Deep learning with noise‐to‐noise training for denoising in SPECT myocardial perfusion imaging

Abstract

Purpose Post‐reconstruction filtering is often applied for noise suppression due to limited data counts in myocardial perfusion imaging (MPI) with single‐photon emission computed tomography (SPECT). We study a deep learning (DL) approach for denoising in conventional SPECT‐MPI acquisitions, and investigate whether it can be more effective for improving the detectability of perfusion defects compared to traditional postfiltering. Methods Owing to the lack of ground truth in clinical studies, we adopt a noise‐to‐noise (N2N) training approach for denoising in SPECT‐MPI images. We consider a coupled U‐Net (CU‐Net) structure which is designed to improve learning efficiency through feature map reuse. For network training we employ a bootstrap procedure to generate multiple noise realizations from list‐mode clinical acquisitions. In the experiments we demonstrated the proposed approach on a set of 895 clinical studies, where the iterative OSEM algorithm with three‐dimensional (3D) Gaussian postfiltering was used to reconstruct the images. We investigated the detection performance of perfusion defects in the reconstructed images using the non‐prewhitening matched filter (NPWMF), evaluated the uniformity of left ventricular (LV) wall in terms of image intensity, and quantified the effect of smoothing on the spatial resolution of the reconstructed LV wall by using its full‐width at half‐maximum (FWHM). Results Compared to OSEM with Gaussian postfiltering, the DL denoised images with CU‐Net significantly improved the detection performance of perfusion defects at all contrast levels (65%, 50%, 35%, and 20%). The signal‐to‐noise ratio (SNRD) in the NPWMF output was increased on average by 8% over optimal Gaussian smoothing (P < 10−4, paired t‐test), while the inter‐subject variability was greatly reduced. The CU‐Net also outperformed a 3D nonlocal means (NLM) filter and a convolutional autoencoder (CAE) denoising network in terms of SNRD. In addition, the FWHM of the LV wall in the reconstructed images was varied by less than 1%. Furthermore, CU‐Net also improved the detection performance when the images were processed with less post‐reconstruction smoothing (a trade‐off of increased noise for better LV resolution), with SNRD improved on average by 23%. Conclusions The proposed DL with N2N training approach can yield additional noise suppression in SPECT‐MPI images over conventional postfiltering. For perfusion defect detection, DL with CU‐Net could outperform conventional 3D Gaussian filtering with optimal setting as well as NLM and CAE.

The promise of artificial intelligence and deep learning in PET and SPECT imaging

“…Compared to spatiotemporal non-local mean (NLM) post-reconstruction filtering, the deep learning algorithm achieved significant improvement in spatial resolution of the left ventricular wall. As an alternative to post-reconstruction filtering, Liu et al employed a coupled-Unet to suppress the noise in conventional SPECT-MPI acquisitions [64]. A noise-to-noise denoising approach was adopted and compared with traditional post-filtering methods owing to the lack of ground truth/reference in clinical studies.…”

Accounting For Inter-Subject Variations in Deep Learning for Reduced-Dose Studies in Cardiac SPECT

“…In this study, we utilized a set of clinical data acquired from 895 subjects (453/442 male/female, BMI: 32.8 ± 6.6 kg•m -2 , age: 62.2 ± 10.6 years) with informed consent under IRB approval [9], [24]. These studies were acquired with standard dose in list-mode on a Philips BrightView SPECT/CT system with 64 projections (3-degree steps) and a 128x128 matrix.…”

“…To combat the increased imaging noise, reconstruction algorithms with resolution recovery as well as attenuation and scatter corrections have been studied to improve the image reconstruction accuracy [17]- [21]. Most recently, CNN based models were employed to suppress noise in low-dose as well as full-dose SPECT studies [9], [10]. In particular, in [9] a convolutional autoencoder (CAE) network was demonstrated to be effective for improving the detectability of perfusion defects in lowdose SPECT images.…”

“…This will ensure that a trained DL network can achieve good generalizability when deployed for clinical images that are unseen during training. Therefore, in our previous developments [9], [10] clinical acquisitions (instead of simulated phantom data) were used for training and evaluating the DL performance.…”

Prediction of an oxygen extraction fraction map by convolutional neural network: validation of input data among MR and PET images

“…Deep learning, which is a type of machine learning with a neural network consisting of numerous layers [ 13 , 14 ], has been recently and widely used in the computer vision area. Deep-learning techniques, such as a convolutional neural network (CNN) and generative adversarial network, have been applied for image synthesis and transformation between different images as follows: denoizing/superresolution [ 15 , 16 ], synthesis of computed tomography (CT) images from MR images [ 17 – 19 ], motion correction [ 20 ], missing data recovery [ 21 ], and image reconstruction [ 22 – 24 ]. To map ischemic stroke, prediction of CBF [ 25 ] and cerebrovascular reserve [ 26 ] maps using the CNN learned with arterial spin labeling (ASL) maps and structural MR images have been proposed.…”