Annual Report 2017–18

26 香港中文大學年報 CUHK Annual Report 2017–2018 肝癌患者佳音 醫學院楊明明教授、余俊豪教授和李傑輝醫 生(上圖)證實,新一代經血管治療「消融化 療栓塞術」與傳統「經血管治療」手術相比, 能增加中期肝癌患者的存活期近兩倍。「消 融化療栓塞術」能有效消滅癌細胞,使部分 病者可接受手術,提高治癒機會。 高效癌症診斷 計算機科學與工程學系王平安教授,研發了 摻合深度學習的人工智能影像識別技術,以 增加電腦斷層掃描及病理組織切片等醫學 影像的效率。研究結果顯示,利用此嶄新技 術分析肺癌及乳腺癌影像,其準確率分別高 達91%及99%,未來或可應用此技術於本地 醫療界。 中風病人好助手 生物醫學工程學系湯啟宇教授及外科學系 李崢教授研發自訂3D打印軟體復康機械 手,幫助中風病人運動手部,加速復元。該 機械手按照用者手部形狀及大小訂製,質 料柔軟,比傳統儀器輕巧,方便病者在家中 鍛練。 Boosting Survival Rate of Liver Cancer Patients A research team led by Prof. Winnie Yeo, Prof. Simon Yu and Dr. Lee Kit-fai of the Faculty of Medicine (top photo) proved that comparing with conventional transarterial chemoembolization, a new transarterial treatment called ‘ablative chemoembolization’ (ACE) could double the progression-free survival period of liver cancer patients at the intermediate stage. ACE can effectively kill off tumour cells, thereby allowing some intermediate liver cancer patients to receive hepatectomy, a treatment method that may not have been available to them in the past. Efficient Cancer Diagnoses To improve the efficiency of CT scan and histopathological imaging in cancer diagnoses, Prof. Pheng Ann Heng of the Department of Computer Science and Engineering devised an automated image processing technology that made use of deep learning. The technology, tested on lung cancer and breast cancer, was able to achieve diagnostic accuracies of 91% and 99%, respectively. This innovative method may be adopted by the local medical sector in future. Giving a Hand to Stroke Patients Prof. Raymond Tong of the Department of Biomedical Engineering and Prof. Zheng Li of the Department of Surgery developed a 3D-printed soft robotic hand as a handy and affordable tool to help stroke patients regain their hand function. The custom-made soft robotic hand, significantly lighter and cheaper than conventional rehabilitation devices, also offers stroke patients the convenience of at-home training. (左起)楊明明教授、余俊豪教授、李傑輝醫生 (From left) Prof. Winnie Yeo, Prof. Simon Yu and Dr. Lee Kit-fai

RkJQdWJsaXNoZXIy NDE2NjYz