Dopamine Neuron Activity Visualization From MRI Sensors Through 3-D MRI-to-SPECT Synthesis for Parkinson’s Disease Assessment

Single Photon Emission Computed Tomography (SPECT) is a radiological imaging technique used for diagnosing and assessing neurological disorders, including Parkinson’s disease. However, direct SPECT imaging requires the injection of radioactive drugs, posing risks of radiation exposure and allergic reactions associated with invasive procedures. In this paper, we propose a novel framework called the 3D-Cycle Conversion Network. Its aim is to extract 3D MRI features captured by MRI sensors and convert them into synthesized SPECT images, enabling the detection of differences in dopamine neuron activity. To the best of our knowledge, we are the first group to effectively synthesize SPECT while preserving important diagnostic features. Inspired by CycleGAN, this architecture uses a cyclic transformation approach to maintain fundamental medical features and includes specifically designed reconstruction modules for 3D image synthesis. During training, a specially designed loss function guides our framework to enhance the accuracy of SPECT image generation. In experimental results, our framework outperforms other state-of-the-art approaches in both quantitative and qualitative evaluations.