Medical Imaging with Deep Learning Conference

Kitware is pleased to have our paper accepted at MIDL 2022. The conference brings the deep learning community together with medical imaging researchers, clinicians, and healthcare organizations. This will provide us the opportunity to share our latest advancements in using AI for medical computing.

Medical Image Quality Assurance using Deep Learning

Wednesday, July 6 from 11:00 am – 12:00 pm (onsite) or 15:20 – 16:20 pm (virtual)

Authors: Dženan Zukić (Kitware), Anne Haley (Kitware), Curtis Lisle (KnowledgeVis LLC), James Klo (SRI International), Kilian M. Pohl (SRI International), Hans J Johnson (University of Iowa), Aashish Chaudhary (Kitware)

In this paper, we present an open source application, Medical Image Quality Assurance (MIQA), for quality control of distributed imaging studies. To minimize the amount of human time and attention spent reviewing the images, we created a neural network to provide an automatic assessment. Reviewer’s are directed to potentially problematic cases, reducing the likelihood of overlooking image quality issues while saving them time. We tested our approach using 5-fold cross-validation on a set of 5,217 magnetic resonance images. [Link to Paper]

Slices of two images: The top has an overall score of 8 and no artifacts present. The bottom has a score of 6 with inhomogeneity, flow, and truncation artifacts. PREDICTHD images are defaced.

Online and Physical Event

ETH Zurich
Rämistrasse 101, 8092 Zürich, Switzerland
ZürichVirtual Location