Application of novel corrections for quantification of 123I SPECT

Brown, Colin McMillan (2018) Application of novel corrections for quantification of 123I SPECT. PhD thesis, University of Glasgow.

Abstract

Introduction:
The quantification of clinical images provides a useful adjunct to visual assessment in the differentiation of disease processes. In nuclear medicine imaging, the accurate quantification of Single Photon Emission Computed Tomography (SPECT) data is challenging due to limited spatial resolution and the corrections required for photon attenuation and scatter.

Specific radionuclides used in SPECT imaging, such as Iodine-123 (123I), pose additional challenges to quantification due to their complex decay schemes. 123I has a predominantly low-energy photon emission of 159keV. However, 123I also has high-energy emissions which, due to septal penetration, are detected within the imaging window. Consequently, absolute quantification of 123I SPECT is not current clinical practice and remains a specialist task.

A novel reconstruction correction scheme has been developed by Hermes Medical Solutions which incorporates Monte Carlo simulation of photon interactions in both the patient and the detector system. This Collimator and Detector Response Modelling (CDRM) algorithm has the potential to enhance image quality and, therefore, the quantitative accuracy of 123I SPECT studies. This thesis aims to optimise 123I SPECT quantification using advanced reconstruction algorithms and, furthermore, to assess the clinical applications of these optimised techniques.

Method:
With the ultimate aim of optimising quantification of 123I SPECT, work was undertaken to assess SPECT spatial uniformity, spatial resolution, contrast recovery, noise and scatter suppression. This work was used to specify the optimum collimator and reconstruction parameters required for accurate quantification.

Using these parameters, absolute quantification was then assessed for accuracy with regard to neurology and oncology studies. The utility of Standardised Uptake Values (SUVs) was evaluated in 123I-DaTSCAN patient studies. Furthermore, human observer studies were used to verify the findings of the quantitative assessment.

Results:
Phantom studies demonstrated that Low Energy High Resolution (LEHR) collimators provide superior image quality for neurology applications where spatial resolution is essential. However, when imaging the torso, this work showed that Medium Energy General Purpose (MELP) collimators, with advanced reconstruction, can improve contrast recovery, noise characteristics and scatter suppression when compared with LEHR data.

The accuracy of quantifying activity concentration for neurology studies was optimised using the novel CDRM correction scheme (measured activity concentration within +/-10% of true concentration). However, the accuracy of quantification in torso studies was shown to vary with lesion location in the Field of View (FOV). Therefore, neurology studies were identified as the best candidates for absolute quantification.

In a subsequent evaluation of patient studies, measuring the mean SUV of the putamen in 123I-DaTSCAN studies marginally outperformed Hermes Medical Solutions BRASS automated analysis application with regard to the differentiation of normality. Direct quantitative assessment has the advantage that it removes the requirement for a normal database.

Furthermore, the evaluation of clinical patient 123I-DaTSCAN studies by human observers demonstrated almost perfect agreement in diagnosis for the novel CDRM reconstruction correction scheme (Kappa coefficient=0.913). Image quality for the CDRM scheme rated significantly higher than current clinical practice (p-value<0.01).

The torso phantom observer study suggested that optimised reconstruction of MELP data demonstrated superior image quality and lesion detectability when compared with LEHR reconstructions.

Conclusions:
For 123I-mIBG oncology studies, including quantification of serial studies, data should be acquired with MELP collimators and reconstructed with advanced corrections for attenuation, scatter and depth-dependent spatial resolution. However, quantification of 123I SPECT body section images for inter-patient comparison is not feasible due to variable accuracy with lesion location in the FOV.

Absolute quantification of 123I-DaTSCAN studies, acquired with LEHR collimators, can be performed routinely with sufficient accuracy using the novel CDRM algorithm.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Keywords: SPECT, quantification, I123, 123I, scatter correction, datscan, collimator choice, gamma camera, reconstruction, iterative reconstruction, mibg, spatial resolution, image quality.
• Subjects: Q Science > QC Physics; R Medicine > RC Internal medicine
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Life Sciences
• Supervisor's Name: Gillen, Dr. Gerry
• Date of Award: 2018
• Depositing User: Mr Colin McMillan Brown
• Unique ID: glathesis:2018-8923
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 28 Mar 2018 10:20
• Last Modified: 11 Dec 2023 11:48
• Thesis DOI: 10.5525/gla.thesis.8923
• URI: https://theses.gla.ac.uk/id/eprint/8923