Iterative reconstruction algorithm improves the image quality without affecting quantitative measurements of computed tomography perfusion in the upper abdomen

A Novel Self-Supervised Learning-Based Method for Dynamic CT Brain Perfusion Imaging

Dynamic computed tomography (CT)-based brain perfusion imaging is a non-invasive technique that can provide quantitative measurements of cerebral blood flow (CBF), cerebral blood volume (CBV), and mean transit time (MTT). However, due to high radiation dose, dynamic CT scan with a low tube voltage and current protocol is commonly used. Because of this reason, the increased noise degrades the quality and reliability of perfusion maps. In this study, we aim to propose and investigate the feasibility of utilizing a convolutional neural network and a bi-directional long short-term memory model with an attention mechanism to self-supervisedly yield the impulse residue function (IRF) from dynamic CT images. Then, the predicted IRF can be used to compute the perfusion parameters. We evaluated the performance of the proposed method using both simulated and real brain perfusion data and compared the results with those obtained from two existing methods: singular value decomposition and tensor total-variation. The simulation results showed that the overall performance of parameter estimation obtained from the proposed method was superior to that obtained from the other two methods. The experimental results showed that the perfusion maps calculated from the three studied methods were visually similar, but small and significant differences in perfusion parameters between the proposed method and the other two methods were found. We also observed that there were several low-CBF and low-CBV lesions (i.e., suspected infarct core) found by all comparing methods, but only the proposed method revealed longer MTT. The proposed method has the potential to self-supervisedly yield reliable perfusion maps from dynamic CT images.

Multiorientation Simultaneous Computation of Back-Projection CT Image Reconstruction Algorithm in Staging Diagnosis of Bladder Cancer

The objective of this research was to investigate the multidirectional synchronous calculation of the back-projection computed tomography (CT) image reconstruction algorithm (MSBP) in the staging diagnosis of bladder cancer. Sixty patients with bladder cancer admitted to the hospital were selected for enhanced CT scanning, all of which were randomly divided into control group (n = 30) and study group (n = 30). The filtered back-projection (FBP) algorithm was employed to reconstruct the scanned image, and the MSBP was additionally applied to the images of the study group. Fringe artifact (SA), overall mass (OQ), effective radiation dose (ED), CT dose-exponential volume (CTDI), and dose-length product (DLP) of the two groups of images were compared and analyzed. The results showed that the total time of the traditional algorithm was 5.473 s, and the total time of MSBP combined with FBP algorithm was 2.832 s, which was significantly higher than that of the traditional algorithm (P < 0.05). CT scan bladder cancer staging results of all patients were compared with surgical pathological staging results, and the results were evaluated according to the coincidence rate. SA in the study group was lower than that in the control group (P < 0.05), and OQ was not statistically significant. The ED of the study group was significantly lower than that of the control group by 33%. The coincidence rate of postoperative pathological staging results and CT staging results was 96%, and T1, T2a, and T4 coincidence rate was 100%, The coincidence rates of T2b, T3a, and T3b were 90%, 83.3%, and 66.67%, respectively. In summary, using MSBP method combined with FBP algorithm can improve OQ while reducing ED of patients. The introduction of MSBP into CT reconstruction image simplified the pixel location operation of projection calculation, showing an important application value in preoperative staging diagnosis of bladder cancer.

Early Diagnosis of Acute Ischemic Stroke by Brain Computed Tomography Perfusion Imaging Combined with Head and Neck Computed Tomography Angiography on Deep Learning Algorithm

The purpose of the research was to discuss the application values of deep learning algorithm-based computed tomography perfusion (CTP) imaging combined with head and neck computed tomography angiography (CTA) in the diagnosis of ultra-early acute ischemic stroke. Firstly, 88 patients with acute ischemic stroke were selected as the research objects and performed with cerebral CTP and CTA examinations. In order to improve the effect of image diagnosis, a new deconvolution network model AD-CNNnet based on deep learning was proposed and used in patient CTP image evaluation. The results showed that the peak signal-to-noise ratio (PSNR) and feature similarity (FSIM) of the AD-CNNnet method were significantly higher than those of traditional methods, while the normalized mean square error (NMSE) was significantly lower than that of traditional algorithms (P < 0.05). 80 cases were positive by CTP-CTA, including 16 cases of hyperacute ischemic stroke and 64 cases of acute ischemic stroke. The diagnostic sensitivity was 93.66%, and the specificity was 96.18%. The cerebral blood flow (CBF), cerebral blood volume (CBV), and the mean transit time (MTT) in the infarcted area were significantly greater than those in the corresponding healthy side area, and the time to peak (TTP) was significantly less than that in the corresponding healthy side area (P < 0.05). The cerebral perfusion parameters CBF, TTP, and MTT in the penumbra were significantly different from those in the infarct central area and the corresponding contralateral area, and TTP was the most sensitive (P < 0.05). To sum up, deep learning algorithm-based CTP combined with CTA could find the location of cerebral infarction lesions as early as possible to provide a reliable diagnostic result for the diagnosis of ultra-early acute ischemic stroke.

The impact of imaging time and contrast agent dose on screening for osteoporosis with contrast-enhanced CT

BACKGROUND

Screening for osteoporosis with contrast-enhanced computed tomography (CT) is promising for identifying high-risk osteoporotic patients. Our aims were (1) to investigate the estimated volume bone mineral density (vBMD) change over time after contrast injection (CT perfusion imaging, CTPI); and (2) to examine the influence of contrast dose on vBMD.

METHODS

Fifteen patients, aged 71 ± 9 years (mean ± standard deviation, range 55-86) underwent a CTPI examination (28 scans within 63 s) of the upper body followed (after a waiting time of 10 min) by a full 4-phase CT examination (4 scans within 4 min). The contrast dose for CTPI was 0.38-0.83 mL/kg, and for 4-phase CT was 0.87-1.29 mL/kg. Vertebrae L1-L3 were analysed totalling 43 vertebrae, using Mindways qCT Pro.

RESULTS

After contrast injection, vBMD showed a near-horizontal line until 17.5 s (non-contrast phase), followed by a steep increase 17.5-41.5 s after contrast injection, i.e., in the arterial phase, which plateaued 41.5 s after, i.e., in the early venous phase. A higher contrast dose per kg yielded significantly higher vBMD increase in both the arterial and venous phase (p < 0.003).

CONCLUSIONS

Both time from contrast administration and contrast dose per kg affected vBMD results. In arterial phase, the steepness of the curve makes vBMD estimation unsure. However, as values plateaued in the venous phase it might be possible to predict the correct vBMD values. Furthermore, contrast dose is a factor that needs to be adjusted for when using such a formula.

Dual-energy CT with virtual monoenergetic images to improve the visualization of pancreatic supplying arteries: the normal anatomy and variations

Background

Pancreatic ductal adenocarcinoma (PDAC) remains a malignancy with poor prognosis, appropriate surgical resection and neoadjuvant therapy depend on the accurate identification of pancreatic supplying arteries. We aim to evaluate the ability of monoenergetic images (MEI [+]) of dual-energy CT (DECT) to improve the visualization of pancreatic supplying arteries compared to conventional polyenergetic images (PEI) and investigate the implications of vascular variation in pancreatic surgery and transarterial interventions.

Results

One hundred patients without pancreatic diseases underwent DECT examinations were retrospectively enrolled in this study. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) at 40-keV MEI (+) were significantly higher than those of PEI (p < 0.05). All subjective MEI (+) scores were significantly higher than those of PEI (p < 0.05). The visualization rates were significantly higher for posterior superior pancreaticoduodenal artery (PSPDA), anterior and posterior inferior pancreaticoduodenal artery (AIPDA, PIPDA), anterior and posterior pancreaticoduodenal arcade (APAC, PPAC), transverse and caudal pancreatic artery (TPA, PCA) at 40-keV MEI (+) than those of PEI (p < 0.05). However, there were no significant differences for visualizing anterior superior pancreaticoduodenal artery (ASPDA), inferior pancreaticoduodenal artery (IPDA), dorsal and magnificent pancreatic artery (DPA, MPA) between 40-keV MEI (+) and PEI (p > 0.05). Four types of variations were observed in the origin of DPA and three to five types in the origin of PSPDA, AIPDA and PIPDA.

Conclusions

40-keV MEI (+) of DECT improves the visualization and objective and subjective image quality of pancreatic supplying arteries compared to PEI. Pancreatic supplying arteries have great variations, which has important implications for preoperative planning of technically challenging surgeries and transarterial interventions.

Comparison of multiphase data from CT perfusion vs clinical 4-phase CT scans with respect to image quality, lesion detection, and LI-RADS classification in HCC patients

Purpose

The aim of this study was to assess the image quality and diagnostic performance of reconstructed arterial (A) and portal venous (PV) phases in computed tomography perfusion (CTP) scans compared to the corresponding phases in standard 4-phase CT and to assess the utility for LI-RADS classification using CTP and 4-phase 4CT.

Methods

A total of 26 scans with each method (CTP and 4-phase CT) from 19 hepatocellular carcinoma patients were analyzed and compared. Arterial and PV phases reconstructed by advanced modeled iterative reconstruction at strength 4 (ADMIRE 4) from raw CTP data were compared with image sets from arterial and PV phases of 4-phase CT (ADMIRE 3) in the same patient with respect to image quality.

Results

Quantitative image analysis showed that reconstructed CTP datasets were equivalent to 4-phase CT image sets. Qualitative image analysis revealed similar lesion detection rates with the 2 methods for patients with an abdominal diameter ≤36 cm and body weight <90 kg, but lower detection rates with CTP for patients with an abdominal diameter >37 cm. There was no difference in Liver Imaging Reporting and Data System (LI-RADS) classifications between the 2 methods.

Conclusion

Reconstructed CTP images can potentially replace 4-phase CT images in patients weighing <90 kg and with a body diameter <37 cm, as the 2 methods are comparable in terms of quantitative image quality and ability to detect and classify lesions based on LI-RADS criteria.

•

Reconstructed A- and PV-CTP images have comparable image quality to 4-phase CT.

•

Reconstructed A- and PV-CTP images can be used for LI-RADS classification of HCC.

•

A-/PV-CTP has the potential to reliably detect lesions in patients weighing <90 kg with body diameter ≤36 cm.

Impact of iterative reconstruction algorithms on image quality and radiation dose in computed tomography scan of patients with malignant pancreatic lesions

Background:

The objective of this study was to investigate the influence of iterative reconstruction (IR) algorithm on radiation dose and image quality of computed tomography (CT) scans of patients with malignant pancreatic lesions by designing a new protocol.

Methods:

The pancreas CT was performed on 40 patients (23 males and 17 females) with a 160-slice CT scan machine. The pancreatic parenchymal phase was performed in two stages: one with a usual dose of radiation and the other one after using a reduced dose of radiation. The images obtained with usual dose were reconstructed with Filtered Back Projection (FBP) method (Protocol A); and the images obtained with the reduced dose were reconstructed with both FBP (Protocol B) and IR method (Protocol C). The quality of images and radiation dose were compared among the three protocols.

Results:

Image noise was significantly lower with Protocol C (10.80) than with Protocol A (14.98) and Protocol B (20.60) (P < 0.001). Signal-to-noise ratio and contrast-to-noise ratio were significantly higher with Protocol C than with Protocol A and Protocol B (P < 0.001). Protocol A and Protocol C were not significantly different in terms of image quality scores. Effective dose was reduced by approximately 48% in Protocol C compared with Protocol A (1.20 ± 0.53 mSv vs. 2.33 ± 0.86 mSv, P < 0.001).

Conclusion:

Results of this study showed that applying the IR method compared to the FBP method can improve objective image quality, maintain subjective image quality, and reduce the radiation dose of the patients undergo pancreas CT.

IMAGE QUALITY AND POTENTIAL DOSE REDUCTION USING ADVANCED MODELED ITERATIVE RECONSTRUCTION (ADMIRE) IN ABDOMINAL CT - A REVIEW

Traditional filtered back projection (FBP) reconstruction methods have served the computed tomography (CT) community well for over 40 years. With the increased use of CT during the last decades, efforts to minimise patient exposure, while maintaining sufficient or improved image quality, have led to the development of model-based iterative reconstruction (MBIR) algorithms from several vendors. The usefulness of the advanced modeled iterative reconstruction (ADMIRE) (Siemens Healthineers) MBIR in abdominal CT is reviewed and its noise suppression and/or dose reduction possibilities explored. Quantitative and qualitative methods with phantom and human subjects were used. Assessment of the quality of phantom images will not always correlate positively with those of patient images, particularly at the higher strength of the ADMIRE algorithm. With few exceptions, ADMIRE Strength 3 typically allows for substantial noise reduction compared to FBP and hence to significant (≈30%) patient dose reductions. The size of the dose reductions depends on the diagnostic task.