Purpose: Motion artifacts in magnetic resonance (MR) images mostly undergo subjective evaluation, which is poorly reproducible, time consuming, and costly. Recently, full-reference image quality assessment (FR–IQA) metrics, such as structural similarity (SSIM), have been used, but they require a reference image and hence cannot be used to evaluate clinical images. We developed a convolutional neural network (CNN) model to quantify motion artifacts without using reference images.
Approach: The brain MR images were obtained from an open dataset. The motion-corrupted images were generated retrospectively, and the peak signal-to-noise ratio, cross-correlation coefficient, and SSIM were calculated. The CNN was trained using these images and their FR–IQA metrics to predict the FR–IQA metrics without reference images. Receiver operating characteristic (ROC) curves were created for binary classification, with artifact scores < 4 indicating the need for rescanning. ROC curve analysis was performed on the binary classification of the real motion images.
Results: The predicted FR–IQA metric having the highest correlation with the subjective evaluation was SSIM, which was able to classify images requiring rescanning with a sensitivity of 89.5%, specificity of 78.2%, and area under the ROC curve (AUC) of 0.930. The real motion artifacts were classified with the AUC of 0.928.
Conclusions: Our CNN model predicts FR–IQA metrics with high accuracy, which enables quantitative assessment of motion artifacts in MR images without reference images. It enables classification of images requiring rescanning with a high AUC, which can improve the workflow of MR imaging examinations.
We developed a novel system for imaging and qualitatively analyzing the surface vessels using near-infrared (NIR) radiation using tuned aperture computed tomography (TACT®). The system consisted of a NIR-sensitive CCD camera surrounded by sixty light emitting diodes (with wavelengths alternating between 700 or 810 nm). This system produced thin NIR tomograms, under 0.5 mm in slice thickness. The venous oxygenation index reflecting oxygen saturation levels calculated from NIR tomograms was more sensitive than that from the NIR images. This novel system makes it possible to noninvasively obtain NIR tomograms and accurately analyze changes in oxygen
We developed a novel system for imaging and qualitative analysis of surface vessels using near infrared (NIR) radiation with tuned aperture computed tomography (TACT®), even if the NIR cannot transmit through thick regions. NIR-sensitive CCD camera was surrounded by sixty light emitting diodes (alternating wavelengths of 700 nm and 810 nm), and could only detect the NIR from the subcutaneous tissue. We obtained multiple near infrared projections of surface vessels at each wavelength in accordance with the optical aperture theory within one second. Then, we created tomograms using the TACT program, and determined the venous oxygenation index (VOI), which reflected the oxygen saturation level, calculated from the image signals at each wavelength. This system produced thinner NIR tomograms under 0.5 mm. The change in VOI after load test calculated from NIR tomograms was more sensitive than that from NIR images without tomography. Our novel system makes it possible to non-invasively obtain NIR tomograms and accurately analyze changes in oxygen saturation.
The goal is non-linear weighted type half-scan algorithm for reconstruction of a long object, which has important applications for clinical CT. Image reconstruction from cone-beam projections collected along a half-scan trajectory is commonly done using the Feldkamp-type half-scan algorithm, which performs well only with a large cone angle. Half-scan CT algorithms are advantageous in terms of temporal resolution, and widely used in fan-beam and cone-beam geometry. We propose a non-linear weight based algorithm to increase the cone angle by several folds to achieve satisfactory image quality at the same radiation dose. In our scheme, we first weighting with respect to half-scan projection data at individual projection angles is changed. Then, distribution of correction coefficients so that they are large near the center of the detector, while taking individual channel data for the detector into account, and smaller near the edges. Finally, three-dimensional back-projection of corrected half-scan projection data. Numerical phantoms are used to assess image quality indexes. Comparison with Feldkamp-type half-scan reconstruction is conducted. Numerical simulation studies are performed to verify the correctness and demonstrate performance. Image quality of the long object reconstruction is similar to that of the short object reconstruction. Our non-linear weighted half-scan reconstruction algorithms allow minimization of redundant data and optimization of temporal resolution, and outperform Feldkamp-type half-scan algorithm. These algorithms seem promising for quantitative and dynamic biomedical applications of cone-beam tomography. We extended our non-linear weighted half-scan method into a solution to the long object problem. Our non-linear weighted half-scan algorithm has a potential for CT in cone-beam geometry.
We devised two kinds of new methods for accurate measurement of the image signal-to-noise ratio (SNR) in parallel magnetic resonance imaging (MRI) because image noise of the parallel MRI was not spatially constant. Using the first (Consecutive) method, more than fifty consecutive scans of the uniform phantom were obtained with identical scan parameters. Then the SNRs in each pixel were calculated from the ratio of mean signal intensity to the standard deviation of the time domain on a pixel-by-pixel. With the second (Remove) method, the phantom was removed after the first scan, and the second scan was done with identical parameters and the RF coil loading device. The SNRs in each pixel were then obtained from the ratio of the signal intensity of the first scan to the second scan (w/o phantom) image which was multiplied by the square root of 2/pi and filtered by the running mean (7 by 7 pixels). Moreover, actual geometry factors were calculated from image SNRs of parallel and no parallel MRI. The image SNR and actual geometry factor of parallel MRI with the Consecutive method agreed with that of the Remove method. The SNRs of the no parallel MRI with the above two methods conformed with that of the conventional SNR method (NEMA standard). Both new methods make it possible to obtain a more detailed determination of SNR in parallel MRI, and to calculate the actual geometry factor.
We investigated an image reconstruction algorithm to reduce cone-beam artifacts in cone-beam CT. To examine the factors involved in the occurrence of cone-beam artifacts, micro-spheres phantom were arranged longitudinally at different positions and a computer simulation was performed. Due to differences in projection angle, data projected onto the detector surface were projected along trajectories shown as different periodic functions depending on the distance and position from the center of rotation. Therefore, projection along several detector channels based on different projection data resulting from different periodic functions is considered responsible for the increase in cone-beam artifacts associated with an increase in the distance of reconstruction planes from the center of rotation. Our new algorithm to reduce such artifacts features: 1) A change in weighting with respect to projection data obtained at different projection angles. 2) Distribution of correction coefficients so that they are larger near the center of the detector, while taking individual channel data for the detector into account, and smaller near the edges. 3) Three-dimensional back-projection of corrected projection data. The effect of the reduction in cone-beam artifacts of an object located at the edges markedly enhanced reconstruction planes at positions further from the center of rotation.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.