Purpose: To assess the physical performance of deep learning image reconstruction (DLIR) compared with those of filtered back projection (FBP) and iterative reconstruction (IR) and to estimate the dose reduction potential of the technique.
Approach: A cylindrical water bath phantom with a diameter of 300 mm including two rods composed of acrylic and soft tissue-equivalent material was scanned using a clinical computed tomography (CT) scanner at four dose levels (CT dose index of 20, 15, 10, and 5 mGy). Phantom images were reconstructed using FBP, DLIR, and IR. The in-plane and z axis task transfer functions (TTFs) and in-plane noise power spectrum (NPS) were measured. The dose reduction potential was estimated by evaluating the system performance function calculated from TTF and NPS. The visibilities of a bar pattern phantom placed in the same water bath phantom were compared.
Results: The use of DLIR resulted in a notable decrease in noise magnitude. The shift in peak NPS frequency was reduced compared with IR. Preservation of in-plane TTF was superior using DLIR than using IR. The estimated dose reduction potentials of DLIR and IR were 39% to 54% and 19% to 29%, respectively. However, the z axis resolution was decreased with DLIR by 6% to 21% compared with FBP. The bar pattern visibilities were approximately consistent with the TTF results in both planes.
Conclusions: The in-plane edge-preserving noise reduction performance of DLIR is superior to that of IR. Moreover, DLIR enables approximately half-dose acquisitions with no deterioration in noise texture in cases that permit some z axis resolution reduction.
Arm artifact, which is type of streak artifact frequently observed in computed tomography (CT) images of polytrauma patients at the arms-down positioning, are known to degrade the image quality. The existing streak artifact reduction algorithms are not effective for arm artifact, as they have not been designed for this purpose. The effects of the latest iterative reconstruction techniques (IRs), which are effective for noise and streak artifact reductions, have not been evaluated for the arm-artifact reduction. In this study, we developed a novel reconstruction algorithm for arm-artifact reduction using an arm-induced noise filtering of the projection data. A phantom resembling a standard adult abdomen with two arms was scanned using a 16-row CT scanner, and then the projection data was downloaded. The proposed algorithm consisted of an arm recognition step after the conventional reconstruction and arm-induced noise filtering (frequency split and attenuation-dependent filtering) of the projection data. The artifact reduction capabilities and image blurring as a side effect of the filtering were compared with those of the latest three IRs (IR1, IR2, and IR3). The proposed algorithm and IR1 significantly reduced the artifacts by 89.4% and 83.5%, respectively. The other two IRs were not effective in terms of arm-artifact reduction. In contrast to IR1 that yielded an apparent image blurring combined with a different noise texture, the proposed algorithm mostly suppressed the image blurring. The proposed algorithm, designed for an arm-artifact reduction, was effective and it is expected to improve the image quality of abdominal CT examinations at the arms-down positioning.
The aim of this study was to propose a calibration method for small dosimeters to measure absorbed doses during dual- source dual-energy computed tomography (DECT) and to compare the axial dose distribution, eye lens dose, and image noise level between DE and standard, single-energy (SE) head CT angiography. Three DE (100/Sn140 kVp 80/Sn140 kVp, and 140/80 kVp) and one SE (120 kVp) acquisitions were performed using a second-generation dual-source CT device and a female head phantom, with an equivalent volumetric CT dose index. The axial absorbed dose distribution at the orbital level and the absorbed doses for the eye lens were measured using radiophotoluminescent glass dosimeters. CT attenuation numbers were obtained in the DE composite images and the SE images of the phantom at the orbital level. The doses absorbed at the orbital level and in the eye lens were lower and standard deviations for the CT attenuation numbers were slightly higher in the DE acquisitions than those in the SE acquisition. The anterior surface dose was especially higher in the SE acquisition than that in the DE acquisitions. Thus, DE head CT angiography can be performed with a radiation dose lower than that required for a standard SE head CT angiography, with a slight increase in the image noise level. The 100/Sn140 kVp acquisition revealed the most balanced axial dose distribution. In addition, our proposed method was effective for calibrating small dosimeters to measure absorbed doses in DECT.
An image-processing technique for separating bones from soft tissue in static chest radiographs has been developed. The present study was performed to evaluate the usefulness of dynamic bone images in quantitative analysis of rib movement. Dynamic chest radiographs of 16 patients were obtained using a dynamic flat-panel detector and processed to create bone images by using commercial software (Clear Read BS, Riverain Technologies). Velocity vectors were measured in local areas on the dynamic images, which formed a map. The velocity maps obtained with bone and original images for scoliosis and normal cases were compared to assess the advantages of bone images. With dynamic bone images, we were able to quantify and distinguish movements of ribs from those of other lung structures accurately. Limited rib movements of scoliosis patients appeared as a reduced rib velocity field, resulting in an asymmetrical distribution of rib movement. Vector maps in all normal cases exhibited left/right symmetric distributions of the velocity field, whereas those in abnormal cases showed asymmetric distributions because of locally limited rib movements. Dynamic bone images were useful for accurate quantitative analysis of rib movements. The present method has a potential for an additional functional examination in chest radiography.
Rib movement during respiration is one of the diagnostic criteria in pulmonary impairments. In general, the rib movement is assessed in fluoroscopy. However, the shadows of lung vessels and bronchi overlapping ribs prevent accurate quantitative analysis of rib movement. Recently, an image-processing technique for separating bones from soft tissue in static chest radiographs, called “bone suppression technique”, has been developed. Our purpose in this study was to evaluate the usefulness of dynamic bone images created by the bone suppression technique in quantitative analysis of rib movement. Dynamic chest radiographs of 10 patients were obtained using a dynamic flat-panel detector (FPD). Bone suppression technique based on a massive-training artificial neural network (MTANN) was applied to the dynamic chest images to create bone images. Velocity vectors were measured in local areas on the dynamic bone images, which formed a map. The velocity maps obtained with bone and original images for scoliosis and normal cases were compared to assess the advantages of bone images. With dynamic bone images, we were able to quantify and distinguish movements of ribs from those of other lung structures accurately. Limited rib movements of scoliosis patients appeared as reduced rib velocity vectors. Vector maps in all normal cases exhibited left-right symmetric distributions, whereas those in abnormal cases showed nonuniform distributions. In conclusion, dynamic bone images were useful for accurate quantitative analysis of rib movements: Limited rib movements were indicated as a reduction of rib movement and left-right asymmetric distribution on vector maps. Thus, dynamic bone images can be a new diagnostic tool for quantitative analysis of rib movements without additional radiation dose.
Reduction of exposure dose and improvement in image quality can be expected to result from advances in the
performance of imaging detectors. We propose a computerized method for determining optimized imaging conditions by
use of simulated images. This study was performed to develop a prototype system for image noise and to ensure
consistency between the resulting images and actual images. An RQA5 X-ray spectrum was used for determination of
input-output characteristics of a flat-panel detector (FPD). The number of incident quantum to the detector per pixel
(counts/pixel) was calculated according to the pixel size of the detector and the quantum number in RQA5 determined in
IEC6220-1. The relationship among tube current-time product (mAs), exposure dose (C/kg) at the detector surface, the
number of incident quanta (counts/pixel), and pixel values measured on the images was addressed, and a conversion
function was then created. The images obtained by the FPD was converted into a map of incident quantum numbers and
input into random-value generator to simulate image noise. In addition, graphic user interface was developed to observe
images with changing image noise and exposure dose levels, which have trade-off relationship. Simulation images
provided at different noise levels were compared with actual images obtained by the FPD system. The results indicated
that image noise was simulated properly both in objective and subjective evaluation. The present system could allow us
to determine necessary dose from image quality and also to estimate image quality from any exposure dose.
This report presents the fundamental temporospatial characteristics of a dynamic flat-panel detector (FPD) system. We
investigated the relationship between pixel value and X-ray pulse output, and examined reproducibility, dependence on
pulse width, tube voltage, and pulse rate. Sequential images were obtained using a direct conversion-type dynamic FPD.
The exposure conditions were: 110 kV, 80 mA, 6.3 ms, 7.5 fps,
source-to-image distance (SID) 1.5 m. X-ray pulse
output was measured using a dosimetry system with a sampling interval of 70 μs, to determine temporal changes in each
X-ray pulse output. Temporal changes in pixel value were measured in the obtained images, and the relationship
between pixel value and X-ray pulse output was examined. Reproducibility was assessed by comparing the results in two
sequential images obtained under the same exposure conditions. Moreover, the relationships and properties were
evaluated by changing the pulse width (12 ms and 25 ms), tube voltage (80 kV, 90 kV, and 100 kV), and pulse rate (3.75
fps and 15 fps). The results showed a good correlation between the
X-ray pulse output and pixel values. Fluctuation of
the pixel value measured in sequential images is thought to be mainly due to changes in X-ray pulse output, and is not
caused by FPD.
There is a concern that image lag may reduce accuracy of real-time target tracking in radiotherapy. This study was
performed to investigate influence of image lag on the accuracy of target tracking in radiotherapy. Fluoroscopic images
were obtained using a direct type of dynamic flat-panel detector (FPD) system under conditions of target tracking during
radiotherapy. The images continued to be read out after
X-irradiations and cutoff, and image lag properties in the system
were then determined. Subsequently, a tungsten materials plate with a precision edge was mounted on to a motor control
device, which provided a constant velocity. The plate was moved into the center of the detector at movement rate of 10
and 20 mm/s, covering lung tumor movement of normal breathing, and MTF and profile curves were measured on the
edges covering and uncovering the detector. A lung tumor with blurred edge due to image lag was simulated using the
results and then superimposed on breathing chest radiographs of a patient. The moving target with and without image lag
was traced using a template-matching technique. In the results, the target could be traced within a margin for error in
external radiotherapy. The results indicated that there was no effect of image lag on target tracking in usual breathing
speed in a radiotherapy situation. Further studies are required to investigate influence by the other factors, such as
exposure dose, target size and shape, imaging rate, and thickness of a patient's body.