{"id":2212,"date":"2022-11-03T11:14:15","date_gmt":"2022-11-03T03:14:15","guid":{"rendered":"https:\/\/www.bebi.ntu.edu.tw\/?p=2212"},"modified":"2022-11-03T11:14:15","modified_gmt":"2022-11-03T03:14:15","slug":"%e7%94%9f%e9%86%ab%e9%9b%bb%e8%b3%87%e6%89%80%e6%95%99%e5%b8%ab%e7%a0%94%e7%a9%b6%e4%ba%ae%e9%bb%9e-111%e5%b9%b411%e6%9c%88%e4%bb%bd%e3%80%8c%e8%8e%8a%e6%9b%9c%e5%ae%87%e6%95%99%e6%8e%88%e3%80%8d","status":"publish","type":"post","link":"https:\/\/www.bebi.ntu.edu.tw\/?p=2212","title":{"rendered":"\u751f\u91ab\u96fb\u8cc7\u6240\u6559\u5e2b\u7814\u7a76\u4eae\u9ede-111\u5e7411\u6708\u4efd\u300c\u838a\u66dc\u5b87\u6559\u6388\u300d"},"content":{"rendered":"

\u7814\u7a76\u4e3b\u984c\uff1a\u61c9\u7528\u6df1\u5ea6\u6a5f\u5668\u5b78\u7fd2\u518dDNA\u5b9a\u5e8f\u4ee5\u53ca\u75c5\u7406\u5207\u7247<\/p>\n

\u64b0\u5beb\u4eba\uff1a\u6f58\u65e5\u5357\u535a\u58eb\u3001\u5b78\u751f\u9673\u7ff0\u5112<\/p>\n

\u7814\u7a76\u5982\u4f55\u5c07\u4eba\u5de5\u667a\u6167\u61c9\u7528\u5230\u6b21\u4e16\u4ee3\u5b9a\u5e8f\u3001\u91ab\u5b78\u5f71\u50cf\u7b49\u9ad8\u901a\u91cf\u8cc7\u6599\u4e2d\u53ef\u4f7f\u6211\u5011\u5c0d\u7cbe\u6e96\u91ab\u5b78\u7684\u9818\u57df\u6709\u66f4\u6df1\u5c64\u7684\u7406\u89e3\u3002\u76ee\u524d\u838a\u66dc\u5b87\u6559\u6388\u7684\u5718\u968a\u5df2\u7d93\u4f7f\u7528\u4e86DNA\u5b9a\u5e8f\u8cc7\u6599\u53ca\u75c5\u7406\u5207\u7247\u5f71\u50cf\u958b\u767c\u4e86\u4e09\u7a2e\u9ad8\u6548\u80fd\u7684\u6df1\u5ea6\u5b78\u7fd2\u5de5\u5177\uff0c\u4e14\u53ef\u61c9\u7528\u65bc\u5206\u985e\u53ca\u8fa8\u8b58\u7279\u5b9a\u76ee\u6a19\u7269\u4e0a\u3002<\/p>\n

\u7b2c\u4e00\u9805\u7814\u7a76\u70ba\u767c\u8868Briefings in bioinformatics<\/em>\u4e2d\u7684 \u201dHigh-performance deep learning pipeline predicts individuals in mixtures of DNA using sequencing data\u201d\u3002\u4f7f\u7528\u57fa\u56e0\u5b9a\u5e8f\u8cc7\u6599\u958b\u767c\u4e00\u5957\u65b0\u7a4e\u4e14\u57fa\u65bc\u6df1\u5ea6\u5b78\u7fd2\u7684\u9810\u6e2c\u65b9\u6cd5\u53bb\u5075\u6e2c\u53ca\u5206\u985e\u6df7\u5408\u7269\u4e2d\u7684\u4e0d\u540c\u500b\u9ad4\u3002\u70ba\u8b49\u660e\u6b64\u6280\u8853\u540c\u6a23\u53ef\u7528\u65bc\u5176\u4ed6\u4e0d\u540c\u7684\u8cc7\u6599\u4e2d\uff0c\u8a72\u6a21\u578b\u5728\u958b\u767c\u7684\u968e\u6bb5\u4f7f\u7528\u4e86\u4f86\u81ea\u4e0d\u540c\u5b9a\u5e8f\u5e73\u53f0\u7684\u8cc7\u6599\uff0c\u5305\u62ec\uff1a1.\u76ee\u6a19\u5340\u9593\u5b9a\u5e8f\u30012.\u5168\u5916\u986f\u5b50\u5b9a\u5e8f\uff08WES\uff09\u3002\u5728\u7b2c\u4e00\u500b\u8cc7\u6599\u4e2d\uff0c\u6211\u5011\u5229\u7528\u76ee\u6a19\u768427\u500b\u77ed\u7247\u6bb5\u91cd\u8907\u5e8f\u5217\u53ca94\u500b\u55ae\u6838\u82f7\u9178\u591a\u578b\u6027\u4f86\u88fd\u5099\u6df7\u96dc\u4e0d\u540c\u500b\u9ad4\u7684DNA\u6a23\u672c\uff0c\u4e26\u4f7f\u7528\u6df1\u5ea6\u5b78\u7fd2\u6a21\u578b\u53bb\u5340\u5206\u51fa\u6bcf\u500b\u500b\u9ad4\u4e26\u53ef\u9054\u523095-97\uff05\u6e96\u78ba\u7387\u3002\u7b2c\u4e8c\u500b\u8cc7\u6599\u96c6\u5247\u4f7f\u7528\u4e73\u764c\u60a3\u8005\u7684WES\u8cc7\u6599\uff0c\u4e26\u53ef\u5b8c\u5168\u6b63\u78ba\u5730\uff08100\uff05\uff09\u9810\u6e2c\u75c5\u60a3\u4e4b\u75be\u75c5\u4e9e\u578b\u3002\u6b64\u5916\u70ba\u514b\u670d\u6bcf\u500b\u5e8f\u5217\u4e4b\u9593\u9577\u5ea6\u7684\u5dee\u7570\uff0c\u6211\u5011\u4f7f\u7528\u65b0\u7684sliding window\u65b9\u6cd5\u53ef\u5927\u5e45\u63d0\u5347\u6a21\u578b\u6548\u80fd\u3002\u7e3d\u7d50\u4f86\u8aaa\uff0c\u672c\u7814\u7a76\u91dd\u5c0d\u5e8f\u5217\u8cc7\u6599\u7684\u8655\u7406\u548c\u6df1\u5ea6\u5b78\u7fd2\uff0c\u63d0\u51fa\u4e00\u9805\u80fd\u9069\u7528\u65bc\u4e0d\u540c\u6b21\u4e16\u4ee3\u5b9a\u5e8f\u5e73\u53f0\u4e0a\u7684\u65b9\u6cd5\u3002<\/p>\n

\u7b2c\u4e8c\u9805\u7814\u7a76\u70ba\u767c\u8868Frontiers in oncology<\/em>\u4e2d\u7684\u201dPredicting Breast Cancer Gene Expression Signature by Applying Deep Convolutional Neural Networks from Unannotated Pathological Images\u201d\u3002\u57fa\u65bc\u75c5\u7406\u5207\u7247\u5f71\u50cf\u4e4b\u6613\u53d6\u5f97\u6027\u548c\u4f7f\u7528\u4e73\u764c\u60a3\u8005\u768470\u500b\u57fa\u56e0\u8a08\u7b97\u51fa\u4f86\u7684\u5fa9\u767c\u98a8\u96aa\uff0c\u7b2c\u4e8c\u9805\u7814\u7a76\u63d0\u51fa\u4e00\u6df1\u5ea6\u5b78\u7fd2\u6a21\u578b\u4f7f\u7528\u75c5\u7406\u5207\u7247\u5f71\u50cf\u9032\u884c\u4e73\u764c\u5fa9\u767c\u7387\u7684\u9810\u6e2c\uff0c\u63d0\u4f9b\u4e00\u5feb\u901f\u3001\u4f4e\u6210\u672c\u4ee5\u4e14\u5065\u5168\u4e4b\u4e73\u764c\u5fa9\u767c\u7387\u9810\u6e2c\u5de5\u5177\uff0c\u5e6b\u52a9\u91ab\u5e2b\u9032\u884c\u6cbb\u7642\u8a08\u756b\u7684\u8a55\u4f30\u3002\u672c\u7814\u7a76\u4f7f\u7528\u516d\u500b\u9810\u8a13\u7df4\u6a21\u578b\u9032\u884c\u9077\u79fb\u5b78\u7fd2\u3002\u5728\u9a57\u8b49\u8cc7\u6599\u4e2d\uff0cpatch-wis\u7684\u65b9\u6cd5\u67090.87\u7684\u6e96\u78ba\u7387;\u4e14patient-wise\u65b9\u6cd5\u4e2d\uff0c\u9ad8\u98a8\u96aa\u53ca\u4f4e\u98a8\u96aa\u985e\u5225\u5206\u5225\u67090.90\u53ca1.00\u7684\u6e96\u78ba\u7387\u3002\u7e3d\u7d50\u4f86\u8aaa\uff0c\u9019\u9805\u7814\u7a76\u8b49\u660e\u4e86\u75c5\u7406\u5207\u7247\u5f71\u50cf\u5728\u672a\u6a19\u6ce8\u7279\u5b9a\u5340\u57df\u7684\u60c5\u6cc1\u4e0b\uff0c\u4ecd\u53ef\u5efa\u7acb\u51fa\u9ad8\u6548\u80fd\u4e4b\u4eba\u5de5\u667a\u6167\u6a21\u578b\u4f86\u9810\u6e2c\u764c\u75c7\u7684\u5fa9\u767c\u7387\u3002<\/p>\n

\u7b2c\u4e09\u7bc7\u7814\u7a76\u70ba\u767c\u8868\u65bcFrontiers in Oncology<\/em>\u7684\u201dPrediction of Breast Cancer Recurrence Using a Deep Convolutional Neural Network Without Region-of-Interest Labeling\u201d\u3002\u5229\u7528\u6df1\u5ea6\u5b78\u7fd2\u9810\u6e2c\u4e73\u764c\u4e9e\u578b\u4e26\u63d0\u4f9b\u4e00\u4fbf\u5229\u4e4b\u4e73\u764c\u8a3a\u65b7\u7b56\u7565\uff0c\u8fd1\u4e00\u6b65\u964d\u4f4e\u9032\u884cmRNA\u8868\u9054\u91cf\u5206\u6790\u4ee5\u53ca\u514d\u75ab\u7d44\u7e54\u5316\u5b78\u67d3\u8272\u7684\u6210\u672c\u3002\u6211\u5011\u671f\u671b\u4f7f\u7528\u4e0a\u4e00\u9805\u7814\u7a76\u6240\u8a13\u7df4\u7684\u6a21\u578b\u6b0a\u91cd\u9032\u884c\u5169\u968e\u6bb5\u9077\u79fb\u5b78\u7fd2\u4e26\u61c9\u7528\u5230\u75c5\u7406\u5207\u7247\u5f71\u50cf\u4e0a\u3002\u6211\u5011\u4f7f\u7528\u4f86\u81ea\u56db\u500b\u9810\u8a13\u7df4\u6a21\u578b\u7684\u6b0a\u91cd\u4ee5\u53caTCGA-BRCA\u7684\u8cc7\u6599\u96c6\u505a\u56db\u7a2e\u4e73\u764c\u4e9e\u578b\u7684\u9810\u6e2c\u6a21\u578b\u3002\u6b64\u5916\uff0c\u4f7f\u7528Imagenet\u6b0a\u91cd\u7684ResNet101\u88ab\u7528\u65bc\u8207\u4e0a\u8ff0\u6a21\u578b\u9032\u884c\u6bd4\u8f03\u3002\u5728\u5206\u985e\u7d50\u679c\u4e0a\uff0c\u6b64\u5169\u968e\u6bb5\u9077\u79fb\u5b78\u7fd2\u6709\u512a\u7570\u7684\u8868\u73fe\uff0cResNet101\u5728slide-wise\u7684\u9810\u6e2c\u6e96\u78ba\u7387\u9054\u52300.913\u3002\u6b64\u6df1\u5ea6\u5b78\u7fd2\u6a21\u578b\u4ea6\u7528\u65bc\u8207\u53e6\u4e00\u5e38\u7528\u7684\u4e73\u764c\u5206\u985e\u5de5\u5177Genefu\u9032\u884c\u6bd4\u8f03\uff0c\u5728\u6bd4\u8f03\u7684\u7d50\u679c\u4e2d\uff0c\u6df1\u5ea6\u5b78\u7fd2\u6a21\u578b\u6709\u8207Genefu\u5ab2\u7f8e\u7684\u8868\u73fe\u4e14\u5728\u7279\u5b9a\u4e73\u764c\u4e9e\u578b\u4e2d\u6709\u66f4\u512a\u7570\u7684\u9810\u6e2c\u80fd\u529b\u3002<\/p>\n

\u6df1\u5ea6\u5b78\u7fd2\u6280\u8853\u5df2\u88ab\u61c9\u7528\u81f3\u8a31\u591a\u7814\u7a76\u4e2d\uff0c\u4e26\u5df2\u88ab\u6574\u5408\u5230\u73fe\u4eca\u7684\u91ab\u7642\u7167\u8b77\u7cfb\u7d71\u4e4b\u4e2d\uff0c\u4ee5\u589e\u9032\u75be\u75c5\u7684\u8a3a\u65b7\u4ee5\u53ca\u9810\u5f8c\u7684\u5224\u5b9a\u3002\u7f8e\u570b\u98df\u54c1\u85e5\u7269\u7ba1\u7406\u5c40\u4e5f\u5df2\u5236\u5b9a\u5b8c\u5584\u7684\u6a5f\u5668\u5b78\u7fd2\u6a19\u6e96\uff0c\u7528\u65bc\u7ba1\u7406\u6df1\u5ea6\u5b78\u7fd2\u53ca\u4eba\u5de5\u667a\u6167\u5de5\u5177\u7684\u61c9\u7528\uff0c\u4e26\u66f4\u9032\u4e00\u6b65\u6210\u70ba\u6a21\u578b\u958b\u767c\u3001\u8cc7\u6599\u96c6\u5efa\u7acb\u548c\u90e8\u7f72\u5230\u91ab\u9662\u7684\u9ec3\u91d1\u6a19\u6e96\u3002<\/p>\n

 <\/p>\n

Application of Machine Learning methods for biomedical research utilizing high-throughput data, such as imaging data and next generation sequencing data, has allowed deeper understanding towards expansion of precision medicine and improvement of public health issues. Deep learning (DL) is the latest sub-branch of ML and has been introduced with the aspiration of bringing ML closer to AI. At Professor Eric Y Chuang\u2019s lab, we develop high performance DL pipelines, applicable universally for classification and identification tasks in this research work, 3 pipelines were developed to process pathological images and DNA sequencing data, respectively.<\/p>\n

For the first study published in the Journal of Briefings in bioinformatics (2021) <\/em>entitled \u201cHigh-performance deep learning pipeline predicts individuals in mixtures of DNA using sequencing data\u201d, a novel DL pipeline was proposed that utilized DNA sequencing data to successfully detect and classify different individuals. To prove the global applicability of the pipeline, it was implemented on datasets generated using different sequencing technologies: (i) targeted sequencing and (ii) whole exome sequencing data. For the first application, individuals were identified with 95-97% accuracy, from mixtures of DNA samples, prepared using targeted 27 short tandem repeats and 94 single nucleotide polymorphisms. WES data from breast cancer patients were used for the second application, and the pipeline could correctly classify all patients (100%) into subtypes. A new sliding window approach was proposed and applied, to overcome the sequence length variation problem of sequencing data, which dramatically improved the model performance. Overall, a complete pipeline, including sequencing data processing steps and DL steps is proposed that is applicable across different NGS platforms.<\/p>\n

The second study published in the journal of Frontiers in oncology<\/em> (2021) entitled \u201cPredicting Breast Cancer Gene Expression Signature by Applying Deep Convolutional Neural Networks from Unannotated Pathological Images\u201d. To leverage the availability of whole slide images data and the recurrence risk score provided by a 70 gene-signature from breast cancer patients, a DL model was proposed for the second study to predict the breast cancer recurrence status using only pathological images. This provides a rapid, cost-effective and robust predictive tool which would assist medical doctor in treatment recommendation. 6 pre-trained models (VGG16, ResNet50, ResNet101, Inception_ResNet, EfficientB5, and Xception) were used for transfer learning and their performances were evaluated based on accuracy, precision, recall, F1 score, confusion matrix, and AUC. Xception demonstrated highest validation performance with an overall accuracy of 0.87 for a patch-wise approach and 0.90 and 1.00 for a patient-wise approach for high-risk and low-risk groups, respectively. Taken together, this study demonstrated the feasibility and high performance of artificial intelligence models trained without region-of-interest labeling for predicting cancer recurrence.<\/p>\n

Finally, Prof. Chuang published another study in the journal of Frontiers in Oncology<\/em> (2021) with the title \u201cPrediction of Breast Cancer Recurrence Using a Deep Convolutional Neural Network Without Region-of-Interest Labeling\u201d.\u00a0Deciphering breast cancer molecular subtypes by DL approaches could provide a convenient and method for the diagnosis of breast cancer patients. It could reduce costs associated with transcriptional profiling and subtyping discrepancy between IHC assays and mRNA expression. Therefore, we aim to develop a highly versatile 2-steps transfer learning pipeline for pathological images using weight obtained from model trained with the 70 gene signature images, for our final study. Weights from 4 pre-trained models namely VGG16, ResNet50, ResNet101, and Xception were used to train TCGA-BRCA datasets to predict 4 intrinsic breast cancer subtypes. Furthermore, ResNet101 model was used for training with weights from ImageNet for comparison with the aforementioned models. The 2-steps DL models showed promising classification results with the overall accuracy of slide-wise prediction as 0.913 with ResNet101 model. The DL model was additionally benchmarked with the common Genefu tool for breast cancer classification. The results demonstrated that the performance of the DL model is comparable to that of Genefu, even superior in certain breast cancer subtypes.<\/p>\n

DL technology is applied routinely in the laboratory and is integrated into the current health care system to facilitate diagnosis and determination of prognosis. Good machine learning protocol has also been released by U.S FDA for managing the applications of DL and artificial intelligence tools and are made golden standard for model development, dataset preparation and deployment into the hospital. Eventually, artificial intelligence tools would make health care system less vulnerable to emergent situations which are otherwise not handled the best under current healthcare protocols.<\/p>\n

 <\/p>\n

References <\/strong><\/p>\n

    \n
  1. Phan, N.N., Chattopadhyay, A., Lee, T.T., Yin, H.I., Lu, T.P., Lai, L.C., Hwa, H.L., Tsai, M.H. and Chuang, E.Y., 2021. High-performance deep learning pipeline predicts individuals in mixtures of DNA using sequencing data.Briefings in bioinformatics<\/em>,\u00a022<\/em>(6), p.bbab283.<\/li>\n
  2. Phan, N.N., Hsu, C.Y., Huang, C.C., Tseng, L.M. and Chuang, E.Y., 2021. Prediction of Breast Cancer Recurrence Using a Deep Convolutional Neural Network Without Region-of-Interest Labeling.Frontiers in Oncology<\/em>,\u00a011<\/em>, pp.734015-734015.<\/li>\n
  3. Phan, N.N., Huang, C.C., Tseng, L.M. and Chuang, E.Y., 2021. Predicting Breast Cancer Gene Expression Signature by Applying Deep Convolutional Neural Networks From Unannotated Pathological Images. Frontiers in oncology<\/em>,\u00a0<\/span>11<\/em>, pp.769447-769447.<\/span><\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"

    \u7814\u7a76\u4e3b\u984c\uff1a\u61c9\u7528\u6df1\u5ea6\u6a5f\u5668\u5b78\u7fd2\u518dDNA\u5b9a\u5e8f\u4ee5\u53ca\u75c5\u7406\u5207\u7247 \u64b0\u5beb\u4eba\uff1a\u6f58\u65e5\u5357\u535a\u58eb\u3001\u5b78\u751f\u9673\u7ff0\u5112 \u7814\u7a76\u5982\u4f55\u5c07\u4eba\u5de5\u667a\u6167\u61c9\u7528\u5230\u6b21 […]\n","protected":false},"author":2,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[19],"tags":[],"class_list":["post-2212","post","type-post","status-publish","format-standard","hentry","category-award"],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=\/wp\/v2\/posts\/2212","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2212"}],"version-history":[{"count":1,"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=\/wp\/v2\/posts\/2212\/revisions"}],"predecessor-version":[{"id":2213,"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=\/wp\/v2\/posts\/2212\/revisions\/2213"}],"wp:attachment":[{"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2212"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2212"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.bebi.ntu.edu.tw\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2212"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}