MR - CT Conversation using Cycle Generative Adversarial Networks evrimsel retici ekismeli Aglar ile MR-BT D n s m

Cuskun Y., Alparslan B., Ertunç H. M.

33rd IEEE Conference on Signal Processing and Communications Applications, SIU 2025, İstanbul, Turkey, 25 - 28 June 2025, (Full Text) identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/siu66497.2025.11111836
  • City: İstanbul
  • Country: Turkey
  • Keywords: CycleGAN, Generative Adversarial Networks (GAN), Medical image translation, MR - CT translation
  • Kocaeli University Affiliated: Yes

Abstract

Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) are widely used medical imaging modalities that provide complementary information for visualizing different tissues. In this study, a Cycle Generative Adversarial Network (CycleGAN)-based model is proposed to perform bidirectional conversion between MRI and CT images. CycleGAN is a type of Generative Adversarial Network (GAN) that enables image-to-image translation without requiring paired data. For training the model, a dataset consisting of 1000 paired brain MRI-CT images was used, and the model was evaluated using 21 image pairs during the testing phase. The numerical performance of the model was analyzed using metrics such as Structural Similarity Index Measure (SSIM), Peak Signal-to-Noise Ratio (PSNR), Mean Absolute Error (MAE), and Universal Quality Index (UQI), while the visual results were assessed by a radiologist. The obtained results indicate that the proposed model successfully performs transformation between MRI and CT modalities. This approach presents significant potential for modality synthesis in the field of medical imaging.