HYPR4D Kernel Method With an Unsupervised 2.5SD+0.5TD Deep Learning Assisted Kernel Matrix

• Published in: IEEE transactions on radiation and plasma medical sciences Vol. 9; no. 1; pp. 20 - 28
• Main Authors: Kevin Cheng, Ju-Chieh, Reimers, Erik, Sossi, Vesna
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.01.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Bias; Convolutional neural networks; Data acquisition; Deep learning; HYPR4D kernel method; Image reconstruction; intrinsic spatiotemporal consistency; Kernel; Noise reduction; Positron emission; Positron emission tomography; single subject training; Smoothing; Spatiotemporal data; Spatiotemporal phenomena; Training; Tuning; unsupervised deep learning (DL); Unsupervised learning
• ISSN: 2469-7311; 2469-7303
• DOI: 10.1109/TRPMS.2024.3442690

We describe a deep learning (DL) assisted HYPR4D kernelized reconstruction which produces low-noise voxel-level time-activity-curves (TACs) while preserving quantification within small structures as well as consistent spatiotemporal patterns/features within measured data. The proposed method consists of the following advantages over other methods: 1) unsupervised single subject network training scheme independent of positron emission tomography (PET) tracers; 2) training data generated on-the-fly during reconstruction; 3) intrinsic spatiotemporal consistency provided by minimizing the <inline-formula> <tex-math notation="LaTeX">L_{2} </tex-math></inline-formula> loss using pseudo 4-D (i.e., 2.5 Spatial Dimension + 0.5 Temporal Dimension or 2.5SD+0.5TD) patches between kernelized OSEM subset estimates; and 4) a final tuning step which minimizes over-smoothing from the network output within the kernel matrix. Contrast phantom, human [18F]FDG and [11C]RAC data acquired on GE SIGNA PET/MR were used for evaluations. The proposed DL HYPR4D kernel method outperformed the standard HYPR4D kernel method as well as TOF-OSEM and TOF-BSREM (Q.Clear) in terms contrast recovery versus noise. The proposed final tuning reduced the underestimation bias due to over-smoothing within a 4-mm target structure from ~15% to ~2% while maintaining low-noise voxel-level TACs. In addition, the proposed unsupervised DL assisted reconstruction also outperformed the supervised DL version in terms of bias reduction along the TACs and kinetic model fittings.