报告摘要：作为生物光子学在临床实践中应用的成功典范，光学相干层析成像（OCT）是一种具有非侵入性、实时、高灵敏度和高分辨率等特点的三维成像技术。OCT可以实现体内“无创显微层析”的光学活检，提高OCT系统的空间分辨率（尤其是横向分辨率）以区分“组织之下的层次—细胞”，具有重要的临床价值。随着硬件水平（如激光器、光纤技术和探测器等）和计算技术（如AI图像处理和计算光学成像）的蓬勃发展，OCT在眼科、皮肤病科、介入心脏病学以及无损检测等领域的应用越来越广泛。微米分辨的光学相干层析成像（μOCT）是目前分辨率最高的 OCT 技术。通过分析μOCT的光谱信息，能够得到不同组织结构的特异性指示。通过多孔径合成技术，能够提高OCT系统的横向分辨率并扩展焦深。通过动物实验和人体实验，能够清晰分辨多种上皮组织的亚细胞结构。同时，我们也积极探索了μOCT技术在多模态成像领域中的应用。
赵杰
东南大学计算机学院 博士生四年级
报告题目：Smart DR for body scanning: Relieving radiographers’ workload and improving accuracy
报告摘要：In a traditional radiography procedure, radiographers’ poor practice of patients positioning and/or exposure parameter setting might result in image retakes, an accident that requires the patient to receive extra ionizing radiation. We propose a smart module to assist radiographers to minimize the possibility of image retakes. The smart module utilizes a RGBD camera group, an embedded computing platform to predict 3D skeleton of the patients using a deep learning model, and combine domain expertise to suggest proper positioning and exposure parameters. Moreover, the smart module automatically drives collimator blades to constrain the X-ray exposure area to an appropriate region of interest. This system was applied in several examinations, and received positive feedback from experienced radiographers. It is promising to extend to full body examinations.