[Restoration filtering based on projection power spectrum for single-photon emission computed tomography]

Area of Science:

• Medical Imaging
• Signal Processing
• Nuclear Medicine

Background:

• Single-photon emission computed tomographic (SPECT) imaging quality is crucial for accurate diagnosis.
• Existing filtering methods like Butterworth and Wiener have limitations in image restoration.
• Developing advanced filters is essential to overcome noise and enhance image resolution in SPECT.

Purpose of the Study:

• To develop and evaluate a novel restoration filter for improving SPECT image quality.
• To compare the performance of the new filter against established Butterworth and Wiener filters.
• To assess the filter's effectiveness in enhancing contrast for clinical brain SPECT imaging.

Main Methods:

• A new restoration filter based on practical "least squares filter" theory was designed.
• The filter utilizes estimated object and noise power spectra derived from projection data.
• Performance was evaluated using Normalized Mean-Squared Errors (NMSE) on phantom studies and clinical brain SPECT images.

Main Results:

• The new filter achieved a lower NMSE (0.76) compared to Butterworth (0.77) and Wiener (0.83) filters in phantom studies.
• Clinical brain SPECT images processed with the new filter demonstrated improved contrast.
• The filter effectively reduced noise while preserving important image features.

Conclusions:

• The developed restoration filter shows significant potential for enhancing SPECT image quality.
• Its ability to improve contrast makes it a valuable tool for the diagnosis of various conditions using SPECT.
• Further clinical validation is warranted to establish its widespread utility in nuclear medicine.