Automatic putamen detection on DTI images. Application to Parkinson's Disease

The putamen, a complex structure located at the base of the forebrain, is connecting neuromotor fibres affected in neurodegenerative diseases, like Parkinson's Disease (PD). We study PD on a total of 47 subjects, with 22 patients clinically diagnosed with PD and 25 control cases. Using Diffusion Tensor Imaging (DTI), a type of Magnetic Resonance Imaging(MRI), we analyze the fractional anisotropy (FA) image stacks. These images hold the information relative to the dopamine flow, one of the neurotransmitters affecting the PD. The present article introduces a new method for the active automatic detection of the putamen. We evaluate our method using the manual tracing of the same anatomical region, validated by a medical expert (neurologist). Our method employs a geometrical approach that updates a classic active contour tracking method by eliminating the inter-patient variability. From the technical point of view we are providing an automated tool similar to an atlas approach, but with more versatility. From the medical point of view the importance of our approach resides in the possibility to provide a specific anatomical detail for further study: the putamen brain structure. This algorithm is part of PDFibAtl@s, our own PD detection system based on DTI image processing and analysis.