Deep Learning based Multi-modal Ultrasound-Photoacoustic Imaging

This study explores the implementation of an artificial intelligence (AI)-assisted multi-modal imaging platform for renal tissue analysis. Leveraging contrast-enhanced ultrasound (CEUS) and photoacoustic (PA) imaging, it provides a comprehensive view of renal tissues. Using deep learning (DL) models like U-Net and nnU-Net, anatomical structures are accurately segmented in medical ultrasound images. Evaluation metrics confirm the effectiveness of DL models. Functional imaging analysis correlates DL predictions with non-linear constrast (NLC) images to understand renal tissue perfusion dynamics. Future work involves using DL predictions for fluence correction in PA images, enhancing tissue absorption accuracy. This multi-modal approach has potential in clinical diagnostics and disease monitoring.