Advanced 3D ultrasound volume reconstruction using 1D linear array

In this work, we aim at developing a deep learning-based approach to improve the quality of 3D volume reconstruction from sequences of beamformed ultrasound data acquired using a conventional 1D linear array transducer. The proposed method reconstructs high-fidelity 3D volumetric images from sparse data, effectively preserving structural details and overcoming the limitations of traditional interpolation techniques. The study offers a flexible and effective solution to replace 2D matrix and row-column-addressed (RCA) arrays, which are limited by their complex and expensive hardware. Experiments are conducted on the datasets collected from the multi-purpose and vascular phantoms using a robotic ultrasound system to precisely control the transducer’s positions. The experimental results demonstrate the superior performance of the proposed approach compared to other interpolation techniques in both quantitative and qualitative analysis.