Our project, GAN4Seg, is a react webapp that is designed to produce segmented images for the purpose of training data for classifier model. The main purpose of our webapp is to deliver top-tier segmented images (skin, bone, brain matter, air, and cerebral spinal fluid), enhancing training data for medical imaging classifiers. This would contribute to the medical community as our project aids physiological analyses of patients and reduces pressure within decision-making situations in a surgical context as previously, the classifier model has to be able to determine parts of a brain scan, but providing this model lots of segmented training data is time-consuming. We intend to implement the GAN model by using specialty packages such as TensorFlow, PyTorch along with our frontend webapp being in react to utilizes UI libraries to help solve our main issues.
We’re working with Dr. Andrew Abumoussa of UNC Neurosurgery to address the main issues in the medical field. We need to speed up the MRI and CT scan analysis process which will reduce waiting times for patients. Our app should offer enhanced precision by reducing human errors, leading to more accurate diagnoses. This project should also help meet increasing demand as the healthcare industry continues to grow. We should also have our solution to these issues to be cost effective as hospitals and clinics can reduce overheads associated with time-consuming manual analysis.
Our overall goal is to improve the accuracy of semantic segmentation for brain CTs and MRIs that addresses a critical need in the healthcare and medical research community. By increasing accuracy, our webapp can lead to better patient care, reduced errors, and advancements in the understanding and treatment of neurological conditions. It aligns with the values of making a positive impact and contributing to the betterment of healthcare and research.