Hybrid kernelised expectation maximisation for Bremsstrahlung SPECT reconstruction in SIRT with ⁹⁰Y micro-spheres

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Radiotherapy Physics

Background:

• Selective Internal Radiation Therapy (SIRT) with Yttrium-90 (Y-90) is effective for liver cancer.
• Y-90 beta particles produce Bremsstrahlung radiation, enabling post-therapy imaging.
• Current imaging suffers from poor quality and inaccurate dosimetry due to continuous energy spectrum challenges.

Purpose of the Study:

• To improve image resolution, contrast, and noise reduction for Y-90 SIRT imaging.
• To investigate the efficacy of a novel reconstruction algorithm for enhanced quantitative SPECT imaging.
• To validate the algorithm's performance using phantom and patient data.

Main Methods:

• Utilized the hybrid kernelised expectation maximisation (HKEM) algorithm.
• Investigated a modified version, frozen HKEM (FHKEM), for further noise reduction.
• Validated the algorithms using a NEMA phantom and clinical data from 5 SIRT patients.

Main Results:

• Achieved up to 56% improvement in region of interest mean recovery coefficient in phantom studies.
• Demonstrated better identification of hot volumes in the NEMA phantom.
• Reported a 47% mean value improvement over the current clinical standard using patient data.

Conclusions:

• The developed algorithm significantly enhances quantitative SPECT imaging for Y-90 SIRT.
• Improved dosimetry calculations and better visualization of liver cancerous lesions are facilitated.
• This advancement holds potential for more precise cancer treatment planning and monitoring.