Abstract:Background: Artificial intelligence-based methods have generated substantial interest in nuclear medicine. An area of significant interest has been the use of deep-learning (DL)-based approaches for denoising images acquired with lower doses, shorter acquisition times, or both. Objective evaluation of these approaches is essential for clinical application. Purpose: DL-based approaches for denoising nuclear-medicine images have typically been evaluated using fidelity-based figures of merit (FoMs) such as root mean squared error (RMSE) and structural similarity index measure (SSIM). However, these images are acquired for clinical tasks and thus should be evaluated based on their performance in these tasks. Our objectives were to: (1) investigate whether evaluation with these FoMs is consistent with objective clinical-task-based evaluation; (2) provide a theoretical analysis for determining the impact of denoising on signal-detection tasks; and (3) demonstrate the utility of virtual imaging trials (VITs) to evaluate DL-based methods. Methods: A VIT to evaluate a DL-based method for denoising myocardial perfusion SPECT (MPS) images was conducted. To conduct this evaluation study, we followed the recently published best practices for the evaluation of AI algorithms for nuclear medicine (the RELAINCE guidelines). An anthropomorphic patient population modeling clinically relevant variability was simulated. Projection data for this patient population at normal and low-dose count levels (20%, 15%, 10%, 5%) were generated using well-validated Monte Carlo-based simulations. The images were reconstructed using a 3-D ordered-subsets expectation maximization-based approach. Next, the low-dose images were denoised using a commonly used convolutional neural network-based approach. The impact of DL-based denoising was evaluated using both fidelity-based FoMs and area under the receiver operating characteristic curve (AUC), which quantified performance on the clinical task of detecting perfusion defects in MPS images as obtained using a model observer with anthropomorphic channels. We then provide a mathematical treatment to probe the impact of post-processing operations on signal-detection tasks and use this treatment to analyze the findings of this study. Results: Based on fidelity-based FoMs, denoising using the considered DL-based method led to significantly superior performance. However, based on ROC analysis, denoising did not improve, and in fact, often degraded detection-task performance. This discordance between fidelity-based FoMs and task-based evaluation was observed at all the low-dose levels and for different cardiac-defect types. Our theoretical analysis revealed that the major reason for this degraded performance was that the denoising method reduced the difference in the means of the reconstructed images and of the channel operator-extracted feature vectors between the defect-absent and defect-present cases. Conclusions: The results show the discrepancy between the evaluation of DL-based methods with fidelity-based metrics versus the evaluation on clinical tasks. This motivates the need for objective task-based evaluation of DL-based denoising approaches. Further, this study shows how VITs provide a mechanism to conduct such evaluations computationally, in a time and resource-efficient setting, and avoid risks such as radiation dose to the patient. Finally, our theoretical treatment reveals insights into the reasons for the limited performance of the denoising approach and may be used to probe the effect of other post-processing operations on signal-detection tasks.

Impact of image denoising on image quality, quantitative parameters and sensitivity of ultra-low-dose volume perfusion CT imaging

Study on the Optimized Scanning Conditions of Cerebral CT Angiography Based on the Angiographic CT Head Phantom

Abstract WP248: Pre-treatment Perfusion CT in Stroke Thrombolysis Can Predict Disability-free Life

Image quality, radiation dose and diagnostic accuracy of 70 kVp whole brain volumetric CT perfusion imaging: a preliminary study

Noise and dose reduction in CT brain perfusion acquisition by projecting time attenuation curves onto lower dimensional spaces

Self-Supervised Dynamic CT Perfusion Image Denoising With Deep Neural Networks

Feasibility of double low techniques in cerebral perfusion using 640-slice volume CT

Feasibility of Dual-Low Scheme Combined with Iterative Reconstruction Technique in Acute Cerebral Infarction Volume CT Whole Brain Perfusion Imaging

Neural network-derived perfusion maps: A model-free approach to computed tomography perfusion in patients with acute ischemic stroke

Adaptive prior image constrained total generalized variation for low-dose dynamic cerebral perfusion CT reconstruction

Abstract TP89: Absolute Cerebral Blood Flow Assessment With Novel Low-Dose Whole Brain Dynamic CT Perfusion Technique

Learning non-local perfusion textures for high-quality computed tomography perfusion imaging

Impact of temporal resolution on perfusion metrics, therapy decision, and radiation dose reduction in brain CT perfusion in patients with suspected stroke

Iterative reconstruction for low-dose cerebral perfusion computed tomography using prior image induced diffusion tensor

Feasibility Study of Change in Normal Cerebral Hemodynamics by Low Dose Whole-brain Perfusion with 640 Slice Volume CT

Dose reduction and image enhancement in micro-CT using deep learning

Need for objective task-based evaluation of deep learning-based denoising methods: A study in the context of myocardial perfusion SPECT

Deep Learning-Based Denoising Enables High-Quality, Fully Diagnostic Neuroradiological Trauma CT at 25% Radiation Dose

Building a Kinetic Induced Voxel-Clustering Filter (KVCF) for Low-Dose Perfusion CT Imaging

A novel use of time separation technique to improve flat detector CT perfusion imaging in stroke patients

CT brain perfusion: A static phantom study of contrast‐to‐noise ratio and radiation dose