Leading Pediatric Hospital Reveals Best AI Models in COVID-19 Grand Challenge

The top 10 results were unveiled in the first of its kind COVID-19 Lung CT Lung Segmentation Grand Challenge, a groundbreaking research competition focused on the development of artificial intelligence (AI) models to help visualize and measure COVID-specific lesions. in the lungs of infected patients, potentially facilitating faster and more patient-specific medical interventions.

Attracting over 1,000 participants worldwide, the competition was presented by the Sheikh Zayed Institute for Pediatric Surgical Innovation at Children’s National Hospital, in collaboration with leading AI technology company NVIDIA and the National Institutes of Health (NIH) . The competition’s AI models used a multi-institutional and multinational dataset provided by public datasets from The Cancer Imaging Archive (National Cancer Institute), NIH, and the University of Arkansas, which were sourced from patients of varying ages, sexes and illnesses. gravity. NVIDIA provided GPUs to the top five winners as prizes, and supported the selection and evaluation process.

“Improving COVID-19 treatment begins with a better understanding of the patient’s condition. However, a previous lack of global data collaboration limited clinicians in their ability to quickly and effectively understand the severity of the disease in adult and pediatric patients, “says Marius George Linguraru, D.Phil., MA, M.Sc., principal investigator at the Sheikh Zayed Institute for Pediatric Surgical Innovation at Children’s National, who led the Grand Challenge initiative. “By harnessing the power of AI through quantitative imaging and machine learning, these findings are helping clinicians better understand the severity of COVID-19 disease and potentially stratify and sort out appropriate treatment protocols to different stages of the disease. “

The top 10 AI algorithms were identified from a highly competitive field of participants who tested the data in November and December 2020. The results were unveiled on January 11, 2021, during a virtual symposium, hosted by Children’s National, which included presentations from top teams, event planners and clinicians.

The developers of the top 10 AI models of the great COVID-19 pulmonary CT lesion segmentation challenge are:

1. Shishuai Hu et al. Northwestern Polytechnic University, China. “Semi-supervised method for segmentation of pulmonary CT lesions COVID-19”
2. Fabian Isensee et al. German Cancer Research Center, Germany. “nnU-Net for Covid segmentation”
3. Claire Tang, Lynbrook High School, United States. “Automated ensemble modeling for the segmentation of COVID-19 CT lesions”
4. Qinji Yu et al. Shanghai JiaoTong University, China. “Segmentation of COVID-19-20 lesions based on nnUNet”
5. Andreas Husch et al. University of Luxembourg, Luxembourg. “Leveraging advanced architectures by enriching training information – a case study”
6. Tong Zheng et al. Nagoya University, Japan. “Fully automated COVID-19-20 segmentation”
7. Vitali Liauchuk. United Institute of Computer Problems (UIIP), Belarus. “Semi-3D CNN with ImageNet pre-training for segmentation of Covid lesions on CT”
8. Ziqi Zhou et al. Shenzhen University, China. “Automated segmentation of chest computed tomography images of COVID-19 with a 3D Unet frame”
9. Jan Hendrik Moltz et al. Fraunhofer Institute for Digital Medicine MEVIS, Germany. “Segmentation of COVID-19 lung injury by CT using nnU-Net”
10. Bruno Oliveira et al. 2Ai – Polytechnic Institute of Cávado and Ave, Portugal. “Automatic detection and segmentation of COVID-19 from pulmonary computed tomography (CT) images using a 3D cascade U-net”

Linguraru added that in addition to an award for the top five AI models, these winning algorithms are now available to partner with clinical institutions around the world to further assess the impact of these quantitative imaging methods. and machine learning on global public health.

Quality annotations are a limiting factor in the development of useful AI models. Using the NVIDIA COVID Lesion Segmentation Model available on our NGC software hub, we were able to quickly tag the NIH dataset, allowing radiologists to make precise annotations in record time. “

Mona Flores, MD, Global Head of Medical AI, NVIDIA

“I congratulate the global academic community of computer science, data science and image processing for quickly coming together to combine multidisciplinary expertise for the development of potential automated and multi-parametric tools for better to investigate and respond to the myriad of unmet clinical needs created by the pandemic, ”said Bradford Wood, MD, director of the NIH Center for Interventional Oncology and head of the Interventional Radiology section of the NIH Clinical Center.“ Thank you to each team for to have locked the arms towards a common cause which unites the scientific community in these difficult times. “