Projects

Fetal brain ultrasound–MRI registration

Atlas-assisted multi-task learning for automatic alignment of three-dimensional fetal brain ultrasound and MRI. The framework combines cross-modality registration with modality-specific image-to-atlas tasks and was evaluated using same-day ultrasound–MRI pairs.

NeuroImage paper (2025)

Three-dimensional shear-wave field shown on intersecting planes.

Volumetric ultrasound elastography

Three-dimensional shear wave absolute vibro-elastography (S-WAVE) with a matrix array transducer, including imaging-system implementation and validation for liver imaging.

Six liver MRI examples comparing ground-truth contours in green, V-net in red, and the presented method in blue.

Liver MRI segmentation

Learning-based liver segmentation using a mean-shape prior and a dense deformation field.

Three-dimensional anatomical model showing ribs and the liver region used for ultrasound–MRI alignment.

Ultrasound–MRI registration

Anatomical segmentation and a rib model support initial alignment, followed by registration of corresponding anatomical features across ultrasound and MRI.

Prostate segmentation with MRI priors

Segmentation of transrectal ultrasound images using a convolutional neural network for the mid-gland and an MRI-derived statistical shape model to guide reconstruction at the base and apex.

This work was presented in my IPCAI talk, Prostate Segmentation in Transrectal Ultrasound Using Magnetic Resonance Imaging Priors.