Radiomics and Radio-Pathomics in Pediatric Brain Cancer
Mentor 1
Peter LaViolette
Start Date
28-4-2023 12:00 AM
Description
Pediatric brain tumors vary in degrees of significance. Benign brain tumors like Pilocytic Astrocytoma have a survivability rate upwards of 95%, while malignant tumors like Glioblastoma have a 5-year survival rate of only 20%. Magnetic resonance imaging (MRI) is the clinical standard in pediatric patients for diagnosis and treatment planning for brain cancer. Current contrast imaging is limited in observing pseudo-progression and undetectable smaller tumor cells, information which is necessary for neurosurgeons and radiologists to perform more effective surgeries. Our study combines MRI scans from pediatric brain cancer patients with image processing and machine learning techniques for differentiating tumor types and detecting hidden characteristics. Our preliminary research studies for adult glioblastoma radio-pathomic models have mapped pseudoprogression and tumor margins beyond MRI result in standard clinical practices. Therefore, we hypothesize that radiomic analyses of pediatric brain cancer imaging can differentiate tumor types and that radio-pathomic techniques can detect pediatric brain tumor invasion beyond traditionally defined margins. Radiological-pathological (Rad-Path) imaging studies of pediatric brain tumors have been limited to biopsy samples aligned to pre-surgical MRI scans. Only a handful of studies have focused on improving pediatric brain tumor radiomics. With access to MR (Magnetic Resonance) imaging of pediatric brain tumors from Children's of Wisconsin and the tumor phenotype information, we will use radiomics to build classification algorithms. Additionally, with access to the world's largest radiology-focused brain bank dedicated to studying brain tumors, we will apply radio-pathomic approaches to pediatric brain cancer imaging to detect brain cancer invasion beyond conventional margins. We anticipate that radio-pathomic maps of tumor cellularity can differentiate different phenotypes of pediatric brain tumors from MRI scans. We expect a clear difference between the appearance of malignant and benign brain tumors based on the presence of hypercellularity beyond contrast enhancement.
Radiomics and Radio-Pathomics in Pediatric Brain Cancer
Pediatric brain tumors vary in degrees of significance. Benign brain tumors like Pilocytic Astrocytoma have a survivability rate upwards of 95%, while malignant tumors like Glioblastoma have a 5-year survival rate of only 20%. Magnetic resonance imaging (MRI) is the clinical standard in pediatric patients for diagnosis and treatment planning for brain cancer. Current contrast imaging is limited in observing pseudo-progression and undetectable smaller tumor cells, information which is necessary for neurosurgeons and radiologists to perform more effective surgeries. Our study combines MRI scans from pediatric brain cancer patients with image processing and machine learning techniques for differentiating tumor types and detecting hidden characteristics. Our preliminary research studies for adult glioblastoma radio-pathomic models have mapped pseudoprogression and tumor margins beyond MRI result in standard clinical practices. Therefore, we hypothesize that radiomic analyses of pediatric brain cancer imaging can differentiate tumor types and that radio-pathomic techniques can detect pediatric brain tumor invasion beyond traditionally defined margins. Radiological-pathological (Rad-Path) imaging studies of pediatric brain tumors have been limited to biopsy samples aligned to pre-surgical MRI scans. Only a handful of studies have focused on improving pediatric brain tumor radiomics. With access to MR (Magnetic Resonance) imaging of pediatric brain tumors from Children's of Wisconsin and the tumor phenotype information, we will use radiomics to build classification algorithms. Additionally, with access to the world's largest radiology-focused brain bank dedicated to studying brain tumors, we will apply radio-pathomic approaches to pediatric brain cancer imaging to detect brain cancer invasion beyond conventional margins. We anticipate that radio-pathomic maps of tumor cellularity can differentiate different phenotypes of pediatric brain tumors from MRI scans. We expect a clear difference between the appearance of malignant and benign brain tumors based on the presence of hypercellularity beyond contrast enhancement.
