MP18-02: Does Deep Learning Reconstruction Improve Ureteral Stone Detection and Subjective Image Quality in the CT Images of Patients with Metal Hardware?
Friday, May 3, 2024 3:30 PM to 5:30 PM · 2 hr. (US/Central)
302B
Abstract
Information
Full Abstract and Figures
Author Block
Ruben Crew*, Jason Smith, Mohammad Kassir, Ala'a Farkouh, Kai Wen Cheng, Bertha Escobar-Poni, Jun Ho Chung, Uy Lae Kim, Jammie-Lyn Quines, Grant Sajdak, Katya Hanessian, Sikai Song, Akin S. Amasyali, Zhamshid Okhunov, Udochukwu Oyoyo, D. Daniel Baldwin, Kerby Oberg, D. Duane Baldwin, Loma Linda, CA
Introduction
Interpreting CT scans in patients with metal hardware may be challenging due to the metal artifact causing image noise, particularly when lower radiation doses are utilized. Those with metal prostheses presenting with signs and symptoms of urinary stones may therefore pose a diagnostic challenge. The purpose of this study was to compare ureteral stone detection and image quality of CT scans with and without deep learning reconstruction (DLR) and metal artifact reduction (MAR), at different radiation doses in the presence of metal hip prostheses.
Methods
Ten urinary system combinations (each with a different combination of ureteral stones sized from 4-6 mm) were separately implanted into a cadaver with bilateral hip prostheses. Each set was scanned under three different radiation doses (Conventional Dose=141 mAs, Low Dose (LD)=30 mAs, and Ultra-low Dose (ULD)=7.0 mAs). For each dose, two scans were obtained: one with DLR and MAR, and a second scan with no additional reconstruction. Utilizing a modified 5-point Likert scale, two blinded radiologists reviewed images and ranked each image in terms of Artifact, Image Noise, Image Sharpness, Overall Quality, and Diagnostic Confidence. The sensitivity at each setting for stone detection was determined.
Results
ULD with DLR and MAR resulted in significantly improved subjective image quality in all 5 measured domains (p<0.05 for all) compared to ULD (Figure 1). For conventional and LD, DLR and MAR significantly improved image quality only in the artifact domain (p<0.05 for both). The sensitivity of ULD for stone detection increased from 25.7% to 50% when DLR and MAR was applied, however this improvement was not significant (p=0.2). The sensitivity of ULD with DLR and MAR was comparable to conventional dose CT (50% vs 57%; p=0.7).
Conclusions
The application of DLR with MAR to ULD resulted in improved subjective image quality across all domains and provided sensitivity comparable to conventional dose CT for stone detection in patients with hip prostheses. Use of DLR with MAR may allow the application of low dose protocols in patients with hip prostheses.
Source Of Funding
None
Author Block
Ruben Crew*, Jason Smith, Mohammad Kassir, Ala'a Farkouh, Kai Wen Cheng, Bertha Escobar-Poni, Jun Ho Chung, Uy Lae Kim, Jammie-Lyn Quines, Grant Sajdak, Katya Hanessian, Sikai Song, Akin S. Amasyali, Zhamshid Okhunov, Udochukwu Oyoyo, D. Daniel Baldwin, Kerby Oberg, D. Duane Baldwin, Loma Linda, CA
Introduction
Interpreting CT scans in patients with metal hardware may be challenging due to the metal artifact causing image noise, particularly when lower radiation doses are utilized. Those with metal prostheses presenting with signs and symptoms of urinary stones may therefore pose a diagnostic challenge. The purpose of this study was to compare ureteral stone detection and image quality of CT scans with and without deep learning reconstruction (DLR) and metal artifact reduction (MAR), at different radiation doses in the presence of metal hip prostheses.
Methods
Ten urinary system combinations (each with a different combination of ureteral stones sized from 4-6 mm) were separately implanted into a cadaver with bilateral hip prostheses. Each set was scanned under three different radiation doses (Conventional Dose=141 mAs, Low Dose (LD)=30 mAs, and Ultra-low Dose (ULD)=7.0 mAs). For each dose, two scans were obtained: one with DLR and MAR, and a second scan with no additional reconstruction. Utilizing a modified 5-point Likert scale, two blinded radiologists reviewed images and ranked each image in terms of Artifact, Image Noise, Image Sharpness, Overall Quality, and Diagnostic Confidence. The sensitivity at each setting for stone detection was determined.
Results
ULD with DLR and MAR resulted in significantly improved subjective image quality in all 5 measured domains (p<0.05 for all) compared to ULD (Figure 1). For conventional and LD, DLR and MAR significantly improved image quality only in the artifact domain (p<0.05 for both). The sensitivity of ULD for stone detection increased from 25.7% to 50% when DLR and MAR was applied, however this improvement was not significant (p=0.2). The sensitivity of ULD with DLR and MAR was comparable to conventional dose CT (50% vs 57%; p=0.7).
Conclusions
The application of DLR with MAR to ULD resulted in improved subjective image quality across all domains and provided sensitivity comparable to conventional dose CT for stone detection in patients with hip prostheses. Use of DLR with MAR may allow the application of low dose protocols in patients with hip prostheses.
Source Of Funding
None
Sessions
MP18: Imaging/Uroradiology I
302B