The brain magnetic resonance (MR) images can present the anatomical information noninvasively for more accurate diagnosis and exploration of the brain. However, it is challenging to acquire adequate resolution medical image clinically.
The super-resolution technology, which provides an economic and efficient approach for reconstructing high-resolution images from the low-resolution counterparts, has attracted lots of attention.
Recently, a research team led by Prof. WANG Shuqiang from the Shenzhen Institute of Advanced Technology (SIAT) of the Chinese Academy of Sciences proposed a fine perceptive generative adversarial network (FP-GANs) that enables super-resolution brain magnetic resonance images.
The study was published in IEEE Transaction of Neural Network and Learning Systems.
The divide-and-conquer method can divide the MR images into low-frequency and high-frequency components with discrete wavelet transformation and super-resolve the sub-band images independently with four GANs. Benefiting from the method, the sensitivity of the proposed model to the fine anatomical structures of the brain is enhanced.
This strategy simplifies the SR task for one single GAN and hence accelerates and stabilizes the training procedure. Furthermore, a novel attention that can adaptively trade off the weight among the sub-band images was proposed to enhance the anatomical textures.
Comprehensive experiments on the MultiRes_7T and ADNI datasets demonstrated that the proposed model achieved finer structure recovery and outperformed the competing methods quantitatively and qualitatively. Moreover, FP-GANs further showed the value by applying the SR results in classification tasks.
