Children’s National Hospital unveils top artificial intelligence models in COVID-19 Grand Challenge to improve lung diagnostics


Foremost pediatric healthcare facility collaborates with NVIDIA and the NIH to identify far more productive diagnostic remedies for COVID-19


January twelve, 2021

The top 10 final results have been unveiled in the 1st-of-its-sort COVID-19 Lung CT Lesion Segmentation Grand Challenge, a groundbreaking investigation level of competition concentrated on acquiring artificial intelligence (AI) styles to assist in the visualization and measurement of COVID unique lesions in the lungs of contaminated individuals, probably facilitating to far more timely and client-unique professional medical interventions.

Attracting far more than one,000 world wide members, the level of competition was introduced by the Sheikh Zayed Institute for Pediatric Surgical Innovation at Children’s National Healthcare facility in collaboration with leading AI technologies corporation NVIDIA and the National Institutes of Well being (NIH). The competition’s AI styles used a multi-institutional, multi-countrywide info established provided by community datasets from The Most cancers Imaging Archive (National Most cancers Institute), NIH and the College of Arkansas, that originated from individuals of unique ages, genders and with variable sickness severity. NVIDIA provided GPUs to the top five winners as prizes, as very well as supported the assortment and judging system.

“Improving COVID-19 cure commences with a clearer comprehension of the patient’s sickness state. Having said that, a prior deficiency of world wide info collaboration minimal clinicians in their skill to promptly and properly realize sickness severity throughout both of those adult and pediatric individuals,” claims Marius George Linguraru, D.Phil., M.A., 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 by quantitative imaging and equipment mastering, these discoveries are supporting clinicians better realize COVID-19 sickness severity and probably stratify and triage into appropriate cure protocols at unique levels of the sickness.”

The top 10 AI algorithms have been recognized from a really aggressive discipline of members who analyzed the info in November and December 2020. The final results have been unveiled on Jan. 11, 2021, in a digital symposium, hosted by Children’s National, that featured presentations from top teams, event organizers and clinicians. 

Developers of the 10 top AI styles from the COVID-19 Lung CT Lesion Segmentation Grand Challenge are

  1. Shishuai Hu, et al. Northwestern Polytechnical College, China. “Semi-supervised Strategy for COVID-19 Lung CT Lesion Segmentation”
  2. Fabian Isensee, et al. German Most cancers Research Heart, Germany. “nnU-Net for Covid Segmentation”
  3. Claire Tang, Lynbrook Higher School, United states. “Automated Ensemble Modeling for COVID-19 CT Lesion Segmentation”
  4. Qinji Yu, et al. Shanghai JiaoTong College, China. “COVID-19-twenty Lesion Segmentation Based on nnUNet”
  5. Andreas Husch, et al. College of Luxembourg, Luxembourg. “Leveraging Point out-of-the-Artwork Architectures by Enriching Training Information and facts – a situation study”
  6. Tong Zheng, et al. Nagoya College, Japan. “Fully-automatic COVID-19-twenty Segmentation”
  7. Vitali Liauchuk. United Institute of Informatics Complications (UIIP), Belarus. “Semi-3D CNN with ImageNet Pretrain for Segmentation of COVID Lesions on CT”
  8. Ziqi Zhou, et al. Shenzhen College, China. “Automated Upper body CT Image Segmentation of COVID-19 with 3D Unet-dependent Framework”
  9. Jan Hendrik Moltz, et al. Fraunhofer Institute for Electronic Medicine MEVIS, Germany. “Segmentation of COVID-19 Lung Lesions in CT Employing nnU-Net”
  10. Bruno Oliveira, et al. 2Ai – Polytechnic Institute of Cávado and Ave, Portugal. “Automatic COVID-19 Detection and Segmentation from Lung Computed Tomography (CT) Visuals Employing 3D Cascade U-net”

Linguraru additional that, in addition to an award for the top five AI styles, these winning algorithms are now obtainable to lover with clinical establishments throughout the world to further assess how these quantitative imaging and equipment mastering approaches may probably impression world wide community health and fitness.

“Quality annotations are a restricting variable in the growth of beneficial AI styles,” said Mona Flores, M.D., world wide head of Healthcare AI, NVIDIA. “Using the NVIDIA COVID lesion segmentation model obtainable on our NGC software package hub, we have been in a position to promptly label the NIH dataset, letting radiologists to do specific annotations in report time.”

“I applaud the laptop or computer science, info science and image processing world wide educational group for swiftly teaming up to combine multi-disciplinary experience towards growth of probable automatic and multi-parametric applications to better examine and deal with the myriad of unmet clinical requirements created by the pandemic,” said Bradford Wooden, M.D., director, NIH Heart for Interventional Oncology and main, Interventional Radiology Segment, NIH Clinical Heart. “Thank you to each individual team for locking arms towards a typical trigger that unites the scientific group in these tough occasions.” 

Media Make contact with:  Diana Troese | 202-476-4500

Previous post Can a mother’s stress impact children’s disease development? Environmental health researcher says there is a connection between trauma and DNA mutation — ScienceDaily
Next post Vitamin D levels during pregnancy linked with child IQ — ScienceDaily