Impacts of Intelligent Automated Quality Control on A Small Animal Apd-Based Digital Pet Scanner

Stable system performance is mandatory to warrant the accuracy and reliability of biological results relying on small animal positron emission tomography (PET) imaging studies. This simple requirement sets the ground for imposing routine quality control (QC) procedures to keep PET scanners at a reliable optimal performance level. However, such procedures can become burdensome to implement for scanner operators, especially taking into account the increasing number of data acquisition channels in newer generation PET scanners. In systems using pixel detectors to achieve enhanced spatial resolution and contrast-to-noise ratio (CNR), the QC workload rapidly increases to unmanageable levels due to the number of independent channels involved. An artificial intelligence based QC system, referred to as Scanner Intelligent Diagnosis for Optimal Performance (SIDOP), was proposed to help reducing the QC workload by performing automatic channel fault detection and diagnosis. SIDOP consists of four high-level modules that employ machine learning methods to perform their tasks: Parameter Extraction, Channel Fault Detection, Fault Prioritization, and Fault Diagnosis. Ultimately, SIDOP submits a prioritized faulty channel list to the operator and proposes actions to correct them. To validate that SIDOP can perform QC procedures adequately, it was deployed on a LabPET (TM) scanner and multiple performance metrics were extracted. After multiple corrections on sub-optimal scanner settings, a 8.5% (with a 95% confidence interval (CI) of [7.6, 9.3]) improvement in the CNR, a 17.0% (CI: [15.3, 18.7]) decrease of the uniformity percentage standard deviation, and a 6.8% gain in global sensitivity were observed. These results confirm that SIDOP can indeed be of assistance in performing QC procedures and restore performance to optimal figures.