, S.HIWA, N.YAGI, T.HIROYASU.<\/li>\n<p><\/p>\n<li>&#8220;MACHINE-LEARNING-BASED AUTOMATIC DIAGNOSTIC SYSTEM USING LINKED COLOR IMAGING FOR HELICOBACTER PYLORI INFECTION: EXAMINATION OF IMAGE AFTER ERADICATION&#8221;<br \/>\u3000M.SEINO, T.YASUDA, H.ICHIKAWA, S.HIWA, N.YAGI, T.HIROYASU.<\/li>\n<\/ul>\n<\/div>\n<p><!--\u3000\u2193\u2193\u2193\u3000\u7d9a\u304d\u306b\u6587\u7ae0\u3092\u5165\u529b\u3057\u3066\u304f\u3060\u3055\u3044\u3000\u2193\u2193\u2193\u3000--><br \/>\n<a href=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/ddw_sokumura.png\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-medium wp-image-6171\" src=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/ddw_sokumura-253x300.png\" alt=\"\" width=\"253\" height=\"300\" \/><\/a> <a href=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/DDW_mseino.png\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-medium wp-image-6172\" src=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/DDW_mseino-252x300.png\" alt=\"\" width=\"252\" height=\"300\" \/><\/a><br \/>\n\u5b66\u4f1a\u306b\u306f\u5171\u540c\u7814\u7a76\u8005\u3067\u3042\u308b\u671d\u65e5\u5927\u5b66\u75c5\u9662\u6d88\u5316\u5668\u5185\u79d1\u306e\u516b\u6728\u5148\u751f\uff0c\u540c\u5fd7\u793e\u5927\u5b66\u751f\u547d\u533b\u79d1\u5b66\u90e8\u306e\u5e02\u5ddd\u5148\u751f\u3082\u53c2\u52a0\u3055\u308c\u307e\u3057\u305f\uff0eDDW\u306f\u533b\u5b66\u5b66\u4f1a\u3067\u3059\u304c\uff0c\u8fd1\u5e74\u81e8\u5e8a\u306b\u304a\u3051\u308bAI\u306e\u6ce8\u76ee\u3082\u3042\u308a\uff0c\u300c\u533b\u7528\u753b\u50cf\u306b\u5bfe\u3059\u308b\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u8a3a\u65ad\u300d\u306e\u30bb\u30c3\u30b7\u30e7\u30f3\u304c\u958b\u8a2d\u3055\u308c\u3066\u304a\u308a\u307e\u3057\u305f\uff0e\u6211\u3005\u306f2\u540d\u3068\u3082\u672c\u30bb\u30c3\u30b7\u30e7\u30f3\u3067\u767a\u8868\u3092\u884c\u3044\u307e\u3057\u305f\uff0e\u5468\u308a\u306b\u306f\u753b\u50cf\u51e6\u7406\u3084\u6a5f\u68b0\u5b66\u7fd2\u3092\u7528\u3044\u3066\u8a3a\u65ad\u652f\u63f4\u30b7\u30b9\u30c6\u30e0\u306e\u69cb\u7bc9\u3092\u884c\u3063\u3066\u3044\u308b\u5148\u751f\u65b9\u304c\u591a\u304f\u304a\u3089\u308c\u307e\u3057\u305f\uff0e\u767a\u8868\u6642\u9593\u306f2\u6642\u9593\u3067\u3057\u305f\u304c\uff0c\u65e5\u672c\u306e\u5148\u751f\u306e\u307f\u306a\u3089\u305a\uff0c\u6d77\u5916\u306e\u5148\u751f\u3084\u5de5\u5b66\u8005\u306e\u65b9\u3005\u3082\u591a\u6570\u6211\u3005\u306e\u7814\u7a76\u306b\u8208\u5473\u3092\u3082\u3063\u3066\u304f\u3060\u3055\u308a\uff0c\u7d76\u3048\u9593\u306a\u304f\u30c7\u30a3\u30b9\u30ab\u30c3\u30b7\u30e7\u30f3\u3092\u884c\u3046\u3053\u3068\u304c\u3067\u304d\u307e\u3057\u305f\uff0e\u672c\u30c7\u30a3\u30b9\u30ab\u30c3\u30b7\u30e7\u30f3\u3088\u308a\uff0c\u73fe\u72b6\u306e\u8ab2\u984c\u304c\u660e\u78ba\u306b\u306a\u308a\uff0c\u307e\u305f\u4ed6\u306e\u767a\u8868\u3092\u805e\u304f\u3053\u3068\u3067\uff0c\u4eca\u5f8c\u306e\u7814\u7a76\u65b9\u91dd\u306e\u30d2\u30f3\u30c8\u3082\u5f97\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u4eca\u56de\u5f97\u305f\u77e5\u898b\u3068\u30e2\u30c1\u30d9\u30fc\u30b7\u30e7\u30f3\u3092\u4eca\u5f8c\u306e\u7814\u7a76\u306b\u6d3b\u304b\u3057\uff0c\u4fee\u4e86\u307e\u3067\u306e\u6b8b\u308a9\u30f6\u6708\u3092\u6709\u610f\u7fa9\u306a\u6642\u9593\u306b\u3057\u305f\u3044\u3068\u601d\u3044\u307e\u3059\uff0e\u307e\u305f\u5f8c\u8f29\u306e\u80b2\u6210\u306b\u3082\u3088\u308a\u4e00\u5c64\u529b\u3092\u5165\u308c\u3066\u3044\u3053\u3046\u3068\u601d\u3044\u307e\u3059\uff0e<br \/>\n<!--\u3000\u2193\u2193\u2193\u3000\u753b\u50cf\u306e\u633f\u5165\uff08\u633f\u5165\u3057\u305f\u3044\u5834\u6240\u306b\u30ab\u30fc\u30bd\u30eb\u3092\u7f6e\u3044\u3066\u304b\u3089\u300c\u30e1\u30c7\u30a3\u30a2\u3092\u8ffd\u52a0\u300d\u3092\u30af\u30ea\u30c3\u30af\u3057\u3066\u8ffd\u52a0\u3059\u308b\uff09\u4f8b\u306f\u5fc5\u305a\u524a\u9664\u3059\u308b\u3053\u3068\u3000\u2193\u2193\u2193\u3000--><br \/>\n<a href=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/IMG_1010.jpg\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-medium wp-image-6190\" src=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/IMG_1010-300x225.jpg\" alt=\"\" width=\"300\" height=\"225\" \/><\/a> <a href=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/IMG_0160.jpg\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-medium wp-image-6189\" src=\"http:\/\/www.is.doshisha.ac.jp\/news\/wp-content\/uploads\/2019\/05\/IMG_0160-300x225.jpg\" alt=\"\" width=\"300\" height=\"225\" \/><\/a><br \/>\n&nbsp;<br \/>\n<!--\u3000\u2193\u2193\u2193\u3000\u4ee5\u4e0b\u306e\u5b66\u5e74\uff0c\u82d7\u5b57\u3092\u7de8\u96c6\u3057\u3066\u304f\u3060\u3055\u3044\uff08\u5b66\u5e74\uff0b\u534a\u89d2\u30b9\u30da\u30fc\u30b9\uff0b\u82d7\u5b57\uff09\u3000\u2193\u2193\u2193\u3000--><br \/>\n\u3010\u6587\u8cac\uff1aM2 \u5965\u6751(\u99ff)\u3011<br \/>\n<!--\u5fc5\u305a\u300cText\u300d\u30bf\u30d6\u3067\u7de8\u96c6\u3057\u3066\u304f\u3060\u3055\u3044\uff0e\u300c\u30d3\u30b8\u30e5\u30a2\u30eb\u300d\u30bf\u30d6\u3067\u7de8\u96c6\u3059\u308b\u3068\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u304c\u5909\u5316\u3059\u308b\u53ef\u80fd\u6027\u304c\u3042\u308a\u307e\u3059\uff0e--><br \/>\n&nbsp;<br \/>\n<!--more--><br \/>\n<strong>\u5b66\u4f1a\u53c2\u52a0\u5831\u544a\u66f8<\/strong><\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"147\"><strong>\u00a0<\/strong><br \/>\n<strong>\u5831\u544a\u8005\u6c0f\u540d<\/strong><\/td>\n<td width=\"373\">&nbsp;<br \/>\n\u5965\u6751\u99ff\u4ecb<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u767a\u8868\u8ad6\u6587\u30bf\u30a4\u30c8\u30eb<\/strong><\/td>\n<td width=\"373\">\u6559\u5e2b\u306a\u3057\u5b66\u7fd2\u306b\u57fa\u3065\u304f\u65e9\u671f\u80c3\u764cNBI\u5185\u8996\u93e1\u753b\u50cf\u306b\u5bfe\u3059\u308bDemarcation Line\u81ea\u52d5\u8a3a\u65ad\u30b7\u30b9\u30c6\u30e0<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u767a\u8868\u8ad6\u6587\u82f1\u30bf\u30a4\u30c8\u30eb<\/strong><\/td>\n<td width=\"373\">UNSUPERVISED\u00a0MACHINE LEARNING BASED AUTOMATIC DEMARCATION LINE DRAWING SYSTEM ON NBI IMAGES OF EARLY GASTRIC CANCER<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u8457\u8005<\/strong><\/td>\n<td width=\"373\">\u5965\u6751\u99ff\u4ecb\uff0c\u5b89\u7530\u525b\u58eb\uff0c\u5e02\u5ddd\u5bdb\uff0c\u65e5\u548c\u609f\uff0c\u516b\u6728\u4fe1\u660e\uff0c\u5ee3\u5b89\u77e5\u4e4b<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u4e3b\u50ac<\/strong><\/td>\n<td width=\"373\">Digestive Disease Week<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u8b1b\u6f14\u4f1a\u540d<\/strong><\/td>\n<td width=\"373\">DDW2019<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u4f1a\u5834<\/strong><\/td>\n<td width=\"373\">San Diego Convention Center<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u958b\u50ac\u65e5\u7a0b<\/strong><\/td>\n<td width=\"373\">2019\/05\/18-2019\/05\/21<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>&nbsp;<br \/>\n&nbsp;<\/p>\n<ol>\n<li>\u8b1b\u6f14\u4f1a\u306e\u8a73\u7d30<\/li>\n<\/ol>\n<p>2019\u5e7405\u670818\u65e5\u304b\u30892019\u5e7405\u670821\u65e5\u306b\u304b\u3051\u3066\uff0cSan Diego Convention Center\u306b\u3066\u958b\u50ac\u3055\u308c\u305fDigestive Disease Week 2019\uff08DDW2019\uff09\u306b\u53c2\u52a0\u3044\u305f\u3057\u307e\u3057\u305f\uff0e\u672c\u5b66\u4f1a\u306f\uff0c\u80c3\u8178\u75c5\u5b66\uff0c\u809d\u81d3\u75c5\u5b66\uff0c\u5185\u8996\u93e1\u691c\u67fb\uff0c\u80c3\u8178\u5916\u79d1\u306e\u5206\u91ce\u306b\u304a\u3051\u308b\u533b\u5e2b\uff0c\u7814\u7a76\u8005\u304a\u3088\u3073\u305d\u306e\u696d\u754c\u306e\u6700\u5927\u7d44\u7e54\u3067\u3042\u308a\uff0c\u4e00\u6d41\u306e\u7814\u7a76\u306b\u5bfe\u3059\u308b\u30a2\u30d7\u30ed\u30fc\u30c1\u3092\u5b66\u3073\u305d\u306e\u5206\u91ce\u306e\u5f15\u5c0e\u8005\u3068\u30b3\u30df\u30e5\u30cb\u30b1\u30fc\u30b7\u30e7\u30f3\u3092\u56f3\u308b\u3053\u3068\u3067\uff0c\u81ea\u5206\u306e\u7814\u7a76\u306b\u5bfe\u3059\u308b\u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u3092\u5f97\u308b\u3053\u3068\u3092\u76ee\u7684\u306b\u958b\u50ac\u3055\u308c\u3066\u3044\u307e\u3059\uff0e<br \/>\n&nbsp;<br \/>\n\u79c1\u306f\u5168\u65e5\u53c2\u52a0\u3044\u305f\u3057\u307e\u3057\u305f\uff0e\u672c\u7814\u7a76\u5ba4\u304b\u3089\u306f\u4ed6\u306b\uff0c\u5ee3\u5b89\u5148\u751f\uff0cM1\u306e\u5b66\u751f\u3068\u3057\u3066\u6e05\u91ce\u304f\u3093\u304c\u53c2\u52a0\u3057\u307e\u3057\u305f\uff0e\u4f01\u696d\u3082\u591a\u304f\u53c2\u52a0\u3057\u3066\u3044\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<ol start=\"2\">\n<li>\u7814\u7a76\u767a\u8868\n<ul>\n<li>\u767a\u8868\u6982\u8981<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n<p>\u79c1\u306f20\u65e5\u306eAGA-Computers in Endoscopy\u306e\u30bb\u30c3\u30b7\u30e7\u30f3\u3067\u767a\u8868\u3092\u884c\u3044\u307e\u3057\u305f\uff0e\u767a\u8868\u306f\u30dd\u30b9\u30bf\u30fc\u5f62\u5f0f\u3067\uff0c\u8a082\u6642\u9593\u53c2\u52a0\u8005\u306e\u65b9\u3068\u8b70\u8ad6\u3092\u884c\u3044\u307e\u3057\u305f\uff0e<br \/>\n\u30bf\u30a4\u30c8\u30eb\u306f\uff0cUNSUPERVISED MACHINE LEARNING BASED AUTOMATIC DEMARCATION LINE DRAWING SYSTEM ON NBI IMAGES OF EARLY GASTRIC CANCER\u3067\uff0c\u6559\u5e2b\u306a\u3057\u5b66\u7fd2\u3068\u8272\u7279\u5fb4\u3092\u7528\u3044\u3066NBI\u62e1\u5927\u89b3\u5bdf\u753b\u50cf\u304b\u3089\u65e9\u671f\u80c3\u764c\u306e\u5883\u754c\u7dda\uff08Demarcation Line\uff09\u306e\u8a3a\u65ad\u3092\u81ea\u52d5\u5316\u3057\uff0c30\u679a\u306e\u753b\u50cf\u306b\u304a\u3051\u308b\u611f\u5ea680.5%\u9054\u6210\u3057\u305f\u3053\u3068\u3092\u5831\u544a\u3057\u307e\u3057\u305f\uff0e\u4ee5\u4e0b\u306b\u6284\u9332\u3092\u8a18\u8f09\u81f4\u3057\u307e\u3059\uff0e<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">Introduction<br \/>\nIn this study, we aim to assist physicians in diagnosing early gastric cancer through an automatic demarcation line (DL) drawing system on NBI images. Recently, research on image diagnosis using supervised machine learning such as deep learning has attracted interest. However, for these methods, it is necessary to prepare a large number of supervised signals for the system to learn and ensure reliability. Moreover, in an image diagnosis by supervised learning methods, the feature to be used is not precise, making the result interpretation complicated. Therefore, in this study, we propose a new method based on unsupervised machine learning.<br \/>\n2.Aims and Methods<br \/>\nThe aim of this study is to automate the DL diagnosis of early gastric cancer for lesion resection. In this study, a diagnostic system using &#8220;unsupervised machine learning,&#8221; which does not require any supervised signals by a physician, is proposed. The proposed method is explained as follows. First, data is classified based on the value of the features, and the lesion site is specified. Specifically, gastric mucosal structures are quantified using 13 types of color features. Second, an NBI image is divided into 400 superpixels (a set of pixels with similar features) and calculated features are stored in each superpixel. Finally, each superpixel is classified based on the features by k-means clustering, and the lesion site is specified. In this system, three patterns of the detected DLs are displayed for each image. Moreover, a physician has the functionality of drawing the DL corresponding to each case. The algorithm of this system is shown in Fig. 1. In this study, we applied our system to early gastric cancer lesions (20 NBI magnified observation images) of 10 cases in which an endoscopic examination was performed at Asahi University Hospital from April 2014 to December 2016. A corresponding DL was detected for each case. The effectiveness of the proposed system was verified by comparing the detected result with the DL determined by the physician.<br \/>\n3.Results The average detection rate of the lesion area by the proposed system was 84.5 % (F-measure). According to the results, the obtained DL can replicate the DL determined by an experienced physician without supervised signals. Experimental images and their detection results are shown in Fig. 2. Thus, this system enabled the automatic detection of early gastric cancer by DL drawing and helped physicians in determining the DL.<br \/>\n4.Conclusion In this study, we developed a system that automatically detected the DL using unsupervised learning. The detection of the lesion area was accurate and it was possible to identify the DL without depending on the physician&#8217;s experience. Moreover, the features that used in this system were apparent and this system helped the physicians to understand the result. The accuracy achieved can be further improved in future.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<ul>\n<li>\u8cea\u7591\u5fdc\u7b54<\/li>\n<\/ul>\n<p>\u4eca\u56de\u306e\u8b1b\u6f14\u767a\u8868\u3067\u306f\uff0c\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u8cea\u7591\u3092\u53d7\u3051\u307e\u3057\u305f\uff0e\u672c\u767a\u8868\u3067\u306f\u8cea\u554f\u8005\u306e\u540d\u524d\u306f\u805e\u3044\u3066\u304a\u308a\u307e\u305b\u3093\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>1<\/strong><br \/>\n\u672a\u5206\u5316\u578b\u764c\u306e\u753b\u50cf\u3082\u6271\u3063\u3066\u3044\u308b\u306e\u304b\uff0c\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u3044\u307e\u306f\u5206\u5316\u578b\u764c\u306e\u307f\u3092\u5bfe\u8c61\u3068\u3057\u3066\u3044\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>2<\/strong><br \/>\ndeep learning\u3068\u306e\u9055\u3044\u306f\u3069\u3053\u304b\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u8b58\u5225\u5668\u306e\u69cb\u7bc9\u306b\u5fc5\u8981\u3068\u306a\u308b\u6559\u5e2b\u30c7\u30fc\u30bf\u3092\u4e00\u5207\u5fc5\u8981\u3068\u3057\u306a\u3044\u70b9\u3068\uff0c\u7279\u5fb4\u91cf\u3092\u81ea\u3089\u304c\u8a2d\u8a08\u3057\u3066\u3044\u308b\u70b9\u306b\u3042\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>3<\/strong><br \/>\nSensitivity\uff0cSpecificity\u306e\u8a55\u4fa1\u5024\uff08%\uff09\u306f\u3069\u306e\u3088\u3046\u306b\u6c7a\u5b9a\u3057\u3066\u3044\u308b\u306e\u304b\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u719f\u7df4\u533b\u306eDL\u3092\u5143\u306bsuperpixel\u5358\u4f4d\u3067\u6b63\u89e3\u30e9\u30d9\u30eb\u3092\u5272\u308a\u5f53\u3066\u3001\u691c\u51fa\u3055\u308c\u305fsuperpixel\u3068\u6b63\u89e3\u30e9\u30d9\u30eb\u306e\u6570\u306e\u6bd4\u8f03\u3067\u8a55\u4fa1\u3057\u3066\u3044\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>4<\/strong><br \/>\n\u7279\u5fb4\u91cf\u306e\u6c7a\u5b9a\u306b\u4f7f\u7528\u3057\u305f\u753b\u50cf\u306f\u4f55\u679a\u3067\uff0cvalidation\u306b\u4f7f\u7528\u3057\u305f\u753b\u50cf\u306f\u4f55\u679a\u304b\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u7279\u5fb4\u91cf\u8a2d\u8a08\u306f50\u679a\u3067\u884c\u3044\uff0c30\u679a\u3067validation\u3092\u884c\u306a\u3063\u3066\u3044\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>5<\/strong><br \/>\n\u30dd\u30b9\u30bf\u30fc\u306b\u8f09\u3063\u3066\u3044\u308b\u62e1\u5927NBI\u306e\u753b\u50cf\u304c\u975e\u5e38\u306b\u9bae\u660e\u306b\u64ae\u5f71\u3055\u308c\u3066\u3044\u308b\u3082\u306e\u304c\u591a\u3044\u304c\uff0c\u63d0\u4f9b\u5143\u306f\u3069\u3053\u304b\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u65e5\u672c\u306e\u671d\u65e5\u5927\u5b66\u75c5\u9662\u3068\u4eac\u90fd\u5e9c\u7acb\u533b\u79d1\u5927\u5b66\u75c5\u9662\u304b\u3089\u63d0\u4f9b\u3057\u3066\u3044\u305f\u3060\u3044\u3066\u3044\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n&nbsp;<br \/>\n&nbsp;<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>6<\/strong><br \/>\n\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u7d44\u3080\u306e\u306b\u4f7f\u7528\u3057\u3066\u3044\u308b\u8a00\u8a9e\u306f\u4f55\u3067\uff0c\u4e00\u679a\u3042\u305f\u308a\u306e\u51e6\u7406\u901f\u5ea6\u306f\u3069\u306e\u7a0b\u5ea6\u304b\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u8a00\u8a9e\u306fpython\u3067\uff0c\u73fe\u5728\u306f\u4e00\u679a\u3042\u305f\u308a20\u79d2\u7a0b\u5ea6\u3067\uff0c\u73fe\u5728\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u9ad8\u901f\u5316\u3092\u3057\u3066\u3044\u306a\u3044\u72b6\u614b\u3067\u3042\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>7<\/strong><br \/>\n\u8178\u4e0a\u76ae\u5316\u751f\u3092\u6709\u3059\u308b\u753b\u50cf\u3082\u6271\u3063\u3066\u3044\u308b\u304b\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u73fe\u5728\u306f\u6271\u3063\u3066\u3044\u306a\u3044\u3068\u7b54\u3048\u307e\u3057\u305f\u304c\uff0cLight Blue Crest\u3092\u8a8d\u3081\u308b\u753b\u50cf\u3082\u4f7f\u7528\u30c7\u30fc\u30bf\u306b\u542b\u307e\u308c\u3066\u3044\u308b\u305f\u3081\uff0c\u304a\u305d\u3089\u304f\u3053\u306e\u56de\u7b54\u306f\u9593\u9055\u3044\u3067\u3042\u308a\uff0c\u8178\u4e0a\u76ae\u5316\u751f\u306e\u753b\u50cf\u3082\u6271\u3063\u3066\u3044\u308b\u3068\u7b54\u3048\u308b\u3079\u304d\u3067\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>8<\/strong><br \/>\n\u4f7f\u7528\u3057\u3066\u3044\u308b\u8272\u7279\u5fb4\u91cf\u306f\u5177\u4f53\u7684\u306b\u3069\u306e\u3088\u3046\u306a\u3082\u306e\u3092\u7528\u3044\u3066\u3044\u308b\u306e\u304b\u3068\u3044\u3046\u8cea\u554f\u3092\u9802\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u73fe\u5728\u306fLab\u8272\u7a7a\u9593\u3068HSV\u8272\u7a7a\u9593\u3092\u7528\u3044\u3066NBI\u753b\u50cf\u3092\u5b9a\u91cf\u5316\u3057\u304a\u308a\uff0csuperpixel\u3054\u3068\u306b\u305d\u308c\u305e\u308c\u306e\u5e73\u5747\u5024\uff0c\u6700\u5927\u5024\uff0c\u6700\u5c0f\u5024\u3092\u683c\u7d0d\u3057\u3066\u7279\u5fb4\u91cf\u3068\u3057\u3066\u3044\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<ul>\n<li>\u611f\u60f3<\/li>\n<\/ul>\n<p>\u4eca\u56de\u306f\u533b\u5b66\u5b66\u4f1a\u3078\u306e\u53c2\u52a0\u3067\u3057\u305f\u304c\uff0cAI\u3068Computing\u306e\u30bb\u30c3\u30b7\u30e7\u30f3\u3067\u767a\u8868\u3057\u307e\u3057\u305f\uff0e\u5468\u308a\u306b\u306f\u79c1\u305f\u3061\u3068\u540c\u3058\u3088\u3046\u306bCAD\u306e\u7814\u7a76\u3092\u884c\u306a\u3063\u3066\u3044\u308b\u5148\u751f\u65b9\u304c\u591a\u304f\u304a\u3089\u308c\u307e\u3057\u305f\uff0e\u79c1\u306f\uff0cDeep Learning\u3092\u306f\u3058\u3081\u3068\u3059\u308b\u6559\u5e2b\u3042\u308a\u5b66\u7fd2\u3067\u5fc5\u8981\u306a\u4fe1\u983c\u5ea6\u306e\u9ad8\u3044\u6559\u5e2b\u4fe1\u53f7\u3092\u4e00\u5207\u5fc5\u8981\u3068\u3057\u306a\u3044\u70b9\uff0c\u307e\u305f\u8aac\u660e\u3067\u304d\u308b\u7279\u5fb4\u3092\u4f7f\u7528\u3057\u3066\u764c\u3092\u691c\u51fa\u3057\u3066\u3044\u308b\u70b9\u3092\u63a8\u3057\u30dd\u30a4\u30f3\u30c8\u3068\u3057\u3066\u767a\u8868\u3057\u307e\u3057\u305f\uff0e\u767a\u8868\u6642\u9593\u306f2\u6642\u9593\u3067\u3057\u305f\u304c\uff0c\u7d76\u3048\u9593\u306a\u304f\u591a\u304f\u306e\u5148\u751f\u65b9\u3068\u30c7\u30a3\u30b9\u30ab\u30c3\u30b7\u30e7\u30f3\u3059\u308b\u3053\u3068\u304c\u3067\u304d\uff0c\u975e\u5e38\u306b\u6709\u610f\u7fa9\u306a\u6642\u9593\u3092\u904e\u3054\u3059\u3053\u3068\u304c\u3067\u304d\u307e\u3057\u305f\uff0e\u30c7\u30a3\u30b9\u30ab\u30c3\u30b7\u30e7\u30f3\u306e\u4e2d\u3067\uff0c\u81e8\u5e8a\u306b\u304a\u3051\u308b\u4eca\u5f8c\u306eAI\u306e\u7acb\u3061\u4f4d\u7f6e\uff0c\u307e\u305f\u300c\u7279\u5fb4\u304c\u7406\u89e3\u3067\u304d\u308b\u30b7\u30b9\u30c6\u30e0\u300d\u306e\u91cd\u8981\u6027\u3092\u518d\u78ba\u8a8d\u3067\u304d\u307e\u3057\u305f\uff0e\u4eca\u56de\u306e\u5b66\u4f1a\u3067\u5f97\u3089\u308c\u305f\u77e5\u898b\u3092\u6d3b\u304b\u3057\uff0c\u4fee\u4e86\u307e\u3067\u306e\u6b8b\u308a9\u30f6\u6708\u3088\u308a\u4e00\u5c64\u7814\u7a76\u306b\u52b1\u307f\u305f\u3044\u3068\u601d\u3044\u307e\u3059\uff0e<br \/>\n&nbsp;<\/p>\n<ol start=\"3\">\n<li>\u8074\u8b1b<\/li>\n<\/ol>\n<p>\u4eca\u56de\u306e\u8b1b\u6f14\u4f1a\u3067\u306f\uff0c\u4e0b\u8a18\u306e4\u4ef6\u306e\u767a\u8868\u3092\u8074\u8b1b\u3057\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a\u3000ARTIFICIAL INTELLIGENCE-ASSISTED ENDOSCOPY IN CHARACTERIZATION OF GASTRIC LESIONS USING MAGNIFYING NARROW BAND IMAGING ENDOSCOPY<br \/>\n\u8457\u8005\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Sergey V. Kashin, Roman Kuvaev, Ekaterina Albertovna Kraynova, Olga Dunaeva, Alexander Rusakov, Evgeny Nikonov<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Poster session<br \/>\nAbstract\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Aims of this study were to develop and evaluate an artificial intelligence based system for histology prediction of gastric lesions using magnifying narrow band imaging (M-NBI) endoscopy. We selected and analyzed 265 endoscopy M-NBI images of gastric lesions from 128 patients who underwent upper M-NBI endoscopy (Olympus Exera GIF Q160Z, Lucera GIF Q260Z). All images were divided into four classes: (1) type A (n=46): non-neoplastic and non-metaplastic lesions with regular circular microsurface (MS) and regular microvascular (MV) patterns; (2) Type B (n=90): intestinal metaplasia with tubulo-villous MS and regular MV patterns; (3) Type C (n=74) neoplastic lesions with irregular MS or MV pattern; (4) artifacts (n=55). During automated classification quadrant areas were calculated on the image, geometrical and topological features were computed for every fragment. Using the greedy forward selection algorithm, the set of five most significant features were selected: three geometric features (the compactness of the MS pattern, the perimeter of the MS pattern, the average of area of the component of the MV pattern) two topological features (the kurtosis of the histogram of the 0-th persistence diagram of the image, the first norm of the 0-th persistence diagram of the signed distance function). Support vector machine (SVM) classifier was used for 4-class automated diagnosis.Training and testing were performed for every image by a k-fold method (k=10).<br \/>\nThe average percentage of correctly recognized areas was 91.4%. Classification precision (positive predictive value), recall (sensitivity), F-score for class A were 96.5 90.4 93.3 for class B were 93.7, 92.0, 92.9, respectively, for class C were 83.3, 91.3, 87.1, respectively, and for artifacts were 99.2, 91.7, 95.3, respectively<br \/>\nThe designed system based on the extraction of the geometrical and topological features from M-NBI image and analysis by SVM could provide effective recognition of three types of gastric mucosal changes.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u306f NBI\u62e1\u5927\u753b\u50cf\u304b\u3089\uff0c\u80c3\u75c5\u5909\u306e\u7d44\u7e54\u5b66\u7684\u5206\u985e\u3092\u4e88\u6e2c\u3057\uff0c\u80c3\u764c\u306b\u306a\u308b\u30ea\u30b9\u30af\u306e\u9ad8\u3044\u90e8\u5206\u3092\u30ab\u30e9\u30fc\u30de\u30c3\u30d7\u3067\u793a\u3059\u7814\u7a76\u3067\u3057\u305f\uff0e5\u7a2e\u985e\u306e\u69cb\u9020\u7279\u5fb4\u91cf\u3092\u4f7f\u7528\u3057\uff0c\u753b\u50cf\u3054\u3068\u306b\u5fae\u5c0f\u8840\u7ba1\u69cb\u7bc9\u50cf\u3068\u7c98\u819c\u5fae\u7d30\u69cb\u9020\u3092\u5143\u306bSVM\u3092\u7528\u3044\u30664\u3064\u306e\u7d44\u7e54\u5b66\u7684\u6240\u898b\u306b\u5206\u985e\u3055\u308c\uff0c\u305d\u3053\u304b\u3089\u764c\u306b\u306a\u308b\u30ea\u30b9\u30af\u304c\u9ad8\u3044\u90e8\u5206\u3092\u691c\u51fa\u3057\u3066\u3044\u307e\u3057\u305f\uff0e<br \/>\n\u4eca\u56de\u306e\u5b66\u4f1a\u306e\u4e2d\u3067\u306f\uff0c\u79c1\u304c\u884c\u306a\u3063\u3066\u3044\u308bNBI\u62e1\u5927\u306e\u7814\u7a76\u3068\u6700\u3082\u985e\u4f3c\u3057\u3066\u3044\u308b\u7814\u7a76\u3060\u3063\u305f\u3068\u601d\u3044\u307e\u3059\uff0e\u7279\u306b\u5fae\u5c0f\u8840\u7ba1\u69cb\u7bc9\u50cf\uff0c\u7c98\u819c\u5fae\u7d30\u69cb\u9020\u306e\u5b9a\u91cf\u5316\u306f\u53d6\u308a\u7d44\u3093\u3067\u307f\u305f\u3044\u5185\u5bb9\u3067\u3042\u3063\u305f\u305f\u3081\uff0c\u3053\u306e\u7814\u7a76\u3067\u884c\u308f\u308c\u3066\u3044\u308b\u5206\u985e\u6307\u6a19\u3084\u4f7f\u7528\u7279\u5fb4\u91cf\u7b49\uff0c\u4eca\u5f8c\u53c2\u8003\u306b\u3057\u305f\u3044\u3068\u601d\u3044\u307e\u3057\u305f\uff0e<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1aARTIFICIAL INTELLIGENCE-ASSISTED POLYP DETECTION SYSTEM FOR COLONOSCOPY<strong>,<\/strong> BASED ON THE LARGEST AVAILABLE COLLECTION OF CLINICAL VIDEO DATA FOR MACHINE LEARNING<br \/>\n\u8457\u8005\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Masashi Misawa<strong>,<\/strong> Shinei Kudo<strong>,<\/strong> Yuichi Mori<strong>,<\/strong> Tomonari Cho<strong>,<\/strong> Shinichi Kataoka<strong>,<\/strong> Yasuharu Maeda<strong>,<\/strong> Yushi Ogawa<strong>,<\/strong> Kenichi Takeda, Hiroki Nakamura<strong>,<\/strong> Katsuro Ichimasa, Naoya Toyoshima<strong>,<\/strong> Noriyuki Ogata<strong>,<\/strong> Toyoki Kudo<strong>,<\/strong> Tomokazu Hisayuki<strong>,<\/strong> Takemasa Hayashi<strong>,<\/strong> Kunihiko Wakamura<strong>,<\/strong> Toshiyuki Baba<strong>,<\/strong> Fumio Ishida<strong>,<\/strong> Hayato Itoh, Masahiro Oda<strong>,<\/strong> Kensaku Mori<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Poster session<br \/>\nAbstract \uff1a Background and aims<br \/>\nEradication of neoplastic polyps during colonoscopy is the most effective means of preventing colorectal cancer. However<strong>,<\/strong> one meta-analysis showed that about 26% of neoplasms are missed per colonoscopy <strong>(<\/strong>van Rijn et al.<strong>,<\/strong> Am J Gastroenterol<strong>,<\/strong> 2006.<strong>)<\/strong>. To tackle this issue<strong>,<\/strong> we have reported a pilot study of a real-time computer-aided detection <strong>(<\/strong>CADe<strong>)<\/strong> system that uses artificial intelligence <strong>(<\/strong>AI<strong>)<\/strong> to assess colonoscopy images <strong>(<\/strong>Misawa et al. Gastroenterology<strong>,<\/strong> 2018<strong>;<\/strong> Figure 1 shows an overview of the system<strong>)<\/strong>. In the current study<strong>,<\/strong> we aimed to create the largest collection of annotated colonoscopy video data <strong>(<\/strong>Mori et al. Endoscopy<strong>,<\/strong> 2017<strong>),<\/strong> re-build the CADe system<strong>,<\/strong> and evaluate its performance.<br \/>\nMethods<br \/>\nTo train the CADe<strong>,<\/strong> we retrospectively collected colonoscopy video data from December 2017 to August 2018. Expert endoscopists annotated every frame of each video as to the presence or absence of polyps. In all<strong>,<\/strong> 3<strong>,<\/strong>017<strong>,<\/strong>088 video frames <strong>(<\/strong>about 28 hours<strong>),<\/strong> including 930 colorectal polyps<strong>,<\/strong> were used to train the CADe. The CADe system uses a three-dimensional convolutional neural network<strong>,<\/strong> which is a deep learning method that is designed for analyzing spatiotemporal data such as video data. To evaluate the performance of the CADe<strong>,<\/strong> we analyzed completely separate video data derived from colonoscopy videos from 64 patients obtained from August 2018 to October 2018. Inclusion criteria were i<strong>)<\/strong> aged over 20 years<strong>;<\/strong> and ii<strong>)<\/strong> consent to participate. Exclusions were i<strong>)<\/strong> inflammatory bowel disease<strong>;<\/strong> and ii<strong>)<\/strong> polyposis. After these videos had been annotated by research assistants and audited by expert endoscopists regarding the presence or absence of polyps<strong>,<\/strong> size<strong>,<\/strong> and macroscopic type<strong>,<\/strong> they were treated as a gold standard. Sensitivity and false positive detection rate <strong>(<\/strong>FP<strong>)<\/strong> were calculated. We defined true positive detection as more than half the frames that included polyps being detected by the CADe. The FP was calculated using videos from patients with no polyps and was defined as the number of non-polyp frames detected by the system divided by the total number of non-polyp frames.<br \/>\nResults<br \/>\nIn all<strong>,<\/strong> 87 colorectal lesions from 47 patients and 17 patients without polyps were analyzed. Thirty-one of the 87 lesions were protruded polyps<strong>,<\/strong> 55 were flat polyps <strong>(<\/strong>including five laterally spreading tumors<strong>)<\/strong> and one was an advanced colon cancer. The median size was 5 <strong>(<\/strong>IQR: 3\u20136<strong>)<\/strong> mm. The sensitivity of the AI system was 86% <strong>(<\/strong>75\/87<strong>)<\/strong> and the FP for the non-polyp frames 26% <strong>(<\/strong>47<strong>,<\/strong>384\/185<strong>,<\/strong>444<strong>)<\/strong>. The sensitivities for diminutive polyps <strong>(<\/strong>\u22645 mm<strong>),<\/strong> protruded polyps<strong>,<\/strong> and flat polyps were 84%<strong>,<\/strong> 84%<strong>,<\/strong> and 87%<strong>,<\/strong> respectively.<br \/>\nConclusion<br \/>\nThe developed AI system has high sensitivity regardless of polyp size and morphology and has potential for automated polyp detection.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u3067\u306f\uff0c\u5927\u8178\u5185\u8996\u93e1\u691c\u67fb\u306b\u304a\u3051\u308b\u816b\u760d\u6027\u30dd\u30ea\u30fc\u30d7\u306e\u898b\u9003\u3057\u3092\u9632\u3050\u305f\u3081\uff0cCNN\u3092\u7528\u3044\u3066\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0\u3067\u30dd\u30ea\u30fc\u30d7\u306e\u767a\u898b\u652f\u63f4\u3092\u884c\u3046\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e\u7d50\u679c\uff0c\u611f\u5ea686%\u3092\u9054\u6210\u3057\uff0c\u30dd\u30ea\u30fc\u30d7\u306e\u5927\u304d\u3055\u3084\u5f62\u72b6\u306b\u304b\u304b\u308f\u3089\u305a\uff0c\u767a\u898b\u53ef\u80fd\u3068\u3044\u3046\u5831\u544a\u3067\u3057\u305f\uff0e\u6211\u3005\u306fCADx\u306e\u306e\u7814\u7a76\u3092\u884c\u306a\u3063\u3066\u304a\u308a\u307e\u3059\u304c\uff0c\u3053\u306e\u7814\u7a76\u306e\u3088\u3046\u306bCADe\u306e\u7814\u7a76\u3082\u4eca\u5f8c\u81e8\u5e8a\u5fdc\u7528\u304c\u8fd1\u3044\u3053\u3068\u3092\u5f37\u304f\u611f\u3058\u307e\u3057\u305f\uff0e\u672c\u7814\u7a76\u306f\uff0c10\u6708\u306e\u5c90\u961c\u770c\u6d88\u5316\u5668\u5185\u8996\u93e1\u61c7\u89aa\u4f1a\u3067\u6211\u3005\u304c\u516c\u6f14\u3092\u4f3a\u3063\u305f\u662d\u548c\u5927\u5b66\u6a2a\u6d5c\u5e02\u5317\u90e8\u75c5\u9662\u306e\u4e09\u6fa4\u5148\u751f\u306b\u3088\u308b\u767a\u8868\u3067\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1aFEASIBILITY STUDY OF ENDOSCOPIC SUBMUCOSAL DISSECTION USING FLEXIBLE THREE DIMENSION-ENDOSCOPE FOR EARLY GASTROINTESTINAL CANCER<br \/>\n\u8457\u8005\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Kensuke Shinmura, Tomonori Yano, Yoichi Yamamoto, Yusuke Yoda, Keisuke Hori, Hiroaki Ikematsu<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Poster session<br \/>\nAbstract \uff1a<br \/>\nBackground and aims<br \/>\nThe three-dimension (3D) endoscopic system is widely used as rigid endoscope in laparoscopic surgery. The advantage of 3D imaging is to obtain depth perception and spatial orientation. 3D system is reported that it can reduce operation time and technical error, especially in the procedure operated by trainees, comparing with conventional two-dimension system in laparoscopic surgery. However, little is known about the utility of 3D imaging system in the luminal endoscopic procedure. Recently, the prototype of 3D flexible endoscope (GIF-Y0080; Olympus Corporation, Tokyo, Japan) have been developed, and the aim of this study is to investigate the safety of endoscopic submucosal dissection using 3D endoscope (3D-ESD) for early gastrointestinal cancer (EGIC).<br \/>\nMethods<br \/>\nThis is a single center, prospective study to evaluate the safety of 3D-ESD for EGIC. From May 2018 to November 2018, patients who had planned 3D-ESD were prospectively enrolled in National Cancer Center Hospital East. The primary endpoint is the incidence rate of adverse events such as the delayed bleeding and the perforation. This study was reviewed and approved by the Institutional Review Board of our hospital.<br \/>\nResults<br \/>\n3 cases of esophageal cancer and 26 cases of gastric cancer were enrolled and analyzed. The median of procedure time in the whole 3D-ESD was 49 min (22-210). The margin-free en bloc resection rate using only 3D scope and curative resection rate was 93.1% (27 cases) and 86.2% (25cases). The incidence rate of the delayed bleeding and the delayed perforation in 3D-ESD for gastric cancer was 3.4% (1 case) and 3.4% (1 case), respectively. In one case, 3D endoscope was changed to conventional two-dimension endoscope because of the location of gastric cancer and the difficulty of manipulation, which is caused by the flexibility of the scope, during ESD. There were no cases of bleeding required transfusion in 3D-ESD for EGIC.<br \/>\nConclusion<br \/>\nThe prototype 3D endoscopic system was validated the safety in ESD for EGIC. Further evaluation is necessary to clarify the utility or efficacy of 3D system in the luminal endoscopic procedure.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u306f\uff0c3\u6b21\u5143\u5185\u8996\u93e1\u30b7\u30b9\u30c6\u30e0\u3092\u7528\u3044\u3066\u65e9\u671f\u306e\u80c3\u8178\u764c\u306b\u304a\u3051\u308bESD\u306e\u6709\u7528\u6027\u3092\u691c\u8a0e\u3059\u308b\u3082\u306e\u3067\u3057\u305f\uff0e\u5b9f\u969b\u306b\u672c\u30b7\u30b9\u30c6\u30e0\u3092\u7528\u3044\u305fESD\u306e3D\u6620\u50cf\u3067\u898b\u305b\u3066\u9802\u304d\u307e\u3057\u305f\uff0e\u5965\u884c\u304d\u304c\u308f\u304b\u308b\u3068\u3044\u3046\u30dd\u30a4\u30f3\u30c8\u306b\u304a\u3044\u3066\u306f\u8840\u7ba1\u306e\u4f4d\u7f6e\u306a\u3069\u304c2\u6b21\u5143\u306b\u6bd4\u3079\u3066\u8a8d\u8b58\u3057\u3084\u3059\u304f\uff0c\u7d20\u4eba\u306e\u79c1\u3067\u3082\u9a5a\u304d\u307e\u3057\u305f\uff0e\u4eca\u5f8c3D\u306b\u3088\u308b\u8a3a\u65ad\u652f\u63f4\u306e\u53ef\u80fd\u6027\u3092\u611f\u3058\u307e\u3057\u305f\uff0e\u672c\u30b7\u30b9\u30c6\u30e0\u306e\u5b9a\u91cf\u7684\u8a55\u4fa1\u304c\u4eca\u5f8c\u306e\u8ab2\u984c\u3068\u304a\u3063\u3057\u3083\u3063\u3066\u304a\u308a\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1aARTIFICIAL INTELLIGENCE USING CONVOLUTIONAL NEURAL NETWORK SHOWS HIGH DIAGNOSTIC PERFORMANCE OF MICROVESSELS ON SUPERFICIAL ESOPHAGEAL SQUAMOUS CELL CARCINOMA SIMILAR TO EXPERTS<br \/>\n\u8457\u8005\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Ryotaro Uema<strong>,<\/strong> Yoshito Hayashi<strong>,<\/strong> Minoru Kato<strong>,<\/strong> Keiichi Kimura<strong>,<\/strong> Takanori Inoue<strong>,<\/strong> Akihiko Sakatani<strong>,<\/strong> Shunsuke Yoshii<strong>,<\/strong> Yoshiki Tsujii<strong>,<\/strong> Shinichiro Shinzaki<strong>,<\/strong> Hideki Iijima<strong>,<\/strong> Tetsuo Takehara<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Poster session<br \/>\nAbstract \uff1a<br \/>\nIntroduction: The morphological diagnosis of microvessels on the surface of superficial esophageal squamous cell carcinoma <strong>(<\/strong>SESCC<strong>)<\/strong> with magnifying endoscopy using narrowband imaging <strong>(<\/strong>NBI-ME<strong>)<\/strong> is widely used in clinical practice. In particular<strong>,<\/strong> the presence of abnormal microvessels without a loop-like formation is very important<strong>,<\/strong> because it represents that the tumor invades muscularis mucosae <strong>(<\/strong>MM<strong>)<\/strong> or deeper<strong>,<\/strong> which is associated with lymph node metastasis. We constructed a convolutional neural network <strong>(<\/strong>CNN<strong>)<\/strong> system which diagnosed microvessels of SESCCs<strong>,<\/strong> and we evaluated the diagnostic performance of the system.<br \/>\nMethod: Endoscopic images of 261 SESCC lesions which underwent magnified endoscopy in our hospital from Jan 2013 to Dec 2017 were retrospectively collected. Then<strong>,<\/strong> 1803 images were cropped from the NBI-ME images into the size of 500\u00d7500 pixels or 430\u00d7430 pixels<strong>,<\/strong> and they were classified into 2 classes based on the Japan esophagus society <strong>(<\/strong>JES<strong>)<\/strong> classification<strong>,<\/strong> &#8216;TypeB1&#8217; <strong>(<\/strong>which is abnormal microvessels with a loop-like formation<strong>)<\/strong> or &#8216;TypeB2\/B3&#8217; <strong>(<\/strong>which are abnormal microvessel without a loop-like formation<strong>),<\/strong> by four expert endoscopists. Images whose diagnoses are matched in three out of four experts or more<strong>,<\/strong> were used for the CNN training. The diagnostic system was constructed by fine-tuning of pre-trained VGG19 model using optimizer Adam. The independent test set of 215 images collected from 34 SESCC lesions from Jan 2018 to June 2018 was used to compare the diagnosis of the trained CNN model<strong>,<\/strong> four experts<strong>,<\/strong> and six trainees. Of the 215 test images<strong>,<\/strong> 173 images from the 27 lesions which underwent endoscopic resection were classified into two groups<strong>,<\/strong> &#8216;EP-LPM&#8217; and &#8216;MM-&#8216;<strong>,<\/strong> by verifying with the mapping images of the specimen. These 173 images were used to evaluate the diagnostic accuracy of the depth of invasion.<br \/>\nResult: In the test set<strong>,<\/strong> the average kappa statistic of the microvessel diagnosis between the 4 experts was 0.78 and the average kappa statistic between the CNN model and the 4 experts was 0.76. The average kappa statistics between each trainee <strong>(<\/strong>Trainee 1<strong>,<\/strong> 2<strong>,<\/strong> 3<strong>,<\/strong> 4<strong>,<\/strong> 5<strong>,<\/strong> and 6<strong>)<\/strong> and the 4 experts were 0.79<strong>,<\/strong> 0.73<strong>,<\/strong> 0.68<strong>,<\/strong> 0.64<strong>,<\/strong> 0.52<strong>,<\/strong> and 0.51<strong>,<\/strong> respectively. The diagnostic time was 5.9s in the CNN model<strong>,<\/strong> 15m50s on average in the experts<strong>,<\/strong> and 23m11s on average in the trainees. In the 173 images whose depth of invasion were already known<strong>,<\/strong> the diagnostic accuracy of the depth of invasion of the CNN model was 85.0%<strong>,<\/strong> and the average diagnostic accuracy of experts and trainees were 84.5% and 72.2%<strong>,<\/strong> respectively. The area under the receiver operating curve for the diagnosis of invasion depth of the CNN model was 0.953.<br \/>\nConclusion: The constructed CNN system had a diagnostic performance equivalent to experts with a shorter time. This system would improve diagnoses of inexperienced endoscopists.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u306f\uff0c\u98df\u9053\u6241\u5e73\u4e0a\u76ae\u764c\u306eNBI\u62e1\u5927\u753b\u50cf\u304b\u3089CNN\u3092\u7528\u3044\u3066\u5fae\u5c0f\u8840\u7ba1\u306e\u5f62\u614b\u5b66\u7684\u8a3a\u65ad\u3092\u81ea\u52d5\u5316\u3059\u308b\u7814\u7a76\u3067\u3057\u305f\uff0e\u30e2\u30c7\u30eb\u306e\u8a3a\u65ad\u7cbe\u5ea6\u304c85%\u3092\u9054\u6210\u3057\uff0c\u719f\u7df4\u533b\u3068\u540c\u7b49\u306e\u7cbe\u5ea6\u304c\u3042\u308b\u3068\u3044\u3046\u5831\u544a\u3067\u3057\u305f\uff0eNBI\u62e1\u5927\u3092\u4f7f\u7528\u3057\u3066\u7814\u7a76\u3057\u3066\u3044\u308b\u6211\u3005\u3068\u3057\u3066\u3082\uff0c\u5fae\u5c0f\u8840\u7ba1\u306e\u5f62\u72b6\u304c\u8b58\u5225\u3067\u304d\u308b\u3088\u3046\u306b\u306a\u308c\u3070\uff0c\u764c\u90e8\u5206\u306e\u691c\u51fa\u306b\u5f79\u7acb\u3064\u305f\u3081\uff0c\u53c2\u8003\u306b\u306a\u308b\u7814\u7a76\u3067\u3057\u305f\uff0e\u5206\u985e\u3057\u3066\u3044\u308b\u5b9a\u91cf\u7684\u306a\u6307\u6a19\u3092\u7406\u89e3\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u308c\u3070\u5c1a\u5b09\u3057\u304b\u3063\u305f\u3067\u3059\uff0e<br \/>\n&nbsp;<br \/>\n\u53c2\u8003\u6587\u732e<br \/>\n<a href=\"https:\/\/ddw.org\/attendee-planning\/online-planner\">https:\/\/ddw.org\/attendee-planning\/online-planner<\/a><br \/>\n&nbsp;<br \/>\n<strong>\u5b66\u4f1a\u53c2\u52a0\u5831\u544a\u66f8<\/strong><\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"147\"><strong>\u00a0<\/strong><br \/>\n<strong>\u5831\u544a\u8005\u6c0f\u540d<\/strong><\/td>\n<td width=\"373\">&nbsp;<br \/>\n\u6e05\u91ce\u5141\u8cb4<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u767a\u8868\u8ad6\u6587\u30bf\u30a4\u30c8\u30eb<\/strong><\/td>\n<td width=\"373\">LCI\u753b\u50cf\u3092\u7528\u3044\u305f\u6a5f\u68b0\u5b66\u7fd2\u306b\u57fa\u3065\u304fHelicobacter Pylori \u611f\u67d3\u81ea\u52d5\u8a3a\u65ad\u30b7\u30b9\u30c6\u30e0-\u9664\u83cc\u5f8c\u753b\u50cf\u306e\u691c\u8a0e-<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u767a\u8868\u8ad6\u6587\u82f1\u30bf\u30a4\u30c8\u30eb<\/strong><\/td>\n<td width=\"373\">MACHINE-LEARNING-BASEDAUTOMATIC DIAGNOSTIC SYSTEM USING LINKED COLOR IMAGING FOR HELICOBACTER PYLORI<br \/>\nINFECTION: EXAMINATION OF IMAGE AFTER ERADICATION<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u8457\u8005<\/strong><\/td>\n<td width=\"373\">\u6e05\u91ce\u5141\u8cb4\uff0c\u5b89\u7530\u525b\u58eb\uff0c\u5e02\u5ddd\u5bdb\uff0c\u65e5\u548c\u609f\uff0c\u516b\u6728\u4fe1\u660e\uff0c\u5ee3\u5b89\u77e5\u4e4b<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u4e3b\u50ac<\/strong><\/td>\n<td width=\"373\">Digestive Disease Week<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u8b1b\u6f14\u4f1a\u540d<\/strong><\/td>\n<td width=\"373\">DDW2019<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u4f1a\u5834<\/strong><\/td>\n<td width=\"373\">San Diego Convention Center<\/td>\n<\/tr>\n<tr>\n<td width=\"147\"><strong>\u958b\u50ac\u65e5\u7a0b<\/strong><\/td>\n<td width=\"373\">2019\/05\/18-2019\/05\/21<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>&nbsp;<br \/>\n&nbsp;<\/p>\n<ol>\n<li>\u8b1b\u6f14\u4f1a\u306e\u8a73\u7d30<\/li>\n<\/ol>\n<p>2019\/05\/18\u304b\u30892019\/05\/21\u306b\u304b\u3051\u3066\uff0cSan Diego Convention Center\u306b\u3066\u958b\u50ac\u3055\u308c\u307e\u3057\u305fDigestive Disease Week 2019\uff08DDW2019\uff09\u306b\u53c2\u52a0\u3044\u305f\u3057\u307e\u3057\u305f\uff0e\u3053\u306eDDW\u306f\u80c3\u8178\u75c5\u5b66\uff0c\u809d\u81d3\u75c5\u5b66\uff0c\u5185\u8996\u93e1\u691c\u67fb\uff0c\u80c3\u8178\u5916\u79d1\u306e\u5206\u91ce\u306b\u304a\u3051\u308b\u533b\u5e2b\uff0c\u7814\u7a76\u8005\u304a\u3088\u3073\u305d\u306e\u696d\u754c\u306e\u6700\u5927\u7d44\u7e54\u3067\u3042\u308a\uff0c\u4e00\u6d41\u306e\u7814\u7a76\u306b\u5bfe\u3059\u308b\u30a2\u30d7\u30ed\u30fc\u30c1\u3092\u5b66\u3073\u305d\u306e\u5206\u91ce\u306e\u5f15\u5c0e\u8005\u3068\u30b3\u30df\u30e5\u30cb\u30b1\u30fc\u30b7\u30e7\u30f3\u3092\u56f3\u308b\u3053\u3068\u3067\uff0c\u81ea\u5206\u306e\u7814\u7a76\u306b\u5bfe\u3059\u308b\u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u3092\u5f97\u308b\u3053\u3068\u3092\u76ee\u7684\u306b\u958b\u50ac\u3055\u308c\u3066\u3044\u307e\u3059\uff0e\u672c\u7814\u7a76\u5ba4\u304b\u3089\u306f\u4ed6\u306b\uff0c\u5ee3\u5b89\u5148\u751f\uff0cM2\u306e\u5b66\u751f\u3068\u3057\u3066\u5965\u6751\uff08\u99ff\uff09\u3055\u3093\u304c\u53c2\u52a0\u3057\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<ol start=\"2\">\n<li>\u7814\u7a76\u767a\u8868\n<ul>\n<li>\u767a\u8868\u6982\u8981<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n<p>\u79c1\u306f20\u65e5\u306e\u30dd\u30b9\u30bf\u30fc\u30bb\u30c3\u30b7\u30e7\u30f3\u3067\u767a\u8868\u3057\u307e\u3057\u305f\uff0e\u767a\u8868\u306f\u30dd\u30b9\u30bf\u30fc\u5f62\u5f0f\u3067\uff0c\u8a082\u6642\u9593\u53c2\u52a0\u8005\u306e\u65b9\u3068\u8b70\u8ad6\u3092\u884c\u3044\u307e\u3057\u305f\uff0e<br \/>\n\u4eca\u56de\u306f\uff0cMACHINE-LEARNING-BASED AUTOMATIC DIAGNOSTIC SYSTEM USING LINKED COLOR IMAGING FOR HELICOBACTER PYLORI INFECTION: EXAMINATION OF IMAGE AFTER ERADICATION\u306b\u3064\u3044\u3066\u767a\u8868\u3057\u307e\u3057\u305f\uff0e\u5185\u5bb9\u306f\uff0c\u30d4\u30ed\u30ea\u83cc\u3092\u9664\u83cc\u3057\u305f\u75c7\u4f8b\u306b\u898b\u3089\u308c\u308b\u5730\u56f3\u72b6\u767a\u8d64\u306e\u7279\u5fb4\u3092\u5b9a\u91cf\u5316\u3057\u305f\u3053\u3068\u3067\uff0c\u5f93\u6765\u306e\u30d4\u30ed\u30ea\u83cc\u611f\u67d3\u8a3a\u65ad\u30b7\u30b9\u30c6\u30e0\u306e\u7cbe\u5ea6\u304c\u5411\u4e0a\u3057\u305f\u3068\u3044\u3046\u3053\u3068\u3067\u3057\u305f\uff0e\u4ee5\u4e0b\u306b\u6284\u9332\u3092\u8a18\u8f09\u81f4\u3057\u307e\u3059\uff0e<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">Introduction<br \/>\nAs a part of our research, we have developed a system for automatically diagnosing the presence or absence of H.pylori (Hp) infection, from the gastric mucosa image obtained by linked color imaging (LCI), using machine learning. This system aids a doctor\u2019s diagnosis. In this study, an experiment involving Hp eradication cases was formulated,and the results emerging from it have been documented. Irrespective of whether eradication has been carried out or not, it is difficult for medical doctors to diagnose whether a patient has been eradicated of Hp. However, if it is possible to diagnose eradication success only by endoscopic diagnosis without performing additional examination, the burden on the patient can be reduced. In addition to formulating the experiment, we have developed a system to detect the success or failure of Hp eradication.<br \/>\nAims and methods<br \/>\nThe characteristic of a gastric mucosa image, representing Hp eradication, is to have a map-like redness. In the proposed system, we quantify this map-like redness for images of gastric mucosa obtained from LCI, and improve the accuracy of diagnosis of Hp positive or negative (post eradication). By using LCI, the map-like redness is observed as lavender color, while background gastric mucosa is observed as apricot color. Figure 1 shows an image with map-like redness.<br \/>\nFirst, a region on the image having a high hue value indicating a lavender color is extracted as a region of interest (ROI). Second, the center of gravity of the ROI is identified on the image. Thereafter, a circle, of radius equivalent to the Euclidean distance of the outermost pixel of the ROI from the center of gravity, is depicted as a circle of interest (COI). Finally, if an image having a high ROI ratio for all pixels and a large hue variance value in the COI is observed, it is identified as an image having map-like redness. Cases where map-like redness is detected in the gastric mucosa image, are considered as sterile and Hp negative. Figure 2 shows a schematic diagram of the conventional system and the proposed method. In this study, 200 images (40 cases; 32 cases are Hp positive and 8 cases are after eradication) of endoscopic examination (LCI observation) at Asahi University Hospital were used to evaluate the system.<br \/>\nResult<br \/>\nIn the conventional system, 29 of the 40 cases were correctly diagnosed. In comparison, by using the proposed system, 37 of the 40 cases were correctly diagnosed. These results show that quantification of the map-like redness, which is characteristic of Hp eradication, leads to an improvement in the accuracy of the system.<br \/>\nConclusion<br \/>\nBy using the proposed system, the presence or absence of Hp infection can be automatically diagnosed with the same precision as expert doctors. And, this system can support the diagnosis of inexperienced doctors.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>&nbsp;<\/p>\n<ul>\n<li>\u8cea\u7591\u5fdc\u7b54<\/li>\n<\/ul>\n<p>\u4eca\u56de\u306e\u8b1b\u6f14\u767a\u8868\u3067\u306f\uff0c\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u8cea\u7591\u3092\u53d7\u3051\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>1<\/strong><br \/>\n\u8cea\u554f\u306f\uff0c\u30b7\u30b9\u30c6\u30e0\u306e\u3069\u3053\u306e\u51e6\u7406\u3067\u6a5f\u68b0\u5b66\u7fd2\u3092\u7528\u3044\u3066\u3044\u308b\u306e\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u8272\u76f8\u306b\u3088\u308b\u753b\u50cf\u306e2\u5206\u985e\u3092\u3057\u305f\u5f8c\u306b\uff0e\u4f4e\u8272\u76f8\u753b\u50cf\u306b\u5bfe\u3057\u3066\u6a5f\u68b0\u5b66\u7fd2\u3092\u7528\u3044\u3066\u8b58\u5225\u3057\u3066\u3044\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>2<\/strong><br \/>\n\u8cea\u554f\u306f\uff0cdeep learning\u3068machine learning\u306e\u9055\u3044\u306f\u4f55\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0cdeep learning\u306f\u4f55\u4e07\u3082\u306e\u30c7\u30fc\u30bf\u304c\u5fc5\u8981\u3067\u3042\u308a\uff0c\u8b58\u5225\u306b\u7528\u3044\u308b\u7279\u5fb4\u91cf\u3092\u81ea\u52d5\u5b9a\u7fa9\u3059\u308b\uff0e\u4e00\u65b9\u3067\uff0cmachine learning\u306f\u5c11\u306a\u3044\u30c7\u30fc\u30bf\u3067\u30e2\u30c7\u30eb\u3092\u4f5c\u6210\u3067\u304d\uff0c\u8b58\u5225\u306b\u7528\u3044\u308b\u7279\u5fb4\u91cf\u3092\u81ea\u5206\u3067\u8a2d\u8a08\u3059\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>3<\/strong><br \/>\n\u8cea\u554f\u306f\uff0c\u30b7\u30b9\u30c6\u30e0\u306e\u30b9\u30c6\u30c3\u30d7\u30a2\u30c3\u30d7\u3068\u3057\u3066\u4eca\u5f8c\u4f55\u3092\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u306e\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0cSVM\u306e\u967d\u6027\u753b\u50cf\u306b\u5bfe\u3059\u308b\u8b58\u5225\u7cbe\u5ea6\u304c\u4f4e\u3044\u3053\u3068\u304c\u8ab2\u984c\u3067\u3042\u308b\uff0e\u305d\u3053\u3067\uff0c\u5b66\u7fd2\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u518d\u8003\u304c\u5fc5\u8981\u3067\u3042\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>4<\/strong><br \/>\n\u8cea\u554f\u306f\uff0c\u30b7\u30b9\u30c6\u30e0\u304c\u30d4\u30ed\u30ea\u83cc\u3092\u9664\u83cc\u3057\u305f\u75c7\u4f8b\u306b\u5bfe\u5fdc\u3067\u304d\u308b\u3088\u3046\u306b\u306a\u3063\u305f\u306e\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u5730\u56f3\u72b6\u767a\u8d64\u3092\u6709\u3057\u3066\u3044\u308b\u75c7\u4f8b\u306b\u306f\u5bfe\u5fdc\u3067\u304d\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n<strong>\u00a0<\/strong><br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>5<\/strong><br \/>\n\u8cea\u554f\u306f\uff0c\u73fe\u72b6\u306e\u30b7\u30b9\u30c6\u30e0\u306e\u8a3a\u65ad\u7d50\u679c\u306f\u3069\u3046\u306a\u306e\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u5171\u540c\u7814\u7a76\u5148\u306eLCI\u3092\u7528\u3044\u3066\u30d4\u30ed\u30ea\u83cc\u611f\u67d3\u306e\u8a3a\u65ad\u3092\u3057\u3066\u3044\u308b\u5148\u751f\u3068\uff0e\u905c\u8272\u306e\u306a\u3044\u7d50\u679c\u3092\u5f97\u3066\u3044\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>6<\/strong><br \/>\n\u8cea\u554f\u306f\u8272\u60c5\u5831\u3060\u3051\u3067\u8a3a\u65ad\u3067\u304d\u306a\u3044\u753b\u50cf\u306f\u5b58\u5728\u3059\u308b\u306e\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u8272\u306e\u307f\u3067\u8a3a\u65ad\u3067\u304d\u306a\u3044\u753b\u50cf\u306f\u3042\u308a\u307e\u3059\uff0e\u7dda\u72b6\u767a\u8d64\u306a\u3069\u306f\u767a\u8d64\u90e8\u5206\u304c\u8d64\u304f\u306a\u308b\u305f\u3081\uff0c\u507d\u967d\u6027\u3068\u306a\u308b\u3053\u3068\u304c\u3042\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n<strong>\u30fb\u8cea\u554f\u5185\u5bb9<\/strong><strong>7<\/strong><br \/>\n\u8cea\u554f\u306f\uff0c\u65b0\u305f\u306b\u30d4\u30ed\u30ea\u83cc\u611f\u67d3\u8a3a\u65ad\u30b7\u30b9\u30c6\u30e0\u3092\u69cb\u7bc9\u3057\u3066\uff0c\u305d\u308c\u305e\u308c\u306e\u30b7\u30b9\u30c6\u30e0\u306e\u8a3a\u65ad\u7d50\u679c\u3092\u591a\u6570\u6c7a\u7684\u306b\u5229\u7528\u3057\u3066\u6700\u7d42\u8a3a\u65ad\u3092\u884c\u3046\u306e\u306f\u3069\u3046\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u3059\u308b\u5fc5\u8981\u306f\u3042\u308b\u304b\u3082\u3057\u308c\u306a\u3044\u304c\uff0c\u591a\u6570\u6c7a\u3067\u533b\u5b66\u7684\u306a\u5224\u65ad\u3092\u884c\u3063\u3066\u3082\u826f\u3044\u306e\u304b\u8003\u3048\u308b\u3068\u3053\u308d\u304c\u3042\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n\u30fb\u8cea\u554f\u5185\u5bb98<br \/>\n\u8cea\u554f\u306f\u30b7\u30b9\u30c6\u30e0\u304c\u5b9f\u969b\u306b\u5c0e\u5165\u3055\u308c\u305f\u6642\uff0c\u304a\u91d1\u306f\u3069\u3046\u3059\u308b\u306e\u304b\u3068\u3044\u3046\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u8cea\u554f\u306b\u5bfe\u3057\u3066\uff0c\u304a\u91d1\u306e\u9762\u306f\u79c1\u304c\u8003\u3048\u308b\u3053\u3068\u3067\u306f\u306a\u3044\u304c\uff0c\u8003\u3048\u308b\u3068\u3059\u308b\u306a\u3089\u3070\uff0c\u9700\u8981\u306e\u9ad8\u3044\u6771\u5357\u30a2\u30b8\u30a2\u306a\u3069\u306e\u767a\u5c55\u9014\u4e0a\u56fd\u3067\u306e\u30b7\u30b9\u30c6\u30e0\u306e\u5c0e\u5165\u306f\uff0c\u91d1\u92ad\u9762\u3092\u691c\u8a0e\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u3068\u7b54\u3048\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<ul>\n<li>\u611f\u60f3<\/li>\n<\/ul>\n<p>\u30fb\u4eca\u56de\u306f\u521d\u306e\u5b66\u4f1a\u304b\u3064\u6d77\u5916\u3067\u3042\u3063\u305f\u306e\u3067\uff0c\u767a\u8868\u306e\u4ed5\u65b9\u3084\u82f1\u8a9e\u3067\u306e\u8aac\u660e\u306e\u96e3\u3057\u3055\u3092\u5b9f\u611f\u3057\u307e\u3057\u305f\uff0e<br \/>\n\u30fb\u79c1\u304c\u767a\u8868\u3057\u305f\uff0cAI\u30bb\u30c3\u30b7\u30e7\u30f3\u3067\u306f\u4ed6\u306e\u90e8\u9580\u306b\u6bd4\u3079\u3066\u8074\u8b1b\u8005\u304c\u591a\u304f\uff0c\u533b\u5b66\u306e\u4e16\u754c\u306b\u304a\u3044\u3066AI\u304c\u6ce8\u76ee\u3055\u308c\u3066\u3044\u308b\u306e\u3060\u3068\u601d\u3044\u307e\u3057\u305f\uff0e<br \/>\n\u30fb\u4ed6\u306eAI\u3092\u7528\u3044\u305f\u7814\u7a76\u3067\u306f\uff0cDeep learning\u3092\u7528\u3044\u305f\u7814\u7a76\u304c\u591a\u304f\uff0c\u307e\u305f\uff0c\u8cea\u554f\u3067\u3082Deep learning\u306e\u8a71\u984c\u3092\u51fa\u3059\u4eba\u304c\u591a\u304b\u3063\u305f\u5370\u8c61\u3067\u3057\u305f\uff0e<br \/>\n\u30fb\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0\u3067\uff0c\u30ac\u30f3\u3084\u30dd\u30ea\u30fc\u30d7\u306e\u691c\u51fa\u3092\u3057\u3066\u3044\u308b\u7814\u7a76\u304c\u591a\u304f\uff0c\u79c1\u306e\u7814\u7a76\u3082\u305d\u306e\u30ec\u30d9\u30eb\u307e\u3067\u5230\u9054\u3057\u305f\u3044\u306a\u3068\u601d\u3044\u307e\u3057\u305f\uff0e<br \/>\n\u30fb\u4ed6\u306e\u65b9\u306e\u30dd\u30b9\u30bf\u30fc\u3092\u62dd\u898b\u3059\u308b\u306b\u3042\u305f\u308a\uff0c\u73fe\u5728\u306e\u8ab2\u984c\u3067\u3042\u308b\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u66f8\u304d\u65b9\u304c\u53c2\u8003\u306b\u306a\u308a\u307e\u3057\u305f\uff0e<br \/>\n\u30fb\u30b7\u30b9\u30c6\u30e0\u306e\u51e6\u7406\u3068\u305d\u306e\u624b\u6cd5\u306e\u90e8\u5206\u3067\u540c\u3058\u8cea\u554f\u3092\u4f55\u5ea6\u304b\u3055\u308c\u305f\u306e\u3067\uff0c\u30b7\u30b9\u30c6\u30e0\u306e\u6982\u8981\u56f3\u306e\u66f8\u304d\u65b9\u3092\u518d\u8003\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u3068\u611f\u3058\u307e\u3057\u305f\uff0e<br \/>\n\u30fb\u82f1\u8a9e\u306e\u8cea\u554f\u306b\u5bfe\u5fdc\u3067\u304d\u305a\uff0c\u5e02\u5ddd\u5148\u751f\u3084\u5ee3\u5b89\u5148\u751f\u306e\u52a9\u3051\u3092\u501f\u308a\u308b\u5834\u9762\u304c\u591a\u3005\u3042\u3063\u305f\u306e\u3067\uff0c\u6b21\u56de\u306e\u56fd\u969b\u5b66\u4f1a\u307e\u3067\u306b\u82f1\u8a9e\u529b\u3092\u9ad8\u3081\u3066\u304a\u304f\u5fc5\u8981\u304c\u3042\u308b\u3068\u601d\u3044\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<ol start=\"3\">\n<li>\u8074\u8b1b<\/li>\n<\/ol>\n<p>\u4eca\u56de\u306e\u8b1b\u6f14\u4f1a\u3067\u306f\uff0c\u4e0b\u8a18\u306e4\u4ef6\u306e\u767a\u8868\u3092\u8074\u8b1b\u3057\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a\u3000AUTOMATIC DETECTION OF GASTRIC CANCER USING SEMANTIC SEGMENTATION BASED ON ARTIFICIAL INTELLIGENCE<br \/>\n\u8457\u8005\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Tomoyuki Shibata, Kazuma Enomoto, Atsushi Teramoto, Hyuga Yamada1, Naoki Ohmiya, Hiroshi Fujita<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Poster session<br \/>\nAbstruct \uff1a Objectives: Gastric cancer carries still a high incidence rate worldwide, and early detection and early treatment are important. However, screening endoscopic examination needs accurate diagnostic skills for individual endoscopist. Therefore, we used deep learning to develop a diagnosis support system for automated detection of gastric cancer from endoscopic images using semantic segmentation technology. The accuracy of detection and extraction of the gastric cancer region were evaluated.<br \/>\nMethods: White light endoscopic images were used for this study. These images were taken during endoscopic examination for gastric cancer evaluation at Fujita Health University Hospital. A fully convolutional network (FCN) with 7 convolution layers, 3 pooling layers, and a deconvolution layer was used for the deep learning model used in this study. The FCN can output a segmented image from an input image of arbitrary size. We resized the endoscopic images to 256 \u00d7 256 pixel and trimmed within the inscribed circle. The infiltrating area of gastric cancer was surrounded by expert endoscopists with lines using paint tool. Subsequently, in order to prevent overfitting owing to the limited number of images, training data were augmented by inversion and rotation processing. We used 42 normal cases (1149 images) and 98 cases (553 images) of gastric cancer. The FCN was trained using 80 cases and the extraction accuracy of gastric cancer region and detection rate were evaluated using 18 cases. If the gastric cancer area by endoscopists and the gastric cancer detection area of FCN were overlapped, the answer was regarded as correct.<br \/>\nResults: The accuracy rate was 96% and positive predictive value (PPV) was 80.6%. Additionally, the number of false positives per image was 0.03.<br \/>\nConclusions: This method showed favorable results and confirmed that the proposed method may be useful to detect gastric cancer in endoscopic images. By using this method, it is possible to mitigate overlook during examination.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u306f\uff0c Deep learning\u3092\u7528\u3044\u3066\u80c3\u304c\u3093\u3092\u81ea\u52d5\u691c\u51fa\u3059\u308b\u30b7\u30b9\u30c6\u30e0\u306e\u958b\u767a\u3067\u3057\u305f\uff0e127800\u679a\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u306f\uff0c25\u679a\u306e\u901a\u5e38\u753b\u50cf\uff0c79\u679a\u306e\u764c\u75c5\u5909\u304c\u542b\u307e\u308c\u305f\u753b\u50cf\u3092\u53cd\u8ee2\u304a\u3088\u3073\u56de\u8ee2\u51e6\u7406\u3059\u308b\u3053\u3068\u3067\u5f97\u3066\u3044\u307e\u3057\u305f\uff0e\u8a3a\u65ad\u7d50\u679c\u306f\uff0cSensitivity\u304c97.6%\uff0cSpecificity\u304c94.8%\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u767a\u8868\u3092\u805e\u3044\u3066\uff0c\u5143\u753b\u50cf\u304c\u764c\u75c5\u5909\u306e\u30d1\u30bf\u30fc\u30f3\u3092\u3069\u308c\u3060\u3051\u7db2\u7f85\u3067\u304d\u3066\u3044\u308b\u306e\u304b\u306f\u5b9a\u304b\u3067\u306f\u306a\u3044\u3067\u3059\u304c\uff0c\u753b\u50cf\u306e\u53cd\u8ee2\u3084\u56de\u8ee2\u3092\u3059\u308b\u3053\u3068\u3067\u5f97\u305f\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3067\uff0c\u3053\u306e\u7cbe\u5ea6\u304c\u51fa\u305f\u3053\u3068\u306b\u9a5a\u304d\u307e\u3057\u305f\uff0e\u3057\u304b\u3057\uff0c Deep learning\u304c\u9762\u767d\u3044\u306e\u306f\u7d50\u679c\u306e\u307f\u3067\uff0c\u305d\u308c\u4ee5\u4e0a\u306e\u8b70\u8ad6\u306b\u767a\u5c55\u3057\u306a\u3044\u306a\u3068\u611f\u3058\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"529\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1aBLI AND LCI IMPROVE POLYP DETECTION AND DELINEATION ACCURACY FOR DEEP LEARNING NETWORKS<br \/>\n\u8457\u8005\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Tom Eelbode, Cesare Hassan, Ingrid Demedts, Philip Roelandt, Emmanuel Coron, Pradeep Bhandari, Helmut Neumann, Oliver Pech, Alessandro Repici, Frederik Maes, Raf Bisschops<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Poster session<br \/>\nAbstruct \uff1a Introduction and aim Current state-of-the-art automated polyp detection and delineation techniques use white light imaging as their base modality. Studies have however suggested that polyp detection rates can be improved by using other modalities such as linked color imaging (LCI) from Fujifilm. This might be true for human observers, but it has not yet been investigated how an artificial intelligence (AI) system is influenced by the choice of modality. The aim of this research is to investigate the influence of the modality (WLI, blue light imaging or BLI and LCI) on the performance of an AI system for polyp detection and delineation.<br \/>\nMethods Complete pull-through colonoscopy videos from 120 patients are included with a total of 280 polyps for training, validation and testing of the system (n = 176, 27, 77 respectively with no overlapping patients). Shorter video clips containing the first apparition of each polyp are extracted and for each clip, only a few frames are annotated by three individual experts. These 758 single-frame manual annotations are automatically propagated over the entire clip. The resulting, much larger annotated dataset of 40887 images is then used to train a recurrent convolutional neural network (CNN) for polyp detection and delineation.<br \/>\nFrame-level sensitivity and specificity are reported for evaluation of the detection power of the network. For delineation accuracy, the Dice score is used which is a measure for the amount of overlap between a delineation map and its ground truth.<br \/>\nResults Table 1 shows that BLI significantly improves sensitivity, specificity and Dice score. Similarly, LCI increases detection performance to a lesser extent, however the LCI Dice score for delineation accuracy decreases significantly in comparison to WLI. Pairwise t-tests show that all differences are significant with a p value &lt;0,00001 (significance level of 0,05).<br \/>\nConclusion The choice of modality has a significant impact on the detection and delineation performance of an AI system. We show that our network performs best for both tasks on BLI and that LCI has a superior detection, but inferior delineation power compared to WLI.<br \/>\nSample size, sensitivity, specificity and Dice score (mean and stdev) for the three different modalities in the test set.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u306f\uff0c\u5927\u8178\u30dd\u30ea\u30fc\u30d7\u306e\u691c\u51fa\u304a\u3088\u3073\u63cf\u5199\u306b\u304a\u3044\u3066WLI\uff0cBLI\uff0cLCI\u3067\u64ae\u5f71\u3055\u308c\u305f\u753b\u50cf\u3092Deep Learning\u3067\u305d\u308c\u305e\u308c\u5b66\u7fd2\u304a\u3088\u3073\u30c6\u30b9\u30c8\u3057\u305f\u7d50\u679c\u3092\u6bd4\u8f03\u3057\u3066\u3044\u307e\u3057\u305f\uff0e\u7d50\u679c\u3068\u3057\u3066\uff0cWLI\uff0cBLI\uff0cLCI\u306e\u305d\u308c\u305e\u308c\u306e\u611f\u5ea6\u306f81%\uff0c92%\uff0c85%\uff0c\u7279\u7570\u5ea6\u306f76%\uff0c85%\uff0c82%\u3067\u3057\u305f\uff0e<br \/>\n\u6587\u732e\u306a\u3069\u3092\u898b\u3066\u3044\u3066\u3082\uff0cWLI\u3068LCI\u306e\u6bd4\u8f03\u306f\u3088\u304f\u898b\u304b\u3051\u307e\u3059\u304c\uff0cBLI\u3068LCI\u306e\u6bd4\u8f03\u306f\u521d\u3081\u3066\u898b\u305f\u306e\u3067\u8208\u5473\u6df1\u304b\u3063\u305f\u3067\u3059\uff0e<br \/>\nDeep Learning\u3092\u7528\u3044\u305f\u30b7\u30b9\u30c6\u30e0\u3067\u3042\u308b\u306e\u3067\uff0cBLI\u306e\u691c\u51fa\u7cbe\u5ea6\u304c\u826f\u304b\u3063\u305f\u539f\u56e0\u304c\u4e0d\u660e\u306a\u90e8\u5206\u304c\u6b8b\u5ff5\u3060\u3068\u601d\u3044\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"539\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1aARTIFICIAL INTELLIGENCE-ASSISTED ENDOSCOPY IN CHARACTERIZATION OF GASTRIC LESIONS USING MAGNIFYING NARROW BAND IMAGING ENDOSCOPY<br \/>\n\u8457\u8005\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Sergey V. Kashin, Roman Kuvaev, Ekaterina Albertovna Kraynova, Olga Dunaeva, Alexander Rusakov, Evgeny Nikonov<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 \uff1a Poster session<br \/>\nAbstruct \uff1aAims of this study were to develop and evaluate an artificial intelligence based system for histology prediction of gastric lesions using magnifying narrow band imaging (M-NBI) endoscopy.<br \/>\nWe selected and analyzed 265 endoscopy M-NBI images of gastric lesions from 128 patients who underwent upper M-NBI endoscopy (Olympus Exera GIF Q160Z, Lucera GIF Q260Z). All images were divided into four classes: (1) type A (n=46): non-neoplastic and non-metaplastic lesions with regular circular microsurface (MS) and regular microvascular (MV) patterns; (2) Type B (n=90): intestinal metaplasia with tubulo-villous MS and regular MV patterns; (3) Type C (n=74) neoplastic lesions with irregular MS or MV pattern; (4) artifacts (n=55). During automated classification quadrant areas were calculated on the image, geometrical and topological features were computed for every fragment. Using the greedy forward selection algorithm, the set of five most significant features were selected: three geometric features (the compactness of the MS pattern, the perimeter of the MS pattern, the average of area of the component of the MV pattern) two topological features (the kurtosis of the histogram of the 0-th persistence diagram of the image, the first norm of the 0-th persistence diagram of the signed distance function). Support vector machine (SVM) classifier was used for 4-class automated diagnosis.Training and testing were performed for every image by a k-fold method (k=10).<br \/>\nThe average percentage of correctly recognized areas was 91.4%. Classification precision (positive predictive value), recall (sensitivity), F-score for class A were 96.5 90.4 93.3 for class B were 93.7, 92.0, 92.9, respectively, for class C were 83.3, 91.3, 87.1, respectively, and for artifacts were 99.2, 91.7, 95.3, respectively<br \/>\nThe designed system based on the extraction of the geometrical and topological features from M-NBI image and analysis by SVM could provide effective recognition of three types of gastric mucosal changes.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u306f\uff0c\u62e1\u5927NBI\u3067\u64ae\u5f71\u3055\u308c\u305f\u80c3\u75c5\u5909\u753b\u50cf\u3092\u533b\u5b66\u7684\u5909\u5316\u304c\u8a8d\u3081\u3089\u308c\u308b\u75c5\u59093\u30af\u30e9\u30b9\uff0c\u3069\u306e\u30af\u30e9\u30b9\u306b\u3082\u5f53\u3066\u306f\u307e\u3089\u306a\u30441\u3064\u306e\u30af\u30e9\u30b9\u306b\u5206\u3051\u3066\u304a\u304d\uff0c5\u6b21\u5143\u306e\u69cb\u9020\u7279\u5fb4\u91cf\u3067\u5b66\u7fd2\u3057\u305fSVM\u3067\u30af\u30e9\u30b9\u5206\u985e\u3057\uff0c\u3069\u306e\u30af\u30e9\u30b9\u306b\u6709\u52b9\u306a\u7279\u5fb4\u91cf\u304c\u8a2d\u8a08\u3055\u308c\u3066\u3044\u308b\u304b\u3092\u793a\u3057\u305f\u3082\u306e\u3067\u3057\u305f\uff0e<br \/>\n\u3053\u306e\u7814\u7a76\u306f\uff0c\u540c\u5fd7\u793eNBI\u7814\u7a76\u3068\u975e\u5e38\u306b\u985e\u4f3c\u3057\u3066\u304a\u308a\uff0cNBI\u7814\u7a76\u306e\u73fe\u5728\u306e\u8ab2\u984c\u3067\u3042\u308b\u764c\u75c5\u5909\u306b\u6709\u52b9\u306a\u69cb\u9020\u7279\u5fb4\u91cf\u3084\uff0c\u6271\u3044\u65b9\u3092\u5b66\u3076\u3053\u3068\u304c\u3067\u304d\u307e\u3057\u305f\uff0e<br \/>\n\u81ea\u5206\u306e\u7814\u7a76\u3067\u306f\uff0c\u73fe\u5728\u8272\u7279\u5fb4\u91cf\u306e\u307f\u3067\u30b7\u30b9\u30c6\u30e0\u3092\u69cb\u7bc9\u3057\u3066\u3044\u308b\u304c\uff0c\u4eca\u5f8c\uff0c\u30b7\u30b9\u30c6\u30e0\u3092\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0\u3067\u7528\u3044\u308b\u969b\u306b\uff0c\u80c3\u306e\u90e8\u4f4d\u306a\u3069\u306e\u5224\u5225\u306b\u69cb\u9020\u7279\u5fb4\u91cf\u3092\u7528\u3044\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u306e\u3067\uff0c\u3053\u306e\u77e5\u8b58\u3092\u6d3b\u304b\u305b\u305f\u3089\u306a\u3068\u601d\u3044\u307e\u3059\uff0e<br \/>\n&nbsp;<\/p>\n<table>\n<tbody>\n<tr>\n<td width=\"539\">\u767a\u8868\u30bf\u30a4\u30c8\u30eb\uff1aFEASIBILITY STUDY OF ENDOSCOPIC SUBMUCOSAL DISSECTION USING FLEXIBLE THREE DIMENSION-ENDOSCOPE FOR EARLY GASTROINTESTINAL CANCER<br \/>\n\u8457\u8005\uff1a Kensuke Shinmura1, Tomonori Yano1, Yoichi Yamamoto1, Yusuke Yoda1, Keisuke Hori1, Hiroaki Ikematsu1<br \/>\n\u30bb\u30c3\u30b7\u30e7\u30f3\u540d\uff1a Poster session<br \/>\nAbstruct \uff1a<br \/>\nBackground and aims<br \/>\nThe three-dimension (3D) endoscopic system is widely used as rigid endoscope in laparoscopic surgery. The advantage of 3D imaging is to obtain depth perception and spatial orientation. 3D system is reported that it can reduce operation time and technical error, especially in the procedure operated by trainees, comparing with conventional two-dimension system in laparoscopic surgery. However, little is known about the utility of 3D imaging system in the luminal endoscopic procedure. Recently, the prototype of 3D flexible endoscope (GIF-Y0080; Olympus Corporation, Tokyo, Japan) have been developed, and the aim of this study is to investigate the safety of endoscopic submucosal dissection using 3D endoscope (3D-ESD) for early gastrointestinal cancer (EGIC).<br \/>\nMethods<br \/>\nThis is a single center, prospective study to evaluate the safety of 3D-ESD for EGIC. From May 2018 to November 2018, patients who had planned 3D-ESD were prospectively enrolled in National Cancer Center Hospital East. The primary endpoint is the incidence rate of adverse events such as the delayed bleeding and the perforation. This study was reviewed and approved by the Institutional Review Board of our hospital.<br \/>\nResults<br \/>\n3 cases of esophageal cancer and 26 cases of gastric cancer were enrolled and analyzed. The median of procedure time in the whole 3D-ESD was 49 min (22-210). The margin-free en bloc resection rate using only 3D scope and curative resection rate was 93.1% (27 cases) and 86.2% (25cases). The incidence rate of the delayed bleeding and the delayed perforation in 3D-ESD for gastric cancer was 3.4% (1 case) and 3.4% (1 case), respectively. In one case, 3D endoscope was changed to conventional two-dimension endoscope because of the location of gastric cancer and the difficulty of manipulation, which is caused by the flexibility of the scope, during ESD. There were no cases of bleeding required transfusion in 3D-ESD for EGIC.<br \/>\nConclusion<br \/>\nThe prototype 3D endoscopic system was validated the safety in ESD for EGIC. Further evaluation is necessary to clarify the utility or efficacy of 3D system in the luminal endoscopic procedure.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u3053\u306e\u7814\u7a76\u306f\uff0c\u4e09\u6b21\u5143\u5185\u8996\u93e1\u30b7\u30b9\u30c6\u30e0\u3092\u7528\u3044\u3066\u7ba1\u8154\u5185\u8996\u93e1\u624b\u8853\u306b\u304a\u3051\u308b\u6709\u7528\u6027\u3092\u691c\u8a0e\u3057\u305f\u3082\u306e\u3067\u3057\u305f\uff0e\u3053\u306e\u30b7\u30b9\u30c6\u30e0\u306f\u6709\u7528\u6027\u306e\u8a55\u4fa1\u304c\u96e3\u3057\u3044\u3068\u304a\u3063\u3057\u3083\u3063\u3066\u3044\u305f\u304c\uff0c\u4e09\u6b21\u5143\u72ec\u81ea\u306e\u5965\u884c\u304d\u3084\u7a7a\u9593\u306e\u8a8d\u8b58\u306f\u5fc5\u305a\u624b\u8853\u306e\u8a3a\u65ad\u3092\u652f\u63f4\u3059\u308b\u3068\u601d\u3044\u307e\u3057\u305f\uff0e\u4e8c\u6b21\u5143\u6620\u50cf\u3092\u4e09\u6b21\u5143\u6620\u50cf\u306b\u5909\u63db\u3059\u308b\u65b9\u6cd5\u306f\u805e\u3051\u306a\u304b\u3063\u305f\u304c\uff0c\u753b\u50cf\u51e6\u7406\u306e\u4e16\u754c\u3067\u306f\u4e09\u6b21\u5143\u30c7\u30fc\u30bf\u306e\u51e6\u7406\u304c\u4e00\u822c\u7684\u3067\u3042\u308b\u3068\u601d\u3046\u306e\u3067\uff0c\u6a5f\u4f1a\u304c\u3042\u308c\u3070\u77e5\u8b58\u3092\u3064\u3051\u305f\u3044\u3068\u601d\u3044\u307e\u3057\u305f\uff0e<br \/>\n&nbsp;<br \/>\n&nbsp;<br \/>\n\u53c2\u8003\u6587\u732e<\/p>\n<ul>\n<li><a href=\"https:\/\/ddw.org\/attendee-planning\/online-planner\">https:\/\/ddw.org\/attendee-planning\/online-planner<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>2019\u5e745\u670818\u65e5\uff5e5\u670821\u65e5\u306b\u304b\u3051\u3066\u30a2\u30e1\u30ea\u30ab \u30b5\u30f3\u30c7\u30a3\u30a8\u30b4\u306b\u3066\u958b\u50ac\u3055\u308c\u307e\u3057\u305fDigestive Disease Week2019\u306b\u53c2\u52a0\u3044\u305f\u3057\u307e\u3057\u305f\uff0e\u672c\u5b66\u4f1a\u306f\uff0c\u80c3\u8178\u75c5\u5b66\uff0c\u809d\u81d3\u75c5\u5b66\uff0c\u5185\u8996\u93e1\u691c\u67fb\uff0c\u80c3\u8178\u5916\u79d1\u306e\u5206\u91ce\u306b\u304a\u3051\u308b\u533b\u5e2b &hellip; <\/p>\n<p class=\"link-more\"><a href=\"https:\/\/is.doshisha.ac.jp\/news\/?p=6167\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;Digestive Disease Week 2019&#8221; \u306e<\/span>\u7d9a\u304d\u3092\u8aad\u3080<\/a><\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3],"tags":[],"class_list":["post-6167","post","type-post","status-publish","format-standard","hentry","category-3"],"_links":{"self":[{"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=\/wp\/v2\/posts\/6167","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6167"}],"version-history":[{"count":1,"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=\/wp\/v2\/posts\/6167\/revisions"}],"predecessor-version":[{"id":7033,"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=\/wp\/v2\/posts\/6167\/revisions\/7033"}],"wp:attachment":[{"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6167"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6167"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/is.doshisha.ac.jp\/news\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6167"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}