Photoacoustic tomography (PAT) is an emerging biomedical imaging technique that combines the advantages of optical and ultrasonic imaging. It relies on the photoacoustic effect, where short laser pulses cause localized heating in biological tissues, leading to thermoelastic expansion and generation of ultrasound waves. These waves are captured by an array of ultrasonic transducers and then reconstructed into images using algorithms such as time reversal (TR), filtered back projection (FBP), or model-based iterative methods. PAT offers deeper tissue penetration and higher resolution than many other imaging modalities, making it particularly suitable for clinical and biological applications. However, practical limitations in transducer array design, such as the spatial impulse response (SIR) and electrical impulse response (EIR), can introduce significant artifacts and blurring into the reconstructed images, especially in systems using a ring-array configuration or having an insufficient number of transducers.
To address these challenges, researchers have developed a variety of image restoration and deconvolution methods. While spatially invariant blur caused by EIR can often be corrected using traditional deconvolution techniques, spatially variant blur due to the geometry of the transducer array is more complex and difficult to eliminate. Furthermore, streak artifacts, often resulting from limited transducer coverage, are commonly mitigated using regularization techniques like total variation or specially designed weighting functions. Despite these efforts, the quality of restored images remains inconsistent, and the computational cost can be high. As such, there is a growing interest in using deep learning for PAT image restoration, particularly through image-to-image post-processing methods, which are more practical and effective than mapping raw signals directly to final images.
In this work, a novel deep learning-based image restoration approach is proposed, utilizing a conditional generative adversarial network (CGAN) architecture enhanced with attention mechanisms. The generator integrates a Residual Shifted Window Transformer Module (RSTM) and is augmented with spatial attention, channel attention, and gamma correction to better capture image features and compensate for degradation. The PatchGAN discriminator is refined using adversarial training to encourage realistic restorations. A comprehensive loss function incorporating adversarial, pixel-level, and feature-level content loss guides the training. The network is trained on a simulated dataset created using the k-Wave toolbox, which includes degraded and corresponding clear PAT images. Experimental results show that the proposed method outperforms existing state-of-the-art techniques by significantly enhancing image clarity, resolving fine structures, and effectively removing artifacts, demonstrating its potential for real-world clinical applications.
International Research Awards on Computer Vision
The International Research Awards on Computer Vision recognize groundbreaking contributions in the field of computer vision, honoring researchers, scientists and innovators whose work has significantly advanced the domain. This prestigious award highlights excellence in fundamental theories, novel algorithms and real-world applications, fostering progress in artificial intelligence, image processing and deep learning.
Visit Our Website : computer.scifat.com
Nominate now : https://computer-vision-conferences.scifat.com/award-nomination/?ecategory=Awards&rcategory=Awardee
Contact us : computersupport@scifat.com
#researchawards #shorts #technology #researchers #conference #awards #professors #teachers #lecturers #biologybiologiest #OpenCV #ArtificialIntelligence #MachineLearning #DeepLearning #NeuralNetworks #DataScience #physicist #coordinator #business #genetics #medicirne #bestreseracher #bestpape
Get Connected Here:
==================
Twitter : x.com/sarkar23498
Youtube : youtube.com/channel/UCUytaCzHX00QdGbrFvHv8zA
Pinterest : pinterest.com/computervision69/
Instagram : instagram.com/saisha.leo/?next=%2F
Tumblr : tumblr.com/blog/computer-vision-research
No comments:
Post a Comment