Deep Learning Super-Resolution Technique Based on Magnetic Resonance Imaging for Application of Image-Guided Diagnosis and Surgery of Trigeminal Neuralgia

Abstract

This study aimed to implement a deep learning-based super-resolution (SR) technique that can assist in the diagnosis and surgery of trigeminal neuralgia (TN) using magnetic resonance imaging (MRI). Experimental methods applied SR to MRI data examined using five techniques, including T2-weighted imaging (T2WI), T1-weighted imaging (T1WI), contrast-enhancement T1WI (CE-T1WI), T2WI turbo spin-echo series volume isotropic turbo spin-echo acquisition (VISTA), and proton density (PD), in patients diagnosed with TN. The image quality was evaluated using the peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). High-quality reconstructed MRI images were assessed using the Leksell coordinate system in gamma knife radiosurgery (GKRS). The results showed that the PSNR and SSIM values achieved by SR were higher than those obtained by image postprocessing techniques, and the coordinates of the images reconstructed in the gamma plan showed no differences from those of the original images. Consequently, SR demonstrated remarkable effects in improving the image quality without discrepancies in the coordinate system, confirming its potential as a useful tool for the diagnosis and surgery of TN.

Keywords: artificial intelligence (AI); deep learning; magnetic resonance imaging (MRI); super resolution (SR); trigeminal neuralgia (TN).

Figure 1

(a–e) Trigeminal nerve images (a): T2WI, (b): T1WI, (c): CE-T1WI, (d): VISTA, and (e): PD.

Figure 2

This Figure shows the proposed VDSR network for improving image quality. This illustrates the series of processes through which the input image passes through the convolution and residual layers, eventually producing a high-resolution image based on SR by summing it with the residual image.

Figure 3

An example of the code used to augment the training dataset.

Figure 4

The process of obtaining high-resolution images through SR using the VISTA’s GT, low-resolution images, and residual images. (a) Represents the GT, (b) the low-resolution image, (c) the residual image, (d) the high-resolution image based on SR, and (e) an enlarged view for comparison of the trigeminal nerve between (b) (left) and (d) (right).

Figure 5

The coordinates system when high-resolution images based on SR were applied to the gamma plan. (a) Represents the GT and (b) shows a high-resolution image based on the SR. The yellow circle indicates a 4 mm shot targeted at the trigeminal nerve, and the X-, Y-, and Z-axis coordinates in (b) match those in (a).

Figure 6

This Figure shows the texture components of the MRI data reconstructed using SR compared with those reconstructed using GT. (a) Represents the GT and (b) displays a high-resolution image based on the SR. This comparison suggests that there were no significant changes in texture composition when the MRI data were reconstructed using SR.