Computer-aided detection of focal cortical dysplasias - ¿
DESCRIPTION (provided by applicant): This project proposes to build computer-aided detection (CAD) software for use in identifying cortical malformations known as focal cortical dysplasia's (FCDs), which are a common cause of epileptic seizures. The intent is for the software, used by a neuroradiologist at a clinical workstation, to decrease the time- intensive nature of the visual search for cortical dysplasia's, while simultaneously increasing sensitivity o dysplasia identification, thus reducing the number of missed lesions and making neuroradiologists more effective and more efficient. Epilepsy is a common neurological disorder, characterized by recurrent unprovoked seizures, that exacts a large toll upon society in terms of both quality of life and health care costs. Malformations of cortical development (MCD) are the most common cause of seizures in children and the second most common cause in adults. Focal cortical dysplasia is a common form of MCD that is responsible for the vast majority of treatment resistant epilepsy in patients with MCD, and when anti-epileptic medication is ineffective, detection of FCD becomes critical to the ability of the epilepsy team to offer surgery which is often these patient's last hope for seizure freedom. Unfortunately, the radiological diagnosis of FCD is exceedingly difficult in a large percentage of cases due to their focal and subtle nature. Thus, while resection of these dysplasias can often cure seizures, they can be missed for years or decades, resulting in increased neurological damage and degradation of quality of life due to chronic seizures. In principle high resolution MRI can be used to increase diagnostic accuracy. While this is becoming more common in clinical practice, the need for high patient throughput, lack of clinical information and inexperience often results i these lesions being missed on routine clinical reads by neuroradiologists. The project will build upon a foundation of existing technology for the generation of quantitative measures of the human brain based on MRI imaging, known in the neuroimaging research domain as FreeSurfer. The project will make use of an MRI dataset of 100 subjects with histologically-confirmed FCD to be labeled by four neuroradiologists, and control subjects with epilepsy that is not due to FCD. The project has three aims: gathering the dataset and expansion of the detection algorithms tested in Phase I to include additional MRI biomarkers; development of an MRI scanner slice prescription component to ensure imaging of an FCD at the optimal visualization plane; and an aim to submit a commercialized version of FreeSurfer for FDA 510(k) clearance. The latter aim is important for the long-term project goal of advancing the state of other clinical detection methods through the building of additional CAD tools making use of FreeSurfer's brain measures, including diseases as varied as Huntington's disease, Alzheimer's disease, tumor monitoring and hydrocephalus.
