Radiography articles within Nature Communications

Featured

• Article
  06 October 2023 | Open Access

  Improving model fairness in image-based computer-aided diagnosis

  Deep learning models can reflect and amplify human bias, potentially resulting inaccurate missed diagnoses. Here, the authors show that by leveraging the marginal pairwise equal opportunity, their model reduces bias in medical image classification by over 35% compared to baseline models, with minimal impact on AUC values.

  • Mingquan Lin
  • , Tianhao Li
  •  & Yifan Peng

• Article
  04 July 2022 | Open Access

  Self-evolving vision transformer for chest X-ray diagnosis through knowledge distillation

  Although deep learning-based computer-aided diagnosis systems have recently achieved expert level performance, developing a robust model requires large, high-quality data with annotations. Here, the authors present a framework which can improve the performance of vision transformer simultaneously with self-supervision and self-training.

  • Sangjoon Park
  • , Gwanghyun Kim
  •  & Jong Chul Ye

• Article
  06 April 2022 | Open Access

  Accurate auto-labeling of chest X-ray images based on quantitative similarity to an explainable AI model

  Here the authors develop a method for accurate auto-labelling of CXR images from large public datasets based on quantitative probability-of similarity to an explainable AI model. The labels can be used to fine-tune the original model through iterative re-training.

  • Doyun Kim
  • , Joowon Chung
  •  & Synho Do

• Article
  04 March 2022 | Open Access

  Active label cleaning for improved dataset quality under resource constraints

  High quality labels are important for model performance, evaluation and selection in medical imaging. As manual labelling is time-consuming and costly, the authors explore and benchmark various resource-effective methods for improving dataset quality.

  • Mélanie Bernhardt
  • , Daniel C. Castro
  •  & Ozan Oktay

• Article
  14 December 2021 | Open Access

  A machine and human reader study on AI diagnosis model safety under attacks of adversarial images

  While active efforts are advancing medical AI model development and clinical translation, safety issues of medical AI models have emerged. Here, the authors investigate the effects on an AI model and on human experts of potential fake/adversarial images for breast cancer diagnosis.

  • Qianwei Zhou
  • , Margarita Zuley
  •  & Shandong Wu

• Article
  16 September 2021 | Open Access

  Automated bone mineral density prediction and fracture risk assessment using plain radiographs via deep learning

  Dual-energy X-ray absorptiometry and the Fracture Risk Assessment Tool are recommended tools for osteoporotic fracture risk evaluation, but are underutilized. Here, the authors present an opportunistic tool to identify fractures, predict bone mineral density and evaluate fracture risk using plain pelvis and lumbar spine radiographs.

  • Chen-I Hsieh
  • , Kang Zheng
  •  & Chang-Fu Kuo

• Article
  16 February 2021 | Open Access

  A scalable physician-level deep learning algorithm detects universal trauma on pelvic radiographs

  Pelvic radiographs (PXRs) are essential for detecting proximal femur and pelvis injuries in trauma patients, but none of the currently available algorithms can detect all kinds of trauma-related radiographic findings. Here, the authors develop a multiscale deep learning algorithm trained with weakly supervised point annotation.

  • Chi-Tung Cheng
  • , Yirui Wang
  •  & Le Lu