Background: Advances in computed tomography (CT) image reconstruction have introduced deep learning–based algorithms such as Deep Learning Image Reconstruction (DLIR), designed to enhance image quality while potentially lowering radiation dose compared with conventional iterative reconstruction (IR) methods. Purpose: To compare DLIR and IR algorithms in computed tomography angiography (CTA) and coronary CTA (CCTA), assessing their impact on radiation dose and image quality. Materials and Methods: This retrospective study included 58 patients who underwent two comparable CTA or CCTA examinations between November 2024 and October 2025 at the University Hospital of Padua—one reconstructed with DLIR and the other with an IR algorithm. Effective dose (ED), image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were quantitatively analysed. Results: The analysis did not show a statistically significant reduction in effective dose with DLIR compared to IR (p = 0.2241), but demonstrated a statistically significant improvement in image quality: lower noise (18.6 vs 26 HU, p < 0.0001), and higher SNR and CNR (p < 0.0001). Conclusion: DLIR enables a significant improvement in image quality compared with iterative reconstruction, while not demonstrating a significant reduction in radiation dose. These findings support the integration of the DLIR algorithm into cardiovascular imaging protocols, particularly for patients requiring repeated follow-up examinations, although further prospective studies on larger and more homogeneous populations are warranted.
Introduzione: Le tecniche di ricostruzione delle immagini in tomografia computerizzata (CT) stanno rapidamente evolvendo con l’introduzione di algoritmi basati sull’intelligenza artificiale. In particolare, l'algoritmo Deep Learning Image Reconstruction (DLIR) è stato proposto come alternativa alle tecniche iterative convenzionali (IR), con l’obiettivo di migliorare la qualità dell’immagine e ridurre la dose di radiazioni. Scopo: Confrontare DLIR con gli algoritmi di ricostruzione iterativa (IR) nelle angiografie TC (CTA) e coronariche (CCTA), valutandone l’impatto sulla dose efficace e sulla qualità dell’immagine. Materiali e Metodi: Sono stati selezionati retrospettivamente 58 pazienti sottoposti a CTA o CCTA presso l’Azienda Ospedaliera di Padova tra novembre 2024 e ottobre 2025, ciascuno con due esami eseguiti rispettivamente con DLIR e IR. Per ogni esame sono stati analizzati dose efficace (ED), rumore, rapporto segnale/rumore (SNR) e rapporto contrasto/rumore (CNR). Risultati: L’analisi non ha mostrato una riduzione significativa della dose efficace con DLIR rispetto IR (p = 0.2241), ma un miglioramento statisticamente significativo della qualità dell’immagine: rumore inferiore (18.6 vs 26 HU, p < 0.0001), SNR e CNR superiori (p < 0.0001). Conclusioni: DLIR consente un miglioramento significativo della qualità delle immagini rispetto alle ricostruzioni iterative, pur non dimostrando significativa riduzione della dose di radiazione. Questi risultati supportano l’integrazione dell’algoritmo DLIR nei protocolli di imaging cardiovascolare, in particolare nei pazienti che richiedono controlli ripetuti, pur richiedendo ulteriori studi prospettici su popolazioni più ampie e omogenee.
CT Effective Dose and Image Quality in CCTA and CTA: comparison between Deep Learning Image Reconstruction and Iterative Algorithms
AGOSTINI, ELENA
2023/2024
Abstract
Background: Advances in computed tomography (CT) image reconstruction have introduced deep learning–based algorithms such as Deep Learning Image Reconstruction (DLIR), designed to enhance image quality while potentially lowering radiation dose compared with conventional iterative reconstruction (IR) methods. Purpose: To compare DLIR and IR algorithms in computed tomography angiography (CTA) and coronary CTA (CCTA), assessing their impact on radiation dose and image quality. Materials and Methods: This retrospective study included 58 patients who underwent two comparable CTA or CCTA examinations between November 2024 and October 2025 at the University Hospital of Padua—one reconstructed with DLIR and the other with an IR algorithm. Effective dose (ED), image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were quantitatively analysed. Results: The analysis did not show a statistically significant reduction in effective dose with DLIR compared to IR (p = 0.2241), but demonstrated a statistically significant improvement in image quality: lower noise (18.6 vs 26 HU, p < 0.0001), and higher SNR and CNR (p < 0.0001). Conclusion: DLIR enables a significant improvement in image quality compared with iterative reconstruction, while not demonstrating a significant reduction in radiation dose. These findings support the integration of the DLIR algorithm into cardiovascular imaging protocols, particularly for patients requiring repeated follow-up examinations, although further prospective studies on larger and more homogeneous populations are warranted.| File | Dimensione | Formato | |
|---|---|---|---|
|
Agostini_Elena.pdf
Accesso riservato
Dimensione
1.32 MB
Formato
Adobe PDF
|
1.32 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/97562