An Artificial Intelligence Tool for Clinical Decision Support and Protocol Selection for Brain MRI

BACKGROUND AND PURPOSE

Protocolling, the process of determining the most appropriate acquisition parameters for an imaging study, is time-consuming and produces variable results depending on the performing physician. The purpose of this study was to assess the potential of an artificial intelligence-based semiautomated tool in reducing the workload and decreasing unwarranted variation in the protocolling process.

MATERIALS AND METHODS

We collected 19,721 MR imaging brain examinations at a large academic medical center. Criterion standard labels were created using physician consensus. A model based on the Long Short-Term Memory network was trained to predict the most appropriate protocol for any imaging request. The model was modified into a clinical decision support tool in which high-confidence predictions, determined by the values the model assigns to each possible choice, produced the best protocol automatically and low confidence predictions provided a shortened list of protocol choices for review.

RESULTS

The model achieved 90.5% accuracy in predicting the criterion standard labels and demonstrated higher agreement than the original protocol assignments, which achieved 85.9% accuracy (κ = 0.84 versus 0.72, P value < .001). As a clinical decision support tool, the model automatically assigned 70% of protocols with 97.3% accuracy and, for the remaining 30% of examinations, achieved 94.7% accuracy when providing the top 2 protocols.

CONCLUSIONS

Our model achieved high accuracy on a standard based on physician consensus. It showed promise as a clinical decision support tool to reduce the workload by automating the protocolling of a sizeable portion of examinations while maintaining high accuracy for the remaining examinations.

Millimeter-wave interferometry and far-forward scattering for density fluctuation measurements on LTX-β

The λ ≈ 1 mm (f = 288 GHz) interferometer for the Lithium Tokamak Experiment-β (LTX-β) will use a chirped-frequency source and a centerstack-mounted retro-reflector mirror to provide electron line density measurements along a single radial chord at the midplane. The interferometer is unique in the use of a single source (narrow-band chirped-frequency interferometry) and a single beam splitter for separating and recombining the probe and reference beams. The current work provides a documentation of the interferometry hardware and evaluates the capabilities of the system as a far-forward collective scattering diagnostic. As such, the current optical setup is estimated to have a detection range of 0.4 ≲ k _⊥ ≲ 1.7 cm^-1, while an improved layout will extend the upper k _⊥ limit to ∼3 cm^-1. Measurements with the diagnostic on LTX are presented, showing interferometry results and scattered signal data. These diagnostics are expected to provide routine measurements on LTX-β for high frequency coherent density oscillations (e.g., Alfvénic modes during neutral beam injection) as well as for broadband turbulence.

A frequency-modulated continuous-wave reflectometer for the Lithium Tokamak Experiment

The frequency-modulated continuous-wave reflectometer on LTX (Lithium Tokamak Experiment) and the data analysis methods used for determining electron density profiles are described. The diagnostic uses a frequency range of 13.1-33.5 GHz, for covering a density range of 0.21-1.4×10¹³ cm^-3 (in O-mode polarization) with a time resolution down to 8 μs. The design of the diagnostic incorporates the concept of an "optimized" source frequency sweep, which minimizes the large variation in the intermediate frequency signal due to a long dispersive transmission line. The quality of the raw data is dictated by the tuning characteristics of the microwave sources, as well as the group delay ripple in the transmission lines, which can generate higher-order nonlinearities in the frequency sweep. Both effects are evaluated for our diagnostic and best practices are presented for minimizing "artifacts" generated in the signals. The quality of the reconstructed profiles is also improved using two additional data analysis methods. First, the reflectometer data are processed as a radar image, where clutter due to echoes from the wall and backscattering from density fluctuations can be easily identified and removed. Second, a weighed least-squares lamination algorithm POLAN (POLynomial ANalysis) is used to reconstruct the electron density profile. Examples of density profiles in LTX are presented, along with comparisons to measurements from the Thomson scattering and the λ = 1 mm interferometer diagnostics.

An exponential growth in incidence of thyroid cancer: trends and impact of CT imaging

BACKGROUND AND PURPOSE

Workup of incidental thyroid nodules detected on CT imaging could be contributing to the increased diagnosis of small thyroid cancers. The purpose of this study was to evaluate recent trends in the incidence of thyroid cancer, and to determine the relationship between annual CT imaging volume and rate of thyroid cancer diagnosis.

MATERIALS AND METHODS

This retrospective cohort study used data bases for thyroid cancer and CT imaging volume. Thyroid cancer data from 1983-2009 were obtained from the Surveillance, Epidemiology, and End Results data base. National Council of Radiation Protection and Measurements Report No. 160 provided data on hospital and nonhospital CT imaging volume for 1993-2006. Trends in thyroid cancer were modeled for overall incidence on the basis of patient age, tumor histologic features, and tumor size and stage. Linear regression analysis was performed to evaluate the strength of the relationship between annual CT scan volume and the incidence of thyroid cancer by tumor size and histologic type.

RESULTS

In 2009, the incidence of thyroid cancer was 14 per 100,000, which represented a 1.9-fold increase compared with 2000. The growth in incidence was exponential compared with a minimal linear increase in thyroid cancer mortality rate. The subgroup with the greatest change was subcentimeter papillary carcinoma, with doubling in incidence approximately every 6.2 years. The linear relationship between annual CT scan volume and the incidence of subcentimeter papillary carcinoma was very strong (R(2) = 0.98; P < .0001).

CONCLUSIONS

The incidence of subcentimeter papillary carcinoma is growing at an exponential rate without significant change in mortality rate. The strong linear relationship between new cases of subcentimeter papillary carcinomas and the number of CT scans per year suggests that an increase in CT scans may increase the detection of incidental thyroid cancers.

Incidental thyroid nodules on CT: evaluation of 2 risk-categorization methods for work-up of nodules

BACKGROUND AND PURPOSE

Thyroid nodules are common incidental findings on CT, but there are no clear guidelines regarding their further diagnostic work-up. This study compares the performance of 2 risk-categorization methods of selecting CT-detected incidental thyroid nodules for work-up.

MATERIALS AND METHODS

The 2 categorization methods were method A, based on nodule size ≥10 mm, and method B, a 3-tiered system based on aggressive imaging features, patient age younger than 35 years or nodule size of ≥15 mm. In part 1, the 2 categorization methods were applied to thyroid cancers in the SEER data base of the National Cancer Institute to compare the cancer capture rates and survival. In part two, 755 CT neck scans at our institution were retrospectively reviewed for the presence of ITNs of ≥5 mm, and the same 2 categorization methods were applied to the CT cases to compare the number of patients who would theoretically meet the criteria for work-up. Comparisons of proportions of subjects captured under methods A and B were made by using the McNemar test.

RESULTS

For 84,720 subjects in the SEER data base, methods A and B each captured 74% (62,708/84,720 and 62,586/84,720, respectively) of malignancies. SEER subjects who would not have met the criteria for further work-up by both methods had equally excellent 10-year cause-specific and relative survival of >99%. For part 2, the prevalence of ITNs of ≥5 mm at our institution was 133/755 (18%). The number of ITNs that would be recommended for work-up by method A was 57/133 (43%) compared with 31/133 (23%) for method B (P < .0005).

CONCLUSIONS

Compared with using a 10-mm cutoff, the 3-tiered risk-stratification method identified fewer ITNs for work-up but captured the same proportion of cancers in a national data base and showed no difference in missing high-mortality cancers.

A Ka-band tunable direct-conversion correlation reflectometer for NSTX

The recent availability of broadband microwave quadrature mixers in the Ka-band (28-40 GHz) of frequencies has allowed the fabrication of low-cost direct-conversion detection circuits for use in the variable-frequency correlation reflectometer on the National Spherical Torus eXperiment (NSTX). The quadrature receiver in this case can be implemented as a simple homodyne circuit, without the complication of a single-sideband modulator or a feedforward tracking circuit present in more typical designs. A pair of direct-conversion receivers is coupled with broadband microwave voltage-controlled oscillators to construct a flexible dual-channel radar system with a fast frequency settling time of ∼160 μs. A detailed description of the design and a full characterization of the hardware are provided. Examples of turbulence measurements from radial and poloidal correlation reflectometry on NSTX using a poloidal array of antennas (oriented normal to the magnetic flux surfaces for conventional reflectometry) are presented.

Intrapallidal lipopolysaccharide injection increases iron and ferritin levels in glia of the rat substantia nigra and induces locomotor deficits

Increasing evidence suggests that abnormal iron handling may be involved in the pathogenesis of Parkinson's disease. The present study investigates the role of iron and the iron-storage protein ferritin in inflammation-induced degeneration of dopaminergic neurons of the substantia nigra pars compacta. Injection of lipopolysaccharide into the globus pallidus of young and middle-aged rats substantially decreased tyrosine hydroxylase immunostaining in substantia nigra pars compacta four weeks after injection. Loss of tyrosine hydroxylase expression was accompanied by increased iron and ferritin levels in glial cells of the substantia nigra pars reticulata. Despite greater increases in nigral iron levels, ferritin induction was less pronounced in older rats, suggesting the regulation of ferritin was compromised with age. Automated movement tracking analyses showed that young rats recovered from LPS-induced locomotor deficits within four weeks, yet older rats failed to improve on measures of speed and total distance moved. Intrapallidal lipopolysaccharide injection also increased expression of alpha-synuclein and ubiquitin in tyrosine hydroxylase-positive neurons of the substantia nigra pars compacta. These results suggest that pallidal inflammation significantly increases stress on dopamine-containing neurons in the substantia nigra pars compacta. Alterations in nigral iron levels and protein handing may increase the vulnerability of nigral neurons to degenerative processes.

A novel fluorescent method for direct visualization of neurofibrillary pathology in Alzheimer's disease

Methods currently available for detecting neurofibrillary pathology are indirect and depend on staining with exogenous chemicals or antibodies. In the present study, we report a novel method named intrinsic fluorescence induction (IFI), which allows direct visualization of neurofibrillary tangles (NFTs), neuropil threads (NTs), and neuritic plaques (NPs) in tissue sections of Alzheimer's disease (AD) brain. The IFI method is based on both induction of a red intrinsic fluorescence and quenching red background autofluorescence. The IFI procedure includes sustained hydrophobic treatment, protein secondary structure enhancement and incubation in high concentration of phosphate buffer. Following this procedure, a unique red fluorescence is generated from the structures of NFTs, NTs, and NPs in brain sections from AD patients. Sequential application of mild permanganate oxidation and 1% sodium borohydride selectively removes the red background autofluorescence, while the latter enhances the intrinsic fluorescence of neurofibrillary pathology. Comparative studies reveal that the IFI method is as sensitive as Gallyas silver staining, and more sensitive than Bielschowsky silver staining or PHF-1 immunostaining in detecting NFTs in the pre-alpha layer of entorhinal cortex and the pri-alpha layer of the entorhinal/transentorhinal cortex. Furthermore, the IFI method is sensitive in displaying plaque neurites and threads, but not NFTs in the hippocampus. This novel finding provides a direct method for detecting neurofibrillary pathology in particular regions of AD brain and a novel tool for AD research.

Central command neurons of the sympathetic nervous system: basis of the fight-or-flight response

During stress, the activity of the sympathetic nervous system is changed in a global fashion, leading to an increase in cardiovascular function and a release of adrenal catecholamines. This response is thought to be regulated by a common set of brain neurons that provide a dual input to the sympathetic preganglionic neurons regulating cardiac and adrenal medullary functions. By using a double-virus transneuronal labeling technique, the existence of such a set of central autonomic neurons in the hypothalamus and brainstem was demonstrated. These neurons innervate both of the sympathetic outflow systems and likely function in circumstances where parallel sympathetic processing occurs, such as in the fight-or-flight response.