Improvement of Image Quality in Low-Count Bone Scintigraphy Using Deep Learning

Abstract

Abstract Objective To improve image quality for low-count bone scintigraphy whole-body images using deep learning and evaluate their applicability in clinical practice.Methods Five hundred fifty patients were included in the study. Low-count Original images (75%, 50%, 25%, 10%, and 5% counts) were generated from Reference images (100% counts) using Poisson resampling. Patients were randomly divided into training (500) and evaluation (50) groups. Output (DL-filtered) images were obtained after training with U-Net using Reference images as teacher data. Gaussian-filtered images were generated for comparison. Peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) to the Reference image were calculated to determine image quality. Artificial neural network (ANN) value, bone scan index (BSI), and number of hotspots (Hs) were computed using BONENAVI analysis for patients with and without bone metastases, to assess diagnostic performance. Accuracy of bone metastasis detection and area under the curve (AUC) were calculated. Original, Gaussian-filtered, and DL-filtered images were compared with Reference images.Results PSNR and SSIM for DL-filtered images were highest in all count percentages. BONENAVI analysis values for DL-filtered images did not differ significantly regardless of the presence or absence of bone metastases. BONENAVI analysis values for Original and Gaussian-filtered images differed significantly at < 25% counts in patients without bone metastases. In patients with bone metastases, BSI and Hs for Original and Gaussian-filtered images differed significantly at < 10% counts, whereas ANN values did not. Accuracy of bone metastasis detection was highest for DL-filtered images in all count percentages; AUC did not differ significantly. Accuracy of Original and Gaussian-filtered images worsened with decreasing count percentage; AUC differed significantly for Original images at < 50% counts and for Gaussian-filtered images at < 25% counts.Conclusions Our deep learning model improved image quality and bone metastasis detection accuracy for low-count bone scintigraphy whole-body images, suggesting its applicability in clinical practice.