Research Projects

AI-Enabled Obstetric Ultrasound in Rural Areas

Development of automated fetal and placenta localization, fetal biometry, and gestational age estimation using Volume Sweep Imaging (VSI) and artificial intelligence. The goal is to provide reliable prenatal monitoring in underserved regions without requiring trained sonographers or radiologists. AI-based image quality transformation improves model transferability between devices and environments.

Automated Breast Cancer Detection with VSI and AI

Integration of 3D segmentation and classification models for breast ultrasound (VSI-B), enabling early diagnosis of breast lesions (benign vs malignant) in low-resource settings. This pipeline eliminates reliance on radiologists, achieving high sensitivity and specificity.

Thyroid VSI Segmentation and Quality Adaptation

Extending VSI-based ultrasound imaging to thyroid evaluations, focusing on segmentation robustness across variable image qualities. After segmentation, measurement of thyroid nodules for clinical assessment based on thresholds.

Liver VSI and Quantitative Ultrasound (QUS)

Applying H-scan quantitative ultrasound to liver VSI acquisitions for scatterer size characterization. Current efforts include replication of liver phantom studies, diabetic foot analysis, and the planned extension toward liver disease assessment.

Lung VSI for Pneumonia Detection

Extending VSI-based ultrasound imaging to lung evaluations, using AI-based segmentations to detect and focus A and B-lines counters for pneumonia assessment in point of care ultrasound devices as Butterfly IQ+.

Computer Vision for Fetal Heart Rate

Use of optical flow techniques to track fetal cardiac motion in noisy point-of-care ultrasound (POCUS) devices, yielding fetal heart rate estimations with accuracy comparable to physician measurements.

Robotic Ultrasound Acquisition

Integration of robotic systems for standardized probe sweeps, reducing operator variability and enhancing reproducibility of VSI protocols.

Elastography in VSI Frameworks

Incorporation of shear wave and strain elastography into VSI-based acquisitions, with the aim of quantifying viscoelastic properties of tissues in both obstetric and abdominal applications.

Volumetric Reconstruction with OptiTrack and Virtual Reality

Development of 3D ultrasound reconstructions from VSI using motion-tracked probe trajectories (OptiTrack). Reconstructions are displayed and manipulated in virtual reality environments, opening avenues for interactive training and remote diagnostics.