NVIDIA infrastructure accelerates AI apps targeting breast cancer care gaps

As the U.S. healthcare system faces a projected shortfall of tens of thousands of radiologists and a majority of women over age 40 skip annual screenings, a wave of artificial intelligence applications accelerated by NVIDIA infrastructure is targeting critical gaps across breast cancer care, from early imaging and risk assessment to treatment planning.

Automated 3D Ultrasound Solves Access and Repeatability Gaps

Approximately 40 million mammograms are performed annually in the U.S., but logistical hurdles keep many women over age 40 from maintaining regular checkups. To bypass the practical barriers of scheduling and clinic location, NVIDIA Inception startup iSono Health deployed its FDA-cleared ATUSA platform.

The wearable, automated 3D quantitative ultrasound system captures a standardized whole-breast volume in roughly two minutes per breast. Conventional handheld ultrasounds can take up to 45 minutes and rely heavily on operator technique.

“Getting the scan closer to the patient is the first breakthrough,” said Neda Razavi, CEO of iSono Health. “Our vision is to make that scan increasingly informative: helping clinicians see what is there, understand what has changed and make more informed decisions.”

Trained on thousands of full-breast scans comprising more than 1.5 million ultrasound frames, the ATUSA system uses NVIDIA GPU acceleration and open-source medical imaging technology. The resulting 3D scan yields a 28% higher sensitivity than a handheld 2D ultrasound, according to the company. Because the hardware captures the tissue uniformly every time, clinicians can track changes across successive scans while minimizing operator-induced variability.

Commercial availability spans partner clinics across California, Texas, Georgia, Tennessee, and Washington D.C., with additional sites deploying the system regularly. A multicenter clinical study involving 3,200 patients is currently underway, anchored by lead research sites at UC Davis and Vanderbilt University Medical Center.

Algorithms Clear the Radiologist Haystack to Reduce False Alarms

Radiologists routinely face a demanding diagnostic environment, reviewing roughly one cancer for every 200 mammograms. Whiterabbit.ai, another participant in the NVIDIA Inception program, builds software designed to ease that workload.

“Every day, breast radiologists face a needle-in-the-haystack problem, trying to find roughly one cancer in every 200 mammograms,” said Jason Su, cofounder and chief technology officer of Whiterabbit.ai. “We hope AI can be a powerful sidekick to radiologists, helping to clear away the hay so they can focus their expertise where it matters most.”

Whiterabbit.ai’s FDA-cleared WRDensity software automatically assesses breast density from mammograms and has already been utilized in the care of hundreds of thousands of patients. The company also developed WRRisk, a clinical decision support tool for estimating long-term breast cancer risk. Active research is exploring advanced AI models capable of detecting more cancers while automating the clearance of negative mammograms.

Training workloads execute on a cluster of NVIDIA GPUs housed at Washington University in St. Louis, backed by supplementary cloud GPU capacity. Inference runs locally on NVIDIA GPUs deployed directly within clinical settings.

Digital Pathology and Deep Learning Accelerate Treatment Predictions

When a patient receives a breast cancer diagnosis, determining the appropriate therapy usually requires genomic assays sent to outside laboratories. These tests often demand a tissue biopsy and a two- to four-week waiting period.

Ataraxis AI bypasses external lab delays by analyzing digital pathology slides already generated during standard patient workups. The system’s models interpret color-coded clusters representing visual patterns in the slides, linking structural variations directly to recurrence risk and chemosensitivity.

“The tools oncologists rely on today to guide therapy decisions were largely trained once, fifteen years ago, and never updated. Our models get stronger every time we acquire more clinical trial data,” said Joseph Cappadona, member of technical staff at Ataraxis AI. “But as we scale our models, the bigger shift is being able to answer more questions to help oncologists personalize therapy across all cancers.”

Ataraxis operates two core models. One predicts whether presurgical chemotherapy will shrink a tumor sufficiently before a patient reaches the operating room. A second model estimates five-year recurrence risk and the potential benefit of follow-up chemotherapy after surgery. Validated across more than ten institutions and multiple clinical trials, these models run on premise, in offsite data centers, and in the cloud using PyTorch accelerated by NVIDIA CUDA.

Multimodal 3D Modeling Guides Complex Surgical Decisions

SimBioSys brings another dimension to precision medicine by generating accurate 3D models of breast tumors, blood vessels, and soft tissues. The company’s technology evaluates multimodal data—including imaging exams, pathology results, and genomic testing—to generate predictive insights for surgical planning.

“We’re building a platform that now allows us to take multimodal data — imaging exams, pathology results, genomic testing when applicable and other biological inputs — and, using AI, bring that all together,” said Stacey Stevens, CEO of SimBioSys, during an October 1 panel in Phoenix, Arizona, marking Breast Cancer Awareness Month. “When we do that, it generates new insights beyond what we had from any individual piece.”

SimBioSys also utilizes a specialized tool to estimate breast cancer recurrence risk using 3D volumetric data derived from patient breast MRIs, tumor pathology, and clinical records. The platform relies on NVIDIA MONAI for training and validation data alongside NVIDIA CUDA-X libraries, including cuBLAS and MONAI Deploy, operating on cloud-hosted NVIDIA GPUs.

“NVIDIA technology gives us the computing power to take hundreds or thousands of images, apply our AI and analyze them quickly,” Stevens said. “That speed matters because patients and physicians need answers quickly. They can’t afford to wait days or weeks.”