Moffitt Cancer Center Uses Machine Learning to Accelerate Cancer Research

MATLAB provided all the necessary tools to successfully apply AI in imaging and radiomics. MATLAB was used for tasks such as data preprocessing, denoising, and image segmentation.

Key Outcomes

  • MATLAB tools for machine learning and deep learning help accelerate scientific discovery for oncologists when researching new approaches to cancer treatment, prognosis, and diagnosis
  • Machine learning makes it possible to extract key features from patient data, including imaging, genomic data, and medical history, to improve diagnosis accuracy and personalize treatment plans
  • Using MATLAB tools for implementing AI explainability and physics-based techniques, in combination with behavioral science, enhances the robustness of AI for its safe application in cancer treatment

Moffitt Cancer Center is a cutting-edge healthcare institution that uses mathematical oncology, computer science, and informatics to research and treat cancer. Studying the treatment of cancer requires a huge amount of data, ranging from medical history to imaging and genomic data. Oncologists at Moffitt Cancer Center apply machine learning and deep learning to this data to extract information and patterns that a single physician alone could not see.

Oncologists at Moffit Cancer Center use MATLAB® tools for cleaning patient data and machine learning. More specifically, with these tools they perform data preprocessing, image deblurring and denoising, feature extraction, and image segmentation. Doing so has made it possible for doctors to achieve more accurate diagnoses and create personalized treatment plans for different patients.  

Using MATLAB interpretability tools for machine learning and behavioral science, Moffitt Cancer Center researchers are working on improving AI transparency and robustness, thus ensuring that data is used contextually to augment physicians’ skills. These steps will help limit AI bias and make it possible to responsibly expand the tools’ applicability in medicine.