GAN-Based ROI Image Translation Method for Predicting Post-Hair Transplant Surgery Images
December 2021
in “
Electronics
”
TLDR The new method predicts post-hair transplant images more accurately than other methods.
The study proposed a novel GAN-based ROI image translation method to predict post-hair transplant surgery images from pre-surgery images, focusing on converting only the region of interest (ROI) while retaining other image areas. By independently performing image translation and segmentation, and using an ensemble method to enhance segmentation, the method improved predictive image generation. The experimental results, using 1,394 pre-transplant and 896 post-transplant images, demonstrated that this method outperformed other GAN-based methods, with improvements of 23% in SSIM, 452% in IoU, and 42% in FID. The ensemble approach also enhanced ROI detection and maintained consistent performance across diverse image angles.