直播預(yù)告 | 人工智能 x 基礎(chǔ)科學(xué)系列論壇：當(dāng)醫(yī)學(xué)插上人工智能的翅膀

眾所周知，醫(yī)學(xué)是人們?nèi)粘Ｉ钪须x不開的重要場景之一，也是人工智能超級應(yīng)用場景之一。如今，利用人工智能尤其是深度學(xué)習(xí)、大模型來進(jìn)行醫(yī)學(xué)檢驗(yàn)、診療以及制造手術(shù)機(jī)器人的案例屢見不鮮，并有愈演愈烈的趨勢。

經(jīng)過幾年的碰撞與磨合，兩大領(lǐng)域進(jìn)入了理性和務(wù)實(shí)的狀態(tài)。高度靈活、可重復(fù)使用的人工智能模型的極快發(fā)展可能會(huì)為醫(yī)學(xué)帶來新的能力，兩者的融合以及催生出的應(yīng)用對現(xiàn)代醫(yī)學(xué)的演進(jìn)會(huì)產(chǎn)生深遠(yuǎn)的影響。

為了更好地促進(jìn)學(xué)術(shù)交流，尤其是交叉學(xué)科和前沿工作的同行交流，中國科學(xué)院自動(dòng)化研究所與機(jī)器之心聯(lián)合舉辦「人工智能 x 基礎(chǔ)科學(xué)系列論壇」，嘗試在更輕松和開放的交流氛圍下，邀請研究者分享近期工作，討論領(lǐng)域熱點(diǎn)問題。

6月29日15:00-17:00，中國科學(xué)院自動(dòng)化研究所、中國科學(xué)院香港創(chuàng)新院人工智能與機(jī)器人創(chuàng)新中心、機(jī)器之心聯(lián)合舉辦系列論壇第六期，以「當(dāng)醫(yī)學(xué)插上人工智能的翅膀」為主題，特邀中科院香港創(chuàng)新院副教授張忠凱博士主持，多位領(lǐng)域?qū)＜覍W(xué)者做技術(shù)分享，聚焦人工智能和醫(yī)學(xué)的深度融合和應(yīng)用，核心受眾為人工智能和醫(yī)學(xué)科學(xué)領(lǐng)域的學(xué)生、學(xué)者。

特邀主持人介紹

張忠凱，博士，中國科學(xué)院創(chuàng)新研究院人工智能與機(jī)器人創(chuàng)新中心副教授，于2023年2月全職歸國加入中科院香港創(chuàng)新研究院并組建醫(yī)學(xué)仿真團(tuán)隊(duì)。該團(tuán)隊(duì)致力于開發(fā)新一代實(shí)時(shí)仿真算法, 并將其應(yīng)用于機(jī)器人控制與手術(shù)模擬訓(xùn)練。2018年畢業(yè)于法國國家信息與自動(dòng)化研究院。博士期間解決了軟體機(jī)器人建模，控制與力反饋的多個(gè)關(guān)鍵性難題。作為核心研發(fā)人員開發(fā)的基于SOFA 的軟體機(jī)器人仿真軟體已成為行業(yè)標(biāo)準(zhǔn)。之后在法國斯特拉斯堡大學(xué)以及法國國家科研中心從事醫(yī)療機(jī)器人與人工智能的博士后研究。在此期間，研發(fā)了世界首臺具有OCT 掃描功能的內(nèi)窺鏡機(jī)器人的自動(dòng)控制算法，并提出了解決人工智能約束問題的新方法。自2021年起擔(dān)任法國農(nóng)業(yè)科學(xué)院數(shù)學(xué)與計(jì)算機(jī)研究中心終身研究員，領(lǐng)導(dǎo)團(tuán)隊(duì)從事移動(dòng)機(jī)器人與人工智能的研究工作。

特邀嘉賓與主題介紹

分享主題：智能手術(shù)器械的多模態(tài)信息感知、處理與應(yīng)用

嘉賓簡介：王廣志，博士，清華大學(xué)醫(yī)學(xué)院長聘教授、生物醫(yī)學(xué)工程系執(zhí)行系主任 。長期從事醫(yī)學(xué)影像處理、影像引導(dǎo)手術(shù)的教學(xué)與研究。發(fā)表學(xué)術(shù)研究論文150多篇，獲發(fā)明專利授權(quán)20多項(xiàng)?，F(xiàn)任中國生物醫(yī)學(xué)工程學(xué)會(huì)副理事長，中國醫(yī)學(xué)影像技術(shù)研究會(huì)副會(huì)長。

分享背景：近年來基于深度學(xué)習(xí)等人工智能方法的輔助診斷取得很大進(jìn)展，手術(shù)機(jī)器人等智能化醫(yī)療器械也日益受到業(yè)界關(guān)注，創(chuàng)新產(chǎn)品不斷涌現(xiàn)，并不斷拓展臨床應(yīng)用領(lǐng)域。

手術(shù)機(jī)器人等智能化器械的核心作用是有效地輔助醫(yī)生更加精準(zhǔn)、安全、高效地完成手術(shù)操作，規(guī)避手術(shù)風(fēng)險(xiǎn)，減少副作用。因此，多模態(tài)信息的智能化感知、識別、處理、決策和操作，已經(jīng)成為新一代智能化手術(shù)器械發(fā)展必不可少的核心技術(shù)。如何更有效地將上述智能化要素與臨床應(yīng)用場景結(jié)合，給醫(yī)生提供更有效的智能化裝備，成為研究的熱點(diǎn)，也提出了新的技術(shù)挑戰(zhàn)，需要密切融合人工智能的最新發(fā)展，形成新的解決方案。

分享摘要：結(jié)合我們的工作體會(huì)，分析影像引導(dǎo)手術(shù)場景中機(jī)器人智能輔助作用形式，并以神經(jīng)外科手術(shù)機(jī)器人為例，從多層次感知與決策角度，探討在臨床應(yīng)用中如何借助多模態(tài)信息感知，提升手術(shù)機(jī)器人的準(zhǔn)確性、便捷性及自主能力。

分享主題：多模態(tài)數(shù)據(jù)導(dǎo)航血管介入手術(shù)機(jī)器人

嘉賓簡介：謝曉亮，中國科學(xué)院自動(dòng)化研究所研究員，主要研究方向?yàn)槭中g(shù)機(jī)器人開發(fā)與臨床應(yīng)用、機(jī)器人智能控制。主持國家自然科學(xué)基金面上項(xiàng)目2項(xiàng)、青年基金1項(xiàng)、參加重點(diǎn)基金項(xiàng)目2項(xiàng);主持國家重點(diǎn)研發(fā)計(jì)劃智能機(jī)器人專項(xiàng)課題1項(xiàng);獲授權(quán)國家發(fā)明專利20余項(xiàng)、PCT專利3項(xiàng);在IEEE Trans. on Neural Networks and Learning Systems、IROS等國際期刊和會(huì)議上發(fā)表論文30余篇。獲機(jī)器人領(lǐng)域頂級會(huì)議IEEE ICRA最佳論文提名獎(jiǎng)1次，IEEE RCAR最佳論文獎(jiǎng)1次，獲得北京協(xié)和醫(yī)院科研成果獎(jiǎng)一等獎(jiǎng)。

分享背景：隨著人口老齡化的不斷加劇和生活水平的不斷提高，心腦血管疾病、腫瘤等具有典型老齡化特征的患者人群迅猛增長。例如，我國各類心腦血管疾病患者總數(shù)超過2.9億。龐大的患者人群給臨床治療與健康生活保障帶來巨大壓力，成為目前重大的國計(jì)民生問題。

血管介入手術(shù)是治療心腦血管最有效的方式。但是純?nèi)斯な中g(shù)存在諸多問題，需要臨床醫(yī)師具備豐富的臨床經(jīng)驗(yàn)、高超的手術(shù)技能、及良好的生理心理素質(zhì)，因此，技術(shù)精湛的醫(yī)生成為稀缺資源。借助機(jī)器人輔助醫(yī)生實(shí)施介入手術(shù)，能有效降低手術(shù)難度、提高手術(shù)精度。鑒于此，立足于當(dāng)前人工智能與先進(jìn)機(jī)器人技術(shù)的發(fā)展，開展了智能化精準(zhǔn)血管介入手術(shù)研究。包括：

開發(fā)了具有多器械協(xié)同遞送功能的血管介入手術(shù)機(jī)器人系統(tǒng)，并對其進(jìn)行精準(zhǔn)控制。根據(jù)同時(shí)協(xié)同遞送不少于2種器械的臨床需求，通過梳理分析各類手術(shù)器械的輸送時(shí)序與特點(diǎn)，結(jié)合醫(yī)生操作需求與仿生學(xué)原理，設(shè)計(jì)了新一代血管介入手術(shù)機(jī)器人，并成功完成了多中心多例臨床試驗(yàn)。

血管介入手術(shù)機(jī)器人多模影像導(dǎo)航。將術(shù)前的血管三維影像信息與術(shù)中影像疊加融合顯示在一起，為醫(yī)生提供了多視角的導(dǎo)航畫面。結(jié)合高級醫(yī)生力覺操控技能學(xué)習(xí)方法，研究了血管介入手術(shù)機(jī)器人擬人化操控策略，提升了機(jī)器人智能化層級，在擴(kuò)展手術(shù)機(jī)器人的適應(yīng)癥的同時(shí)，有效降低介入手術(shù)難度。

分享摘要：復(fù)雜血管環(huán)境對血管介入手術(shù)機(jī)器人的器械安全精準(zhǔn)操控提出了更大挑戰(zhàn)，本報(bào)告將從機(jī)器人本體設(shè)計(jì)、術(shù)中導(dǎo)航方法、臨床應(yīng)用三個(gè)層面介紹最新研究進(jìn)展。

相關(guān)項(xiàng)目：

[1] 中國科學(xué)院自動(dòng)化研究所，2035創(chuàng)新任務(wù)——多模態(tài)數(shù)據(jù)導(dǎo)航的血管介入手術(shù)機(jī)器人，2020/05至2023/04，負(fù)責(zé)人。

[2] 科技部，國家重點(diǎn)研發(fā)計(jì)劃，2019YFB1311700，面向復(fù)雜病變的多器械協(xié)同遞送血管介入手術(shù)機(jī)器人關(guān)鍵技術(shù)及應(yīng)用研究，2019/12至2022/11，課題1負(fù)責(zé)人。

[3] 國家自然科學(xué)基金委員會(huì)，面上項(xiàng)目，面向復(fù)雜病變的血管介入手術(shù)機(jī)器人自主操控關(guān)鍵技術(shù)研究，2021/01至2024/12，負(fù)責(zé)人。

相關(guān)論文：

[1] Xiaohu Zhou, Xiaoliang Xie, Zhenqiu Feng, Zengguang Hou, Guibin Bian, Ruiqi Li, Zhenliang Ni, Shiqi Liu, and Yanjie Zhou, “A mult and multimodal-fusion architecture for simultaneous recognition of endovascular manipulations and assessment of technical skills,” IEEE Transactions on Cybernetics, vol. 52, no. 4, pp. 2565-2577, April, 2022.

[2] Xiaohu Zhou, Xiaoliang Xie, Shiqi Liu, Zhenliang Ni, Yanjie Zhou, Ruiqi Li, Meijiang Gui, Chenchen Fan, Zhenqiu Feng, Guibin Bian, and Zengguang Hou, “Learning skill characteristics from manipulations,” IEEE Transactions on Neural Networks and Learning Systems, DOI: 10.1109/TNNLS.2022.3160159, March, 2022.

[3] Xiaohu Zhou, Xiaoliang Xie, Shiqi Liu, Zhenqiu Feng, Meijiang Gui, Jinli Wang, Hao Li, Tianyu Xiang, Guibin Bian, and Zengguang Hou, “Surgical skill assessment d on dynamic warping manipulations,” IEEE Transactions on Medical Robotics and Bionics, vol. 4, no. 1, pp. 50-61, February, 2022.

[4] Ruiqi Li, Xiaoliang Xie, Xiaohu Zhou, Shiqi Liu, Zhenliang Ni, Yanjie Zhou, Guibin Bian, and Zengguang Hou, “Real-time multi-guidewire endpoint localization in fluoroscopy images,”IEEE Transactions on Medical Imaging, vol. 40, no. 8, pp. 2002-2014, August, 2021.

[5] Yanjie Zhou, Xiaoliang Xie, Xiaohu Zhou, Shiqi Liu, Guibin Bian, and Zengguang Hou, “A real-time multi-functional work for guidewire morphological and positional analysis in interventional X-ray fluoroscopy,”IEEE Transactions on Cognitive and Developmental Systems, vol. 13, no. 3, pp. 657-667, September, 2021.

分享主題：Computational Analysis of Non-small Cell Lung Cancer Drug Resistance

嘉賓簡介：Hong Yan received his PhD degree from Yale University. He was Professor of Imaging Science at the University of Sydney and currently is Wong Chun Hong Professor of Data Engineering and Chair Professor of Computer Engineering at City University of Hong Kong. Professor Yan's research interests include image processing, pattern recognition, and bioinformatics. He has over 600 journal and conference publications in these areas. Professor Yan is an IEEE Fellow and IAPR Fellow. He received the 2016 Norbert Wiener Award from the IEEE SMC Society for contributions to image and biomolecular pattern recognition techniques. He is a foreign member of the European Academy of Sciences and Arts and a Fellow of the US National Academy of Inventors.

分享背景：Lung cancer has the highest incidence and mortality rates among all cancer types in the world. The mutation of a protein called the epidermal growth factor receptor (EGFR) is a major cause of non-small cell lung cancer (NSCLC). Anti-EGFR tyrosine kinase inhibitors (TKIs) can work effectively initially, which reduce tumor size and increase the patient survival time. However, almost all patients develop drug resistance after about a year of treatment due to one or more additional mutations of EGFR.

In collaboration with medical doctors, our research group has studied EGFR mutations at the molecular and atomic levels. We have collected EGFR mutations from the literature and from clinical cases at Queen Mary Hospital. Some of the mutations observed locally are rare and have never been reported in scientific journals before. d on computational models, we analyzed how the 3D structure of EGFR will change due to a mutation. Then for each drug, we computed its binding strength with EGFR before and after the secondary mutation. The reduction in the binding strength reflects the degradation of the drug effectiveness. We have built and published a 3D structural data of EGFR mutants. We have analyzed the characteristics of all known EGFR mutations.

Biomolecular surface complementarity is important in protein-drug interactions. We can find EGFR surface atoms using the alpha-shape model. The solid angles at these atoms represent the surface curvature, which is used as a geometric property in our analysis. Concave shapes around the binding sites are more likely to offer opportunities for drug binding than convex shapes. If the drug binds to EGFR mutants very tightly, then it will strengthen the response. We have studied the EGFR mutation-induced drug resistance by analyzing the solid angles around the binding sites. We have introduced the concept of eigen-binding surface to investigate common properties of biomolecular surfaces that are significant in protein-drug complexes.

EGFR is a member of the ErbB family. A pair of this family members can form a dimer to perform various biological functions. We have investigated the EGFR (ErbB-1) and ErbB-3 heterodimerization, regarded as the origin of intracellular signaling pathways. We combined the molecular interaction in EGFR heterodimerization with that between the EGFR tyrosine kinase and its inhibitor. For 168 clinical subjects, we characterized their corresponding EGFR mutations using molecular interactions, with three potential dimerization partners (ErbB-2, IGF-1R and c-Met) of EGFR and two of its small molecule inhibitors (gefitinib and erlotinib). d on molecular dynamics simulations and structural analysis, we modeled these mutant-partner or mutant-inhibitor interactions using the binding free energy and its components.

Our work leads to a deeper understanding of the mechanisms of cancer drug resistance. The knowledge gained can help the design of new and more effective drugs. The data s and computer algorithms developed provide a useful reference to medical doctors for assessment of drug resistance level for different EGFR mutants and for planning personalized treatment of lung cancer patients.

分享摘要：The mutation of a protein called the epidermal growth factor receptor (EGFR) is a major cause of non-small cell lung cancer (NSCLC). Almost all patients develop drug resistance after about a year of treatment due to one or more additional mutations of EGFR. In collaboration with medical doctors, we have built and published a 3D structural data of all known EGFR mutants. We have developed an alpha-shape d model to characterize the contact surfaces of EGFR-drug complexes and introduced the concept of eigen-binding surface to investigate common properties of biomolecular surfaces that are significant in protein-drug complexes. Our work leads to a deeper understanding of the mechanisms of cancer drug resistance. The knowledge gained can help the design of new and more effective drugs. The data s and computer algorithms developed provide a useful reference to medical doctors for assessment of drug resistance level for different EGFR mutants and for planning personalized treatment of lung cancer patients.

相關(guān)鏈接：

https://bcc.ee.cityu.edu.hk/med/