Daguang Xu

Research Manager of Deep Learning in medical imaging , NVIDIA, USA

Annotate, Build, and Adapt Models for Medical Imaging with the Clara Train SDK

Deep Learning in medical imaging has shown great potential for disease detection, localization, and classification within radiology. Deep Learning holds the potential to create solutions that can detect conditions that might have been overlooked and can improve the efficiency and effectiveness of the radiology team. However, for this to happen data scientists and radiologists need to collaborate to develop high-quality AI algorithms. The NVIDIA Clara AI toolkit, now generally available, lowers the barrier to the adoption of AI. The Clara Train SDK, part of the Clara AI toolkit, gives data scientists and developers the tools to accelerate data annotation, development, and adaptation of AI algorithms for medical imaging. In this talk, I will cover the features of the Clara Train SDK and NVIDIA’s solutions in the medical imaging sector.

Holger Roth

NVIDIA, USA

Tackling the Challenges of Next-generation Healthcare

Healthcare is currently undergoing and digital transformation. The Association of American Medical Colleges predicts a shortage of up to 120,000 physicians by 2030. Finding ways to assist doctors and avoid burnout will be critical. Over 70 percent of medical imaging research is already using deep learning for preventative care, population health and precision medicine. It is estimated that by 2025, 65 percent of all automated healthcare delivery processes will involve some form of AI. In this talk, I will present some of the ongoing challenges in medical imaging research and how NVIDIA’s applied research team addresses them, focusing on applications in medical image segmentation, classification, automatic machine learning, and its clinical applications.

Cheyu Hsu

Department of Oncology Imaging, National Taiwan University Hospital, Taiwan

Medical AI-driven Radiographic Phenotypic Characterization for Precision Cancer Treatment

Wen-Jeng Lee

Department of Medical Imaging, National Taiwan University Hospital, Taiwan

Building a cardiovascular imaging database for AI research

Artificial intelligence (AI) has recently been successful applied to many applications, including medical imaging. The basic steps of medical imaging AI development are: 1. Find a clinical unmet need. 2. Define data cohort 3. Collect medical images and clinical information 4. Data cleaning and de-identification 5. Dataset annotation 6. AI model building, validation and testing .

Our team is composed of well-experienced cardiologists and radiologists from various medical centers in Taiwan, as well as researchers specialized at AI/ database management/imaging processing/law of science and technology from different universities. The goal is to build a multi-modality coronary artery disease-based imaging bank, including coronary computed tomographic angiography (CCTA), single photon emission computed tomography (SPECT), invasive coronary angiography (ICA), intravascular ultrasound (IVUS), optical coherence tomography (OCT) and fractional flow reserve (FFR). In this presentation, I will talk about our experience in building a large imaging database for AI research and our preliminary results.

Junting Chen

Graduate Program of Data Science, National Taiwan University and Academia Sinica, Taiwan

Severe Stenosis Detection Using 2D Convolutional Recurrent Network

Stenosis detection is a critical criterion in the diagnosis of coronary artery disease (CAD). Manually detecting stenosis over complete coronary artery can be time-consuming. Our deep-learning-based stenoses detector can detect all stenoses that were greater than 70% in less than 20 seconds per patient, which achieves a significant time reduction in detection time.

Qingyi Tao

Solutions Architect , NVIDIA AI Technology Center, Singapore

Improving Deep Lesion Detection Using Attentive Feature Learning

Lesion detection from computed tomography (CT) scans is challenging compared to natural object detection because of two major reasons: small lesion size and small inter-class variation. To tackle both problems, we need to enrich the feature representation and improve the feature discriminativeness. Therefore, we introduce a dual-attention mechanism to the 3D contextual lesion detection framework, including the cross-slice contextual attention to selectively aggregate the information from different slices through a soft re-sampling process. Moreover,we propose intra-slice spatial attention to focus the feature learning in the most prominent regions. The results show that our model can significantly boost the results of the baseline lesion detector (with 3D contextual information) but using much fewer slices.

Tsung-Ming Tai

Solution Architect , NVIDIA, TAIWAN

Win-win strategy in Brain Age Prediction challenge - collaboration between data scientists and domain experts

This speech will share the experiences in building a brain age predictor in PAC2019 challenge in just one month, with a collaboration between YMU and NVIDIA experts.

Chiu-Wang Tseng

Solution Architect , NVIDIA, TAIWAN

Task Guided Modality Generation with MRI Image based on Deep Learning

Tinghui Wu

Research assistant of Institute of Applied Mathematical Sciences, National Taiwan University.

Deep Learning Accurately Distinguishes Pancreatic Cancer from Non-cancerous Pancreas

Pochuan Wang

Ph.D. Student , Department of Computer Science and Information Engineering, National Taiwan University

Using Multiple Cross-Country Datasets for Deep Learning Based Pancreas and Tumor Segmentation

Pochuan Wang is a Ph.D. student in the Department of Computer Science and Information Engineering, National Taiwan University. He received his BS in mathematics and MS in applied mathematics at National Taiwan University. His interests include parallel computing, distributed computing, GPU acceleration, and their applications. Currently, he is working on GPU acceleration on machine learning algorithms and medical images analysis with deep learning.