Channelized Hotelling observer (CHO), which has been shown to be well correlated with human observer performance in many clinical CT tasks, has a great potential to become the method of choice for objective image quality assessment. However, its use has been quite limited in routine CT practice due to lack of efficient implementation. In this work, a CHO model optimized for the most widely used ACR CT accreditation phantom was applied to evaluate the low-contrast detectability of a deep-learning based reconstruction (DLIR) equipped on a GE Revolution scanner. The commercially available DLIR reconstruction method showed consistent increase in low-contrast detectability over the FBP and the IR method at routine dose levels, which suggests potential dose reduction to the FBP reconstruction by up to 27.5%.
In large part from concerns about radiation exposure from computed tomography (CT), iterative reconstruction (IR) has emerged as a popular technique for dose reduction. Although IR clearly reduces image noise and improves resolution, its ability to maintain or improve low-contrast detectability over (possibly post-processed) filtered backprojection (FBP) reconstructions is unclear. In this work, we have scanned a low contrast phantom encased in an acrylic oval using two vendors’ scanners at 120 kVp at three dose levels for axial and helical acquisitions with and without automatic exposure control. Using the local noise power spectra of the FBP and IR images to guide the filter design, we developed a two-dimensional angularly-dependent Gaussian filter in the frequency domain that can be optimized to minimize the root-mean-square error between the image-domain filtered FBP and IR reconstructions. The filter is extended to three-dimensions by applying a through-slice Gaussian filter in the image domain. Using this three-dimensional, non-isotropic filtering approach on data with non-uniform statistics from both scanners, we were able to process the FBP reconstructions to closely match the low-contrast performance of IR images reconstructed from the same raw data. From this, we conclude that most or all of the benefits of noise reduction and low-contrast performance of advanced reconstruction can be achieved with adaptive linear filtering of FBP reconstructions in the image domain.
In this study, the feasibility of differentiating uric acid from
non-uric acid kidney stones in
the presence of iodinated contrast material was evaluated using
dual-energy CT (DECT). Iodine
subtraction was accomplished with a commercial three material decomposition algorithm to create a
virtual non-contrast (VNC) image set. VNC images were then used to segment stone regions from
tissue background. The DE ratio of each stone was calculated using the CT images acquired at two
different energies with DECT using the stone map generated from the VNC images. The
performance of DE ratio-based stone differentiation was evaluated at five different iodine
concentrations (21, 42, 63, 84 and 105 mg/ml). The DE ratio of stones in iodine solution was found
larger than those obtained in non-iodine cases. This is mainly caused by the partial volume effect
around the boundary between the stone and iodine solution. The overestimation of the DE ratio leads
to substantial overlap between different stone types. To address the partial volume effect, an
expectation-maximization (EM) approach was implemented to estimate the contribution of iodine
and stone within each image pixel in their mixture area. The DE ratio of each stone was corrected to
maximally remove the influence of iodine solutions. The separation of uric-acid and non-uric-acid
stone was improved in the presence of iodine solution.
Multi-spectral Near Infrared tomographic imaging has the potential to provide information about patho-physiological
function of soft tissue. However, the specific choice of wavelengths used is crucial for the accurate separation of such
parameters. It will be demonstrated that the conventionally believed choice of large set of wavelengths can be
detrimental in accurate recovery of tissue specific functions. The method of determining a set of optimized bands of
wavelengths will be presented and are tested using simulations and experimental data. It will be shown that the
optimization method achieves images as accurate as using the full spectrum, but improves crosstalk between
parameters. Additionally, a Jacobian normalization technique is presented which takes into account the varying
magnitude of different optical parameters within image reconstruction, creating a more uniform update within a
spectral image reconstruction model.
We present a wavelength optimization approach incorporating the spectral derivative method in near infrared (NIR)
reconstruction. The similarity between spectral features of
oxy-hemoglobin and water leads to difficulty in separating
them with any spectral reconstruction method. By taking the difference of wavelength pairs and using a spectral
derivative fitting method, this similarity can be broken down to better differentiate each set of two chromophores.
Optimal wavelength permutations were chosen based on the criteria of minimum correlation of the differences in
adjacent wavelengths. The direct and derivative methods were then compared with optimal wavelength sets in two and
three dimensional reconstructions. The wavelength optimized derivative method shows superior results in the recovery
of chromophore concentrations. This approach can not only yield reduced coupling and geometry errors, it also shows
better separation of one chromophore from others.
Multispectral near-infrared (NIR) tomographic imaging has the potential to provide information about molecules absorbing light in tissue, as well as subcellular structures scattering light, based on transmission measurements. However, the choice of possible wavelengths used is crucial for the accurate separation of these parameters, as well as for diminishing crosstalk between the contributing chromophores. While multispectral systems are often restricted by the wavelengths of laser diodes available, continuous-wave broadband systems exist that have the advantage of providing broadband NIR spectroscopy data, albeit without the benefit of the temporal data. In this work, the use of large spectral NIR datasets is analyzed, and an objective function to find optimal spectral ranges (windows) is examined. The optimally identified wavelength bands derived from this method are tested using both simulations and experimental data. It is found that the proposed method achieves images as qualitatively accurate as using the full spectrum, but improves crosstalk between parameters. Additionally, the judicious use of these spectral windows reduces the amount of data needed for full spectral tomographic imaging by 50%, therefore increasing computation time dramatically.
Near-infrared (NIR) region-based spectroscopy is examined for accuracy with spectral recovery using frequency domain data at a discrete number of wavelengths, as compared to that with broadband continuous wave data. Data with more wavelengths in the frequency domain always produce superior quantitative spectroscopy results with reduced noise and error in the chromophore concentrations. Performance of the algorithm in the situation of doing region-guided spectroscopy within the MRI is also considered, and the issue of false positive prior regions being identified is examined to see the effect of added wavelengths. The results indicate that broadband frequency domain data are required for maximal accuracy. A broadband frequency domain experimental system was used to validate the predictions, using a mode-locked Ti:sapphire laser for the source between 690- and 850-nm wavelengths. The 80-MHz pulsed signal is heterodyned with photomultiplier tube detection, to lower frequency for data acquisition. Tissue-phantom experiments with known hemoglobin absorption and tissue-like scatter values are used to validate the system, using measurements every 10 nm. More wavelengths clearly provide superior quantification of total hemoglobin values. The system and algorithms developed here should provide an optimal way to quantify regions with the goal of image-guided breast tissue spectroscopy within the MRI.
NIR tomography image reconstruction can be improved by incorporating spectral constraints and prior spatial
information. The convergence of scattering power was studied based on the distribution of projection error with different
parameters. The reason that scattering properties are harder to recover than chromophore concentrations was discussed.
Using "hard prior" spectral reconstruction, the role of stopping criteria was found to be important. Multiple wavelength
simulations were used to choose suitable stopping criteria. Preliminary tests using a wavelength tunable Ti-Sapphire
laser shows promise for frequency domain measurements covering a wide range of wavelengths.
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.