Recently the Organization Committee of MICCAI2020 (International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI) released the list of accepted papers. One paper written by the research group led by Professor Zhang Xiaohong of the School of Big Data and Software Engineering is among those accepted.

MICCAI is organized by Medical Image Computing and Computer Assisted Intervention Society and is a top comprehensive academic conference on MIC and CAI. Despite the pandemic, the number of papers received by the Organization Committee this year is no less than that received in previous years. The number of papers hit a historic high in 2019, and that in 2020 remained at the same level with that in last year. MICCAI, which is planned to be held from October 4 to 8, 2020 in Lima, the capital of Peru, is expected to take the virtual form for the first time.

The paper submitted by Professor Zhang’s research group is titled “Class-Aware Multi-Window Adversarial Lung Nodule Synthesis Conditioned on Semantic Features”. This paper proposes a data enhancement method for benign and malignant pulmonary nodules based on antagonistic learning. As shown in Figure 1, different from the existing generation confrontation network model, this model integrates the clinical observation experience of pulmonary nodules for the first time, and uses different CT windows to conduct targeted learning on medical signs of pulmonary nodules, so that the model can not only control the CT window of the generated image, but also control the performance of medical signs in different windows. As such, model can be used for the data of pulmonary nodules. The enhancement is more specific and improves the ability to control the medical signs of pulmonary nodules (Fig. 2).

Fig. 1 Pulmonary nodule generation network model integrated with multi window and multi signs

Fig. 2 Impact of benign and malignant settings on imaging medical signs of pulmonary nodules

Wang Qiuli, a doctoral candidate and a member of the research group led by Professor Zhang Xiaohong, is the first author of the paper. Zhang Xiaohong is the corresponding author. Chongqing University is the first completing author. The research project has been supported by the National Natural Science Foundation of China, the key research and development program of China and the special funds for major themed projects allocated by the government of Chongqing.