Deep learning-based reconstruction can improve the image quality of low radiation dose head CT.

OBJECTIVES: To evaluate the image quality of deep learning-based reconstruction (DLR), model-based (MBIR), and hybrid iterative reconstruction (HIR) algorithms for lower-dose (LD) unenhanced head CT and compare it with those of standard-dose (STD) HIR images. METHODS: This retrospective study included 114 patients who underwent unenhanced head CT using the STD (n = 57) or LD (n = 57) protocol on a 320-row CT. STD images were reconstructed with HIR; LD images were reconstructed with HIR (LD-HIR), MBIR (LD-MBIR), and DLR (LD-DLR). The image noise, gray and white matter (GM-WM) contrast, and contrast-to-noise ratio (CNR) at the basal ganglia and posterior fossa levels were quantified. The noise magnitude, noise texture, GM-WM contrast, image sharpness, streak artifact, and subjective acceptability were independently scored by three radiologists (1 = worst, 5 = best). The lesion conspicuity of LD-HIR, LD-MBIR, and LD-DLR was ranked through side-by-side assessments (1 = worst, 3 = best). Reconstruction times of three algorithms were measured. RESULTS: The effective dose of LD was 25% lower than that of STD. Lower image noise, higher GM-WM contrast, and higher CNR were observed in LD-DLR and LD-MBIR than those in STD (all, p ≤ 0.035). Compared with STD, the noise texture, image sharpness, and subjective acceptability were inferior for LD-MBIR and superior for LD-DLR (all, p < 0.001). The lesion conspicuity of LD-DLR (2.9 ± 0.2) was higher than that of HIR (1.2 ± 0.3) and MBIR (1.8 ± 0.4) (all, p < 0.001). Reconstruction times of HIR, MBIR, and DLR were 11 ± 1, 319 ± 17, and 24 ± 1 s, respectively. CONCLUSION: DLR can enhance the image quality of head CT while preserving low radiation dose level and short reconstruction time. KEY POINTS: • For unenhanced head CT, DLR reduced the image noise and improved the GM-WM contrast and lesion delineation without sacrificing the natural noise texture and image sharpness relative to HIR. • The subjective and objective image quality of DLR was better than that of HIR even at 25% reduced dose without considerably increasing the image reconstruction times (24 s vs. 11 s). • Despite the strong noise reduction and improved GM-WM contrast performance, MBIR degraded the noise texture, sharpness, and subjective acceptance with prolonged reconstruction times relative to HIR, potentially hampering its feasibility.

Super-resolution deep-learning reconstruction for cardiac CT: impact of radiation dose and focal spot size on task-based image quality.

This study aimed to evaluate the impact of radiation dose and focal spot size on the image quality of super-resolution deep-learning reconstruction (SR-DLR) in comparison with iterative reconstruction (IR) and normal-resolution DLR (NR-DLR) algorithms for cardiac CT. Catphan-700 phantom was scanned on a 320-row scanner at six radiation doses (small and large focal spots at 1.4-4.3 and 5.8-8.8 mGy, respectively). Images were reconstructed using hybrid-IR, model-based-IR, NR-DLR, and SR-DLR algorithms. Noise properties were evaluated through plotting noise power spectrum (NPS). Spatial resolution was quantified with task-based transfer function (TTF); Polystyrene, Delrin, and Bone-50% inserts were used for low-, intermediate, and high-contrast spatial resolution. The detectability index (d') was calculated. Image noise, noise texture, edge sharpness of low- and intermediate-contrast objects, delineation of fine high-contrast objects, and overall quality of four reconstructions were visually ranked. Results indicated that among four reconstructions, SR-DLR yielded the lowest noise magnitude and NPS peak, as well as the highest average NPS frequency, TTF50%, d' values, and visual rank at each radiation dose. For all reconstructions, the intermediate- to high-contrast spatial resolution was maximized at 4.3 mGy, while the lowest noise magnitude and highest d' were attained at 8.8 mGy. SR-DLR at 4.3 mGy exhibited superior noise performance, intermediate- to high-contrast spatial resolution, d' values, and visual rank compared to the other reconstructions at 8.8 mGy. Therefore, SR-DLR may yield superior diagnostic image quality and facilitate radiation dose reduction compared to the other reconstructions, particularly when combined with small focal spot scanning.

[Reduction of radiation exposure and image quality using dose reduction tool on computed tomography fluoroscopy].

The purpose of our study was to measure the reduction rate of radiation dose and variability of image noise using the angular beam modulation (ABM) on computed tomography (CT) fluoroscopy. The Alderson-Rando phantom and the homemade phantom were used in our study. These phantoms were scanned at on-center and off-center positions at -12 cm along y-axis with and without ABM technique. Regarding the technique, the x-ray tube is turned off in a 100-degree angle sector at the center of 12 o'clock, 10 o'clock, and 2 o'clock positions during CT fluoroscopy. CT fluoroscopic images were obtained with tube voltages, 120 kV; tube current-time product per reconstructed image, 30 mAs; rotation time, 0.5 s/rot; slice thickness, 4.8 mm; and reconstruction kernel B30s in each scanning. After CT scanning, radiation exposure and image noise were measured and the image artifacts were evaluated with and without the technique. The reduction rate for radiation exposure was 75-80% with and without the technique at on-center position regardless of each angle position. In the case of the off-center position at -12 cm, the reduction rate was 50% with and without the technique. In contrast, image noise remained constant with and without the technique. Visual inspection for image artifacts almost have the same scores with and without the technique and no statistical significance was found in both techniques (p>0.05). ABM is an appropriate tool for reducing radiation exposure and maintaining image-noise and artifacts during CT fluoroscopy.

Assessment of cardiac implantable electric device lead perforation using a metal artifact reduction algorithm in cardiac computed tomography.

PURPOSE: CT is considered the non-invasive gold standard for evaluating cardiac implantable electronic devices (CIEDs) lead perforation, but metal artifacts caused by the lead tip affect the image quality and make a definitive diagnosis challenging. We compared the performances of the metal artifact reduction (MAR) algorithm and the conventional algorithm for identification of the right ventricular (RV) lead tip position in cardiac CT studies of patients with CIEDs. METHOD: Forty-seven consecutive patients (26 men; age 70.3 ±â€¯15.4 years) with CIEDs underwent cardiac CT. Using the conventional and MAR algorithm, two image reconstructions were performed for each scan. We calculated the artifact index (AI) to assess the quantitative capability of the MAR algorithm for artifact reduction and visually assessed the RV lead tip position on both images as follows: non-perforation, perforation, and equivocal. RESULTS: The mean AIs were significantly lower with the MAR algorithm than with the conventional algorithm (96.7 ±â€¯40.1 HU vs. 284.6 ±â€¯134.1 HU, P < 0.001). Thirteen (27.7 %) patients were diagnosed as equivocal using the conventional algorithm but were diagnosed with perforation (2 patients) and non-perforation (11 patients) using the MAR algorithm (equivocal rate: 27.7 % vs. 0%, P < 0.001). Using the MAR algorithm, all cases were diagnosed with perforation (6 patients, 12.8 %) or non-perforation (41 patients, 87.2 %). CONCLUSIONS: The MAR algorithm effectively reduced metal artifacts and allowed us to diagnose the presence or absence of perforation in all cases, whereas definitive diagnosis was difficult with the use of conventional algorithm in 27.7 % of cases.

[A fundamental study of non-contrast enhanced MR angiography using ECG gated-3D fast spin echo at 3.0 T].

Contrast-enhanced magnetic resonance angiography (CE-MRA) is frequently performed in body and extremity studies because of its superior ability to detect the vascular stenosis. However, nephrotoxicity of the contrast medium has been emphasized in recent years. Non-contrast MRA using the three-dimensional electrocardiogram-synchronized fast spin echo method (FBI, NATIVE and TRANCE) is recommended as a substitute for CE-MRA. There are a few reports in the literature that evaluate the detectability of vascular stenosis using non-contrast MRA on 3.0 T MRI. The purpose of this study was to evaluate the detectability of vascular stenosis using non-contrast MRA at 3.0 T with an original vascular phantom. The vascular phantom consisted of silicon tubes. 30% and 70% stenosis of luminal diameter were made. Each silicon tube connected a pump producing a pulsatile flow. A flowing material to was used in this study to show the similarity of the intensity to blood on MRI. MRA without a contrast medium (NATIVE sequence) were performed in the vascular phantom by changing the image matrix, static magnetic field strength and flow velocity. In addition, the NATIVE sequence was used with or without flow compensation. Vascular stenosis was quantitatively estimated by measurement of the signal intensities in non-contrast MRA images. MRA with NATIVE sequence demonstrated an accurate estimation of 30% vascular stenosis at slow flow velocity. However, 30% stenosis was overestimated in cases of high flow velocity. Estimation was improved by using a flow compensation sequence. 70% stenosis was overestimated on MRA with NATIVE sequence. Estimation of 70% stenosis was improved by using a flow compensation sequence. Accurate estimation of vascular stenosis in MRA with a NATIVE sequence is improved by using the flow compensation technique. MRA with NATIVE sequence is considered to be a promising method for the evaluation of patients with severe renal dysfunction as a substitute for CT angiography or CE-MRA.

Optimization of acquisition parameters and accuracy of target motion trajectory for four-dimensional cone-beam computed tomography with a dynamic thorax phantom.

Our purpose in this study was to evaluate the performance of four-dimensional computed tomography (4D-CBCT) and to optimize the acquisition parameters. We evaluated the relationship between the acquisition parameters of 4D-CBCT and the accuracy of the target motion trajectory using a dynamic thorax phantom. The target motion was created three dimensionally using target sizes of 2 and 3 cm, respiratory cycles of 4 and 8 s, and amplitudes of 1 and 2 cm. The 4D-CBCT data were acquired under two detector configurations: "small mode" and "medium mode". The projection data acquired with scan times ranging from 1 to 4 min were sorted into 2, 5, 10, and 15 phase bins. The accuracy of the measured target motion trajectories was evaluated by means of the root mean square error (RMSE) from the setup values. For the respiratory cycle of 4 s, the measured trajectories were within 2 mm of the setup values for all acquisition times and target sizes. Similarly, the errors for the respiratory cycle of 8 s were <4 mm. When we used 10 or more phase bins, the measured trajectory errors were within 2 mm of the setup values. The trajectory errors for the two detector configurations showed similar trends. The acquisition times for achieving an RMSE of 1 mm for target sizes of 2 and 3 cm were 2 and 1 min, respectively, for respiratory cycles of 4 s. The results obtained in this study enable optimization of the acquisition parameters for target size, respiratory cycle, and desired measurement accuracy.