This project will develop novel, widely applicable statistical methods for the accurate differential diagnosis and longitudinal follow-up of dopaminergic function in normal aging in healthy subjects, and neuronal degeneration in patients with disorders such as Parkinson's disease (PD), using single photon emission tomography (SPECT). While numerous methods exist for the pixel-based comparison groups of subjects (such as Statistical Parametric Mapping), applying them to the diagnosis of individual subjects is inaccurate. The techniques developed in this study will use pixel-based, data-driven methods for diagnosing individual subjects. PD is a long-term and devastating disorders, afflicting over a million Americans and costing society over $25 billion annually. Existing diagnostic techniques, based on clinical symptoms and anatomical and functional imaging, are inaccurate and cannot easily distinguish the parkinsonian symptoms associated with this disorder, particularly when the patient presents at an early stage of the disease. Of particular importance is the diagnosis of disease at a very early stage, or the screening of """"""""at risk"""""""" individuals who may present before clinical symptoms become apparent. It is also important to provide a statistically valid analysis of longitudinal follow-up studies, where subjects are studied multiple times to measure the effects of normal aging, or disease progression and the efficacy of treatment. Conventional region-of-interest (ROI) analysis techniques are time-consuming and subjective and known to be prone to operator bias. ROI analysis also dilutes small focal changes in brain behavior, reducing the sensitivity and specificity of the method leading to misdiagnoses. This project will apply and develop novel automated, pixel-based statistical techniques leading to accurate and unbiased diagnoses, and improved patient management. Neuronal loss in the dopaminergic, nigrostriatal system will be monitored using SPECT imaging"""""""" """""""" of [99m Tc]TRODAT-1 binding to dopamine transporters. Two methods, a pixel-based logistic discriminant analysis system and a channelized Hotelling observer, will be """"""""trained"""""""" on known images, and then used to distinguish between patients and controls, based on regional differences in neuronal degeneration. The optimal combination of discriminatory factors will be established for the disorder, and the methodology applied to patients with more equivocal diagnoses presenting earlier in the course of the disease. In summary, this project will develop widely applicable and powerful neuroimaging analysis tools for the accurate, objective differential diagnosis of neurodegenerative disorders, and for the longitudinal analysis of follow-up studies.
| Bal, H; Bal, G; Acton, P D (2007) Diagnosis of Parkinsonian disorders using a channelized Hotelling observer model: proof of principle. Med Phys 34:3987-95 |
| Acton, Paul D; Newberg, Andrew; Plossl, Karl et al. (2006) Comparison of region-of-interest analysis and human observers in the diagnosis of Parkinson's disease using [99mTc]TRODAT-1 and SPECT. Phys Med Biol 51:575-85 |
| Acton, Paul D; Newberg, Andrew (2006) Artificial neural network classifier for the diagnosis of Parkinson's disease using [99mTc]TRODAT-1 and SPECT. Phys Med Biol 51:3057-66 |
