Ultrasound imaging is commonly used for aiding diseasediagnosis and treatment, with advantages in noninvasive. Lately, medicalultrasound shows prospects revolving from expensive big-size machines inhospitals to economical hand-held devices in wider use. The barrier is thatultrasound examination with a handheld device has the drawback of low imagingquality due to hardware limitations. Toward this, ultrasound image enhancementprovides a potential low-cost solution. Restoring high-quality images fromlow-quality ones using computer algorithms would exempt requirements forhardware improvements and promote ultrasound device revolutions and widerapplications.
We propose to hold the challenge of enhancement for ultrasound images in conjunction with MICCAI 2023. We will provide various ultrasound data of five organs, including the thyroid, carotid artery, liver, breast, and kidney. The challenging task is reconstructing high-quality ultrasound images from low-quality ones. A total of 3000 ultrasound images (1500 pairs of low- and high-quality images) from 109 patients will be provided in the challenge.
References:
1. Zhou Z,Guo Y, Wang, Y. Handheld ultrasound video high-quality reconstruction using alow-rank representation multipathway generative adversarial network. IEEETransactions on Neural Networks and Learning Systems, 2021, 32(2): 575-588.
2. Zhou Z,Wang Y, Guo Y, Qi Y, Yu J. Image quality improvement of hand-held ultrasounddevices with a two-stage generative adversarial network. IEEE Transactions onBiomedical Engineering, 2020, 67(1): 298-311.
3. Xia M, Yang H, Qu Y, Guo Y, Zhou G, Zhang F, Wang Y. Multilevel structure-preserved GAN for domain adaptation in intravascular ultrasound analysis. Medical Image Analysis, 2022, 82: 102614.
4. Huang L, Zhou Z,Guo Y, Wang Y. A stability-enhanced CycleGAN for effective domaintransformation of unpaired ultrasound images. Biomedical Signal Processing andControl, 2022, 77: 103831.