Highlight

Prostate cancer is among the most prevalent cancers in men and a leading cause of cancer-related deaths globally. Early detection is critical for effective treatment, yet conventional prostate ultrasound techniques suffer from limited sensitivity and specificity. Ultrasound tomography (UST) offers a potential solution by quantifying parameters such as speed of sound (SOS), but achieving high image quality is challenging due to the narrow data acquisition aperture imposed by the prostate's anatomy. This study explores the use of a convolutional neural network (CNN) to enhance prostate imaging under these constraints. Our approach leverages supervised learning to manage the extremely narrow data acquisition apertures that hinder traditional full-waveform inversion (FWI) methods. We validated our method using synthetic prostate phantoms and finite difference ultrasound data simulations. The results demonstrate that our CNN-based approach surpasses traditional FWI in both accuracy and efficiency, reconstructing high-resolution SOS maps swiftly once the model is trained. This advancement promises significant improvements in the accuracy and efficiency of prostate cancer diagnosis and treatment planning, potentially reducing the need for invasive procedures and enabling better-informed clinical decisions.

The full paper can be found on SPIE-Medical Imaging:

Mask-enhanced Deep-learning for Prostate Ultrasound Tomography with Narrow Data Acquisition Aperture