Dynamic Cone-beam CT Reconstruction using Spatial and Temporal Implicit Neural Representation Learning (STINR)

TL;DR

This paper introduces STINR, a novel neural network-based method for dynamic cone-beam CT reconstruction that effectively captures complex motion and anatomy variations with limited projection data.

Contribution

STINR combines spatial and temporal neural representations with PCA-based motion models to improve dynamic CBCT reconstruction accuracy under challenging conditions.

Findings

STINR outperforms traditional PCA and polynomial methods in image quality and motion tracking.

Achieves tumor tracking with <2 mm error and <10% relative error in reconstructions.

Effectively handles various lung motion scenarios and anatomical variations.

Abstract

Objective: Dynamic cone-beam CT (CBCT) imaging is highly desired in image-guided radiation therapy to provide volumetric images with high spatial and temporal resolutions to enable applications including tumor motion tracking/prediction and intra-delivery dose calculation/accumulation. However, the dynamic CBCT reconstruction is a substantially challenging spatiotemporal inverse problem, due to the extremely limited projection sample available for each CBCT reconstruction (one projection for one CBCT volume). Approach: We developed a simultaneous spatial and temporal implicit neural representation (STINR) method for dynamic CBCT reconstruction. STINR mapped the unknown image and the evolution of its motion into spatial and temporal multi-layer perceptrons (MLPs), and iteratively optimized the neuron weighting of the MLPs via acquired projections to represent the dynamic CBCT series. In addition to the MLPs, we also introduced prior knowledge, in form of principal component analysis (PCA)-based patient-specific motion models, to reduce the complexity of the temporal INRs to address the ill-conditioned dynamic CBCT reconstruction problem. We used the extended cardiac torso (XCAT) phantom to simulate different lung motion/anatomy scenarios to evaluate STINR. The scenarios contain motion variations including motion baseline shifts, motion amplitude/frequency variations, and motion non-periodicity. The scenarios also contain inter-scan anatomical variations including tumor shrinkage and tumor position change. Main results: STINR shows consistently higher image reconstruction and motion tracking accuracy than a traditional PCA-based method and a polynomial-fitting based neural representation method. STINR tracks the lung tumor to an averaged center-of-mass error of <2 mm, with corresponding relative errors of reconstructed dynamic CBCTs <10%.