皮肤癌的AI诊断课件.pptx

1、PPT制作、主讲 朱朱明哲明哲文献查找、翻译、Word文档制作 李卓李卓生物医学传感领域的新突破生物医学传感领域的新突破百度搜索引擎都膜拜的第一百度搜索引擎都膜拜的第一作者作者荣获斯坦福大学荣获斯坦福大学AI领域的领域的理学博士理学博士并修德克萨斯大学奥斯汀并修德克萨斯大学奥斯汀分校电子计算机工程及纯分校电子计算机工程及纯数学双专业数学双专业Dermatologist-level classification of skin cancerwith deep neural networks利用深度神经网络，进行皮肤科专家级别的皮肤癌诊断这项研究成果被x-mol官网评为2017年年2月全球科学技术

2、十大突破之五月全球科学技术十大突破之五这种所谓的深度神经网络这种所谓的深度神经网络就是近年来经常提及就是近年来经常提及 但但又充满神秘的又充满神秘的从数学上看从数学上看深度神经网络深度神经网络(Deep Neural Networks)卷积神经网络卷积神经网络(Convolutional Neural Networks)浅析浅析CNNsLogistic回归模型浅析浅析CNNs卷积的处理看起来使得问题复杂了，这是否会增大计算难度，降低效率？课堂讨论环节显然让课堂效率提高了而它正是卷积过程的体现Layer L1在座近100位同学都提出了自己的观点Layer L2通过小组讨论，初步得出问题的解答La

3、yer L3对每个小组的讨论结果进行评估，考虑其对问题的偏离程度（权重），得出最终结果CNNs的工作就是将全局映射转化为区域映射的过程。另一个现实中的例子是图像处理Classical diagnosing methodInitial clinical screening Dermoscopic analysisBiopsyHistopathological examinationObserving undermicroscopeDiagnosticreportFormation of AI dermatologist A process of machine learningFigure 1

4、| Deep CNN layout. Formed by Google Inception v3 CNN architectureStep1 They demonstrated classification of skin lesions using a single CNN, trained end-to-end from images directly, using only pixels and disease labels as inputs.Figure 2 | A schematic illustration of the taxonomy and example test set

5、 imagesa, A subset of the top of the tree-structured taxonomy of skin disease.b, Malignant and benign example images from two disease classes.Step 2 They trained a CNN using a dataset of 129,450 clinical images, consisting of 2,032 different diseases.Step 3 Grasp the characteristic message of pictur

6、es by analyzing pixels.Figure 3 | Skin cancer classification performance of dermatologists (a) and CNN (b).Practical diagnose Analyzing up-loaded picture from mobile devicesFigure 4 | t-SNE visualization of the last hidden layer representations in the CNN for four disease classesReal case analysis a

7、s support information Outfitted with deep neural networks, mobile devices can potentially extend the reach of dermatologists outside of the clinic. It is projected that 6.3 billion smartphone subscriptions will exist by the year 2021 and can therefore potentially provide low-cost universal access to

8、 vital diagnostic care.Ericsson Mobility Report, 2016 Deep learning algorithms, powered by advances in computation and very large datasets, have recently been shown to exceed human performance in visual tasks such as playing Atari games, strategic board games like Go and object recognition.AI应用于生活的几个例子2017年5月23日5月27日 “第二次围棋人机大战” 论决策和价值评估，人类论决策和价值评估，人类真的输了！真的输了！Demis Hassabis，创业公司DeepMind创始人。 “AI是人类探索世界的工具。”像AlphaGo一样思考v对历史的旁征博引v对当今的运筹帷幄v对未来的高瞻远瞩“人生如棋”Karl Popper(1902.7.281994.9.17) “一切伟大的科学理论都意味着对未知领域的新征服。”