Registration and display of brain SPECT and MRI using external markers

Area of Science:

• Medical Imaging
• Neuroscience
• Computer Science

Background:

• Accurate anatomical localization of brain abnormalities in SPECT imaging is challenging.
• Correlating SPECT data with other tomographic modalities like MRI can improve localization.
• Existing methods may lack efficient tools for integrated analysis and display.

Purpose of the Study:

• To develop and evaluate software for registering, analyzing, and displaying brain perfusion SPECT and MRI data.
• To improve the anatomical accuracy of SPECT findings by integrating with MRI.
• To provide a tool for enhanced visualization of neurological abnormalities.

Main Methods:

• Developed C-coded software using X Window and OSF/Motif for SPECT-MRI registration and display.
• Utilized external skin markers containing 99mTc for SPECT/MRI data registration.
• Employed a noniterative least-squares method with singular value decomposition for registration.
• Implemented 2D side-by-side and 3D surface-shaded/cut-plane display modes.

Main Results:

• Achieved an average registration error of 4 mm in phantom studies.
• Successfully displayed co-registered SPECT and MRI slices at identical tomographic levels.
• Demonstrated the software's utility in visualizing abnormalities in patients with cerebral infarcts, brain tumors, encephalitis, and epilepsy.

Conclusions:

• The developed software effectively registers and integrates brain perfusion SPECT and MRI data.
• This method enhances the anatomical localization of abnormalities detected by SPECT.
• The tool shows significant potential for clinical diagnosis and research in neurology.