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

Muller, Florence M; Maebe, Jens; Vanhove, Christian; Vandenberghe, Stefaan

doi:10.1002/mp.16385

Cited by 5 publications

(1 citation statement)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Dose reduction strategies with DL denoising are also applicable in micro-CT imaging. In our previous work [19] a neural network was trained and evaluated for denoising low dose micro-CT acquisitions to realize higher quality micro-CT imaging at reduced doses (by a factor of 3). In preclinical research, low dose multi-modality studies offer promising future prospects for managing the cumulative effects of radiation in longitudinal imaging studies.…”

Section: Discussionmentioning

confidence: 99%

Image Denoising of Low Dose PET Mouse Scans with Deep Learning: Validation Study for Cross-Tracer Preclinical Imaging Applicability

Muller

Vervenne

Maebe

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Purpose: Positron Emission Tomography (PET) image quality can be improved by higher injected activity and/or longer acquisition time, but both may often not be practical in preclinical imaging. Common preclinical radioactive doses (10 MBq) have been shown to cause deterministic changes in biological pathways. Reducing the injected tracer activity and/or shortening the scan time inevitably results in low-count acquisitions which poses a challenge because of the inherent noise introduction. We present an image-based deep learning (DL) framework for denoising lower count micro-PET images. Procedures: For 36 mice, a 15-min 18F-FDG (8.15 ± 1.34 MBq) PET scan was acquired at 40 min post-injection on the Molecubes b-CUBE (in list mode). The 15-min acquisition (high-count) was parsed into smaller time fractions of 7.50, 3.75, 1.50 and 0.75 mins to emulate images reconstructed at 50, 25, 10 and 5% of the full counts, respectively. A 2D U-Net was trained with mean-squared-error loss on 28 high-low count image pairs. Results: The DL algorithms were visually and quantitatively compared to spatial and edge-preserving denoising filters; the DL-based methods effectively removed image noise and recovered image details much better while keeping quantitative (SUV) accuracy. The largest improvement in image quality was seen in the images reconstructed with 10 and 5% of the counts (equivalent to sub-1-MBq or sub-1-min mouse imaging). The DL-based denoising framework was also successfully applied on the NEMA-NU4 phantom and different tracer studies (18F-PSMA, 18F-FAPI and 68Ga-FAPI). Conclusion: Visual and quantitative results support the superior performance and robustness in image denoising of the implemented DL models for low statistics micro-PET. This offers tremendous flexibility in optimizing scan protocols with reduced tracer doses or shorter durations.

show abstract

Section: Discussionmentioning

confidence: 99%

Image Denoising of Low Dose PET Mouse Scans with Deep Learning: Validation Study for Cross-Tracer Preclinical Imaging Applicability

Muller

Vervenne

Maebe

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Image Denoising of Low-Dose PET Mouse Scans with Deep Learning: Validation Study for Preclinical Imaging Applicability

Muller,

Vervenne,

Maebe

et al. 2023

Mol Imaging Biol

Self Cite

View full text Add to dashboard Cite

Positron Emission Tomography (PET) image quality can be improved by higher injected activity and/or longer acquisition time, but both may often not be practical in preclinical imaging. Common preclinical radioactive doses (10 MBq) have been shown to cause deterministic changes in biological pathways.Reducing the injected tracer activity and/or shortening the scan time inevitably results in low-count acquisitions which poses a challenge because of the inherent noise introduction. We present an imagebased deep learning (DL) framework for denoising lower count micro-PET images. Procedures:For 36 mice, a 15-min 18 F-FDG (8.15 ± 1.34 MBq) PET scan was acquired at 40 min post-injection on the Molecubes b-CUBE (in list mode). The 15-min acquisition (high-count) was parsed into smaller time fractions of 7.50, 3.75, 1.50 and 0.75 mins to emulate images reconstructed at 50, 25, 10 and 5% of the full counts, respectively. A 2D U-Net was trained with mean-squared-error loss on 28 high-low count image pairs. Results:The DL algorithms were visually and quantitatively compared to spatial and edge-preserving denoising lters; the DL-based methods effectively removed image noise and recovered image details much better while keeping quantitative (SUV) accuracy. The largest improvement in image quality was seen in the images reconstructed with 10 and 5% of the counts (equivalent to sub-1-MBq or sub-1-min mouse imaging). The DL-based denoising framework was also successfully applied on the NEMA-NU4 phantom and different tracer studies ( 18 F-PSMA, 18 F-FAPI and 68 Ga-FAPI). Conclusion:Visual and quantitative results support the superior performance and robustness in image denoising of the implemented DL models for low statistics micro-PET. This offers tremendous exibility in optimizing scan protocols with reduced tracer doses or shorter durations.

show abstract

Recent progress of ultrasound-responsive titanium dioxide sonosensitizers in cancer treatment

Cui,

Zhu,

Yue

et al. 2024

Chinese Chemical Letters

View full text Add to dashboard Cite

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

Abstract: Investigate the use of deep learning (DL) approaches for denoising low dose micro-CT images. 2. Compare the performance with other methods to show and quantify the impact of DL-based image enhancement on micro-CT datasets.

Cited by 5 publications

References 41 publications

Image Denoising of Low Dose PET Mouse Scans with Deep Learning: Validation Study for Cross-Tracer Preclinical Imaging Applicability

Image Denoising of Low Dose PET Mouse Scans with Deep Learning: Validation Study for Cross-Tracer Preclinical Imaging Applicability

Image Denoising of Low-Dose PET Mouse Scans with Deep Learning: Validation Study for Preclinical Imaging Applicability

Recent progress of ultrasound-responsive titanium dioxide sonosensitizers in cancer treatment

Contact Info

Product

Resources

About