Rhabdomyosarcoma (RMS), which usually begins in muscles that are attached to bones, is the most common type of soft tissue sarcoma in children. Having an outlook of the patient prognosis is crucial for determining treatment options. The objective of this proposal is to design and develop deep learning tools to provide RMS patient prognosis prediction from whole slide images (WSIs), genetic data and clinical data. The rationale underlying this proposal is that the development of the deep learning tools will provide objective measurements and judgements of the disease and make pathologists and physicians better informed to make precise diagnosis and treatment suggestions. The goal will be real...

Read More

Rhabdomyosarcoma (RMS), which usually begins in muscles that are attached to bones, is the most common type of soft tissue sarcoma in children. Having an outlook of the patient prognosis is crucial for determining treatment options. The objective of this proposal is to design and develop deep learning tools to provide RMS patient prognosis prediction from whole slide images (WSIs), genetic data and clinical data. The rationale underlying this proposal is that the development of the deep learning tools will provide objective measurements and judgements of the disease and make pathologists and physicians better informed to make precise diagnosis and treatment suggestions. The goal will be realized by pursuing three specific aims: 1) Develop deep learning models for RSM patient outcome prediction from digitized histopathological WSIs; 2) Include genetic and clinical information in addition to WSI image data and create a multimodal deep learning model to make prognostic predictions of the patients; 3) Develop a user-friendly web-portal to facilitate pediatric RMS research and patient care. The proposed research is significant as the completion of it will provide viable tools to aid pathologists and physicians to improve RMS diagnosis and treatments, and it could be extendable to other malignant diseases. In summary, we have assembled a multi-disciplinary research team with complementary research expertise. We will fully leverage our accumulated data and extensive experience to solve the challenge of developing computational algorithms for pathology imaging analysis and outcome prediction for pediatric RMS. This will greatly facilitate treatment planning for individual RMS patients and will have an important impact on clinical care.

Read Less