Breast Cancer Screening

Low-cost, quantitative, high-resolution, and ionizing-radiation-free imaging of the whole breast based on ultrasound computed tomography

Ultrasound Computed Tomography for Breast Cancer Screening

Ultrasound computed tomography (USCT) provides a different approach to breast imaging than conventional pulse-echo ultrasound. Rather than relying only on backscattered echoes, a ring of transducers records ultrasound transmitted through the breast as well as reflections generated within the tissue. These measurements provide complementary information about the acoustic properties of the breast and enable quantitative image reconstruction. The goal is to develop a low-cost, radiation-free alternative for whole-breast imaging that combines the quantitative information of tomography with the spatial resolution of modern wave-equation imaging.

Ring-based UST/USCT records transmissions through and reflections from tissue. In-vivo comparison of FWI-based UST to MRI for breast imaging.

Quantitative Breast Imaging with Full-Waveform Inversion

Full-waveform inversion (FWI) reconstructs tissue properties by matching the measured ultrasound waveforms to waveforms simulated through a model of the breast. In transmission, sound speed primarily controls the phase and arrival time of the waveform, while attenuation controls its amplitude. By explicitly modeling wave propagation, FWI can use these effects to reconstruct quantitative maps of sound speed and attenuation rather than producing only a qualitative echo image. This approach enables the reconstruction of breast anatomy directly from the acoustic properties of tissue. The resulting images can reveal lesions through changes in sound speed and attenuation while retaining the quantitative character of tomographic imaging. Numerical and experimental studies demonstrate that FWI can recover these properties in breast phantoms, including both benign and malignant lesions (Ali et al., 2024).

FWI Reconstruction of Sound Speed and Attenuation in the Breast (Ali et al., 2024). (Top) Benign Cyst. (Bottom) Malignancy.

From Screening to Diagnosis

The resolution of FWI reconstructions can be increased by incorporating progressively higher frequencies. Lower frequencies provide more robust recovery of the large-scale sound-speed structure, while higher frequencies provide finer spatial detail. This creates a natural tradeoff between image resolution and computational cost: a screening scan can first identify suspicious regions, after which additional high-frequency reconstruction can be used to characterize those regions in greater detail. This suggests a workflow in which USCT serves not only as a screening modality, but also as a quantitative tool for investigating suspicious findings. Rather than acquiring a single image optimized for every diagnostic task, the reconstruction can be adapted to the clinical question by varying the frequency content and computational effort.

Screening to Diagnosis. In each case, the spatial resolution of the FWI sound speed reconstruction increases with frequency but may requires longer compute times to confirm suspected cancers.

Comparison with MRI

The quantitative nature of FWI makes USCT particularly interesting as a potential alternative to MRI for breast imaging. MRI provides excellent soft-tissue contrast but requires expensive equipment. USCT uses non-ionizing ultrasound and can acquire whole-breast measurements using a surrounding transducer array. The goal is not necessarily to replace MRI but to reduce the clinical burden on MRI by enabling ultrasound to answer the same fundamental clinical questions. The FWI reconstructions below show that quantitative sound-speed imaging can produce structural and contrast information that is visually comparable to T1-weighted breast MRI.

UST/USCT FWI Sound Speed Reconstruction vs. Contrast-Enhanced (T1-Weighted) MRI for Breast Imaging.

Detecting Mammographically Occult Lesions

Mammographically Occult Breast Cancer. (Left) Mammography. (Right) Cancer visualized on B-mode, sound speed, and attenuation.

A particularly important application is the detection of cancers that are difficult to identify with mammography. Dense breast tissue and other imaging limitations can obscure lesions in conventional mammographic images. Because USCT provides complementary measurements and reconstructs quantitative acoustic properties, lesions that are inconspicuous on mammography may become visible in sound-speed, attenuation, or reflection images. The examples above and below illustrate this complementary information across multiple breast lesions, including invasive ductal carcinomas and cases in which the cancer is occult on mammography. The combination of FWI-derived quantitative images with conventional reflection imaging provides multiple signatures with which to identify and characterize suspicious tissue.

Three Case Examples of Breast Cancers Detected using UST. (Top Row) Fibroadenoma. (Middle Row) Invasive Ductal Carcinoma Visible on Mammogram. (Bottom Row) Invasive Ductal Carcinoma Occult on Mammogram. (Left Column) Mammography. (Middle Column) FWI sound speed. (Right Column) B-mode reflection.

Toward Quantitative Whole-Breast Imaging and Screening

The broader objective is to combine the advantages of ultrasound tomography, full-waveform inversion, and modern computational imaging into a practical breast-screening system. Transmission and reflection measurements provide complementary sensitivity to tissue properties, while FWI allows those measurements to be interpreted through a physics-based model of wave propagation. The resulting framework moves breast ultrasound from qualitative detection toward quantitative tissue characterization: instead of asking only whether a structure reflects ultrasound, the reconstruction asks how its acoustic properties differ from surrounding tissue. Improving the accuracy, resolution, and computational efficiency of these reconstructions could ultimately enable USCT to provide a radiation-free and quantitative approach to whole-breast screening and follow-up diagnosis. Future direction for whole-breast USCT include 3D volumetric FWI and USCT acquisition systems (Ali et al., 2024), and data efficiency for rapid screening via transmit-receive downsampling (Nketia et al., 2026).

References

2026

Optimizing computational efficiency and image quality in clinical breast ultrasound tomography through data downsampling

Victoria Nketia, Trevor Mitcham, Maria Camila Gonzalez Garcia, and 4 more authors

In Medical Imaging 2026: Ultrasonic Imaging and Tomography, 2026

Abs DOI Bib PDF

Ultrasound tomography (UST) offers quantitative, high-resolution breast imaging, but widespread clinical adoption remains limited by long acquisition times, large raw data volumes, and high computational demands for image reconstruction. Some current systems employ a 1024-element ring transducer array, producing more than 2 GB of raw data per slice and nearly 200 GB for a full breast volume, resulting in considerable memory and storage requirements, hardware, and processing burdens. This study systematically investigates downsampling as a practical strategy to reduce data size, computational load, and reconstruction time while preserving diagnostic image quality. Using clinical UST data from a ring-array system, raw data were downsampled from 1024×1024 to 512×512 and 256×256, with additional transmit-only reductions applied by factors of 4, 8, 16, and 32. Images were reconstructed using a Full Waveform Inversion (FWI) algorithm, and the quality of the reconstructed images was evaluated across consistent anatomical Regions of Interest (ROIs). Quantitative metrics included Signal-to-Noise Ratio (SNR), Contrast-to-Noise Ratio (CNR), and Root Mean Square Error (RMSE). Results show that downsampling reduces reconstruction time by up to a factor of three while decreasing memory size. Although tumor SNR decreased with extreme downsampling, CNR remained largely comparable across many configurations, indicating preserved tumor detectability. The 64×256 configuration provided an optimal balance, offering strong CNR performance with a reconstruction 3 times faster compared to the 3.0 minutes of the full dataset. These findings demonstrate that strategic downsampling enables substantial efficiency gains with minimal loss of clinically relevant image quality, supporting the development of lower-cost, UST systems suitable for clinical use.
@inproceedings{Nketia2026Downsampling, title = {Optimizing computational efficiency and image quality in clinical breast ultrasound tomography through data downsampling}, author = {Nketia, Victoria and Mitcham, Trevor and Garcia, Maria Camila Gonzalez and Ali, Rehman and Owolabi, Israel and Singh, Melanie and Duric, Neb}, booktitle = {Medical Imaging 2026: Ultrasonic Imaging and Tomography}, volume = {13931}, pages = {118--133}, year = {2026}, organization = {SPIE}, doi = {10.1117/12.3082370}, url = {https://www.spiedigitallibrary.org/conference-proceedings-of-spie/13931/3082370/Optimizing-computational-efficiency-and-image-quality-in-clinical-breast-ultrasound/10.1117/12.3082370.full} }

2024

2-D Slicewise Waveform Inversion of Sound Speed and Acoustic Attenuation for Ring Array Ultrasound Tomography Based on a Block LU Solver

Rehman Ali, Trevor M Mitcham, Thurston Brevett, and 5 more authors

IEEE transactions on medical imaging, 2024

Abs DOI Bib PDF Code

Ultrasound tomography is an emerging imaging modality that uses the transmission of ultrasound through tissue to reconstruct images of its mechanical properties. Initially, ray-based methods were used to reconstruct these images, but their inability to account for diffraction often resulted in poor resolution. Waveform inversion overcame this limitation, providing high-resolution images of the tissue. Most clinical implementations, often directed at breast cancer imaging, currently rely on a frequency-domain waveform inversion to reduce computation time. For ring arrays, ray tomography was long considered a necessary step prior to waveform inversion in order to avoid cycle skipping. However, in this paper, we demonstrate that frequency-domain waveform inversion can reliably reconstruct high-resolution images of sound speed and attenuation without relying on ray tomography to provide an initial model. We provide a detailed description of our frequency-domain waveform inversion algorithm with open-source code and data that we make publicly available.
@article{Ali2024_BlockLU_2DFWI, title = {2-D Slicewise Waveform Inversion of Sound Speed and Acoustic Attenuation for Ring Array Ultrasound Tomography Based on a Block LU Solver}, author = {Ali, Rehman and Mitcham, Trevor M and Brevett, Thurston and Agudo, {\`O}scar Calder{\'o}n and Martinez, Cristina Dur{\'a}n and Li, Cuiping and Doyley, Marvin M and Duric, Nebojsa}, journal = {IEEE transactions on medical imaging}, volume = {43}, number = {8}, pages = {2988--3000}, year = {2024}, doi = {10.1109/TMI.2024.3383816}, publisher = {IEEE}, url = {https://ieeexplore.ieee.org/document/10486959}, }