Artificial intelligence (AI) is transforming breast cancer care by addressing significant challenges in diagnosis and treatment. Despite breast cancer being the most commonly diagnosed cancer among American women, many women over 40 skip recommended annual screenings, radiologists face increasing workloads with fewer colleagues, and treatment decisions can be delayed by weeks-long test results. AI applications developed by companies in the NVIDIA Inception program are supporting clinicians across these critical points, including imaging, risk assessment, and treatment planning, by leveraging advanced computational power.
Automated Imaging and Enhanced Detection
Access to timely and convenient breast cancer screening is a major barrier. iSono Health, an NVIDIA Inception startup, has developed the FDA-cleared ATUSA platform, a wearable, automated 3D quantitative ultrasound system. This system captures a standardized breast volume in about two minutes per breast, a significant reduction compared to the up to 45 minutes a conventional handheld ultrasound can take. Its AI, trained on thousands of full-breast scans (over 1.5 million ultrasound frames), automates image acquisition and is reported to be 28% more sensitive than a handheld 2D ultrasound. Unlike handheld ultrasounds, which can vary based on the operator, ATUSA captures the whole breast consistently, allowing for repeatable comparisons of tissue changes across successive scans and reducing operator variability.
iSono Health’s AI capabilities extend to lesion detection, 3D segmentation, and lesion classification. The company plans to integrate its AI pipeline with multimodal diagnostic intelligence, combining 3D ultrasound, mammography, MRI, and clinical information. The ATUSA system is commercially available through partner clinics in several U.S. states, including California, Texas, Georgia, Tennessee, and Washington D.C., with more sites expected to come online. A multicenter clinical study involving 3,200 patients is also underway at lead research sites like UC Davis and Vanderbilt University Medical Center to further validate the platform's performance.
Another NVIDIA Inception company, Whiterabbit.ai, focuses on improving breast cancer screening with AI. Their FDA-cleared WRDensity software automatically assesses breast density from mammograms, having been used for hundreds of thousands of patients. Additionally, Whiterabbit.ai offers WRRisk, a clinical decision support software that estimates a patient's long-term risk of developing breast cancer. The company is also developing a new generation of AI for mammography aimed at helping radiologists detect more cancers and automating the screening of negative mammograms. This initiative seeks to alleviate the burden on radiologists, accelerate results, reduce unnecessary patient callbacks, and lower healthcare costs. Whiterabbit.ai trains its AI models on NVIDIA GPUs at Washington University in St. Louis and in the cloud, with inference running on NVIDIA GPUs in clinics.
Predicting Treatment Response and Recurrence Risk
Once breast cancer is diagnosed, predicting how a patient will respond to treatment is crucial but currently limited in scope and accuracy, often requiring lengthy biopsy wait times. Ataraxis AI is developing clinical intelligence that predicts patient outcomes and responses to various therapies using digital data, such as pathology slides that are already part of standard patient workups. Their AI models analyze patterns in these slides to associate variations with differences in recurrence risk and chemosensitivity. These models are designed to improve over time with more clinical trial data, aiming to personalize therapy across all cancers. One model predicts the likelihood of presurgical chemotherapy shrinking a tumor, while another estimates five-year recurrence risk and the benefit of chemotherapy after surgery. Both models have been validated across over 10 institutions and multiple clinical trials and are actively used, running on NVIDIA GPUs using PyTorch accelerated by NVIDIA CUDA.
SimBioSys, also an NVIDIA Inception company, creates AI-powered precision medicine technology that generates accurate 3D models of breast tumors, veins, and soft tissue. These models provide critical insights to guide surgeries and influence treatment plans. The company has also built a tool that estimates breast cancer recurrence risk based on 3D volumetric data from MRI scans, tumor pathology, and clinical data. SimBioSys integrates multimodal data—imaging exams, pathology results, genomic testing, and other biological inputs—using AI to generate insights beyond what individual data points could provide. SimBioSys uses NVIDIA MONAI for training and validation data, and NVIDIA CUDA-X libraries, including cuBLAS and MONAI Deploy, for its imaging technology, which runs on NVIDIA GPUs in the cloud. This provides the computing power needed to process and analyze complex medical data effectively.