The migration of imaging reports to electronic medical record systems holds great potential in terms of advancing radiology research and practice by leveraging the large volume of data continuously being updated, integrated, and shared. However, there are significant challenges as well, largely due to the heterogeneity of how these data are formatted. Indeed, although there is movement toward structured reporting in radiology (ie, hierarchically itemized reporting with use of standardized terminology), the majority of radiology reports remain unstructured and use free-form language. To effectively "mine" these large datasets for hypothesis testing, a robust strategy for extracting the necessary information is needed. Manual extraction of information is a time-consuming and often unmanageable task. "Intelligent" search engines that instead rely on natural language processing (NLP), a computer-based approach to analyzing free-form text or speech, can be used to automate this data mining task. The overall goal of NLP is to translate natural human language into a structured format (ie, a fixed collection of elements), each with a standardized set of choices for its value, that is easily manipulated by computer programs to (among other things) order into subcategories or query for the presence or absence of a finding. The authors review the fundamentals of NLP and describe various techniques that constitute NLP in radiology, along with some key applications.

OBJECTIVE:Results of the use of the transluminal attenuation gradient (TAG) at coronary CT angiography (CCTA) to predict hemodynamically significant disease vary widely. This study tested whether diagnostic performance of TAG to predict fractional flow reserve (FFR) ≤ 0.8 is improved when applied separately to subsets of coronary arteries that carry similar physiological flow. METHODS:28 patients with 64 × 0.5 mm CCTA and invasive FFR in ≥1 major coronary artery were retrospectively evaluated. Two readers assessed TAG in each artery. The receiver operating characteristic (ROC) area under the curve (AUC) was used to assess the diagnostic performance of TAG to detect hemodynamically significant disease following a clinical use rule [negative: FFR > 0.8 or ≤ 25% diameter stenosis (DS) at invasive catheter angiography; positive: FFR ≤ 0.8 or ≥ 90% DS at invasive catheter angiography]. ROC AUC was compared for all arteries pooled together, vs separately for arteries carrying similar physiological flow (Group 1: all left anterior descending plus right-dominant left circumflex; Group 2: right-dominant RCA plus left/co-dominant left circumflex). RESULTS:Of the 84 arteries, 30 had FFR measurements, 30 had ≤25% DS and 13 had ≥90% DS. 11 arteries with 26-89% DS and no FFR measurement were excluded. TAG interobserver reproducibility was excellent (Pearson r = 0.954, Bland-Altman bias: 0.224 Hounsfield unit cm-1). ROC AUC to detect hemodynamically significant disease was higher when considering arteries separately (Group 1 AUC = 0.841, p = 0.039; Group 2 AUC = 0.840, p = 0.188), than when pooling all arteries together (AUC = 0.661). CONCLUSION:Incorporating information on the physiology of coronary flow via the particular vessel interrogated and coronary dominance may improve the accuracy of TAG, a simple measurement that can be quickly performed at the time of CCTA interpretation to detect hemodynamically significant stenosis in individual coronary arteries. Advances in knowledge: The interpretation of TAG may benefit by incorporating information regarding which coronary artery is being interrogated.

Objective

To test the hypothesis that wide area detector face transplant surgical planning CT angiograms with simulated lower radiation dose and iterative reconstruction (AIDR3D) are comparable in image quality to those with standard tube current and filtered back projection (FBP) reconstruction.

Materials and methods

The sinograms from 320-detector row CT angiography of four clinical candidates for face transplantation were processed utilizing standard FBP, FBP with simulated 75, 62, and 50% tube current, and AIDR3D with corresponding dose reduction. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were measured at muscle, fat, artery, and vein. Image quality for each reconstruction strategy was assessed by two independent readers using a 4-point scale.

Results

Compared to FBP, the median SNR and CNR for AIDR3D images were higher at all sites for all 4 different tube currents. The AIDR3D with simulated 50% tube current achieved comparable SNR and CNR to FBP with standard dose (median muscle SNR: 5.77 vs. 6.23; fat SNR: 6.40 vs. 5.75; artery SNR: 43.8 vs. 45.0; vein SNR: 54.9 vs. 55.7; artery CNR: 38.1 vs. 38.6; vein CNR: 49.0 vs. 48.7; all p-values >0.19). The interobserver agreement in the image quality score was good (weighted κ = 0.7). The overall score and the scores for smaller arteries were significantly lower when FBP with 50% dose reduction was used. The AIDR3D reconstruction images with 4 different simulated doses achieved a mean score ranging from 3.68 to 3.82 that were comparable to the scores from images reconstructed using FBP with original dose (3.68-3.77).

Conclusions

Simulated radiation dose reduction applied to clinical CT angiography for face transplant planning suggests that AIDR3D allows for a 50% reduction in radiation dose, as compared to FBP, while preserving image quality.

Background

The effects of reduced radiation dose CT for the generation of maxillofacial bone STL models for 3D printing is currently unknown. Images of two full-face transplantation patients scanned with non-contrast 320-detector row CT were reconstructed at fractions of the acquisition radiation dose using noise simulation software and both filtered back-projection (FBP) and Adaptive Iterative Dose Reduction 3D (AIDR3D). The maxillofacial bone STL model segmented with thresholding from AIDR3D images at 100 % dose was considered the reference. For all other dose/reconstruction method combinations, a "residual STL volume" was calculated as the topologic subtraction of the STL model derived from that dataset from the reference and correlated to radiation dose.

Results

The residual volume decreased with increasing radiation dose and was lower for AIDR3D compared to FBP reconstructions at all doses. As a fraction of the reference STL volume, the residual volume decreased from 2.9 % (20 % dose) to 1.4 % (50 % dose) in patient 1, and from 4.1 % to 1.9 %, respectively in patient 2 for AIDR3D reconstructions. For FBP reconstructions it decreased from 3.3 % (20 % dose) to 1.0 % (100 % dose) in patient 1, and from 5.5 % to 1.6 %, respectively in patient 2. Its morphology resembled a thin shell on the osseous surface with average thickness <0.1 mm.

Conclusion

The residual volume, a topological difference metric of STL models of tissue depicted in DICOM images supports that reduction of CT dose by up to 80 % of the clinical acquisition in conjunction with iterative reconstruction yields maxillofacial bone models accurate for 3D printing.

While use of advanced visualization in radiology is instrumental in diagnosis and communication with referring clinicians, there is an unmet need to render Digital Imaging and Communications in Medicine (DICOM) images as three-dimensional (3D) printed models capable of providing both tactile feedback and tangible depth information about anatomic and pathologic states. Three-dimensional printed models, already entrenched in the nonmedical sciences, are rapidly being embraced in medicine as well as in the lay community. Incorporating 3D printing from images generated and interpreted by radiologists presents particular challenges, including training, materials and equipment, and guidelines. The overall costs of a 3D printing laboratory must be balanced by the clinical benefits. It is expected that the number of 3D-printed models generated from DICOM images for planning interventions and fabricating implants will grow exponentially. Radiologists should at a minimum be familiar with 3D printing as it relates to their field, including types of 3D printing technologies and materials used to create 3D-printed anatomic models, published applications of models to date, and clinical benefits in radiology. Online supplemental material is available for this article.

Objective

Upper extremity (UE) transplantation is the most commonly performed composite tissue allotransplantation worldwide. However, there is a lack of imaging standards for pre- and posttransplant evaluation. This study highlights the protocols and findings of UE allotransplantation toward standardization and implementation for clinical trials.

Methods

Multimodality imaging protocols for a unilateral hand transplant candidate and a bilateral mid-forearm level UE transplant recipient include radiography, computed tomography (CT), magnetic resonance (MR) imaging, catheter angiography, and vascular ultrasonography. Pre- and posttransplant findings, including dynamic CT and MR performed for assessment of motor activity of transplanted hands, are assessed, and image quality of vessels and bones on CT and MR evaluated.

Results

Preoperative imaging demonstrates extensive skeletal deformity and variation in vascular anatomy and vessel patency. Posttransplant images confirm bony union in anatomical alignment and patency of vascular anastomoses. Mild differences in rate of vascular enhancement and extent of vascular networks are noted between the 2 transplanted limbs. Dynamic CT and MR demonstrate a 15° to 30° range of motion at metacarpophalangeal joints and 90° to 110° at proximal interphalangeal joints of both transplanted hands at 8 months posttransplant. Image quality was slightly better for CT than for MR in the first subject, while MR was slightly better in the second subject.

Conclusion

Advanced vascular and musculoskeletal imaging play an important role in surgical planning and can provide novel posttransplantation data to monitor the success of the procedure. Implementation of more standardized protocols should enable a more comprehensive assessment to evaluate the efficacy in clinical trials.

Background

If undetected, infrarenal Abdominal Aortic Aneurysm (AAA) growth can lead to rupture, a high-mortality complication. Some AAA patients exhibit inhomogeneous luminal contrast attenuation at first-pass CT angiography (CTA). This study assesses the association between this observation and aneurysm growth.

Methods

Sixty-seven consecutive pre-repair AAA CTAs were included in this retrospective study. The "Gravitational Gradient" (GG), defined as the ratio of the mean attenuation in a region-of-interest placed posteriorly to that in a region-of-interest placed anteriorly within the lumen of the aortic aneurysm on a single axial slice, and the maximum aneurysm diameter were measured from each CT data set. "AAA Contrast Inhomogeneity" was defined as the absolute value of the difference between the GG and 1.0. Univariate and multivariate logistic regression was used to assess the association of aneurysm growth >0.4 and >1.0 cm/year to AAA Contrast Inhomogeneity, aneurysm diameter, patient characteristics and cardiovascular co-morbidities.

Results

AAA Contrast Inhomogeneity was not correlated to aneurysm diameter (p = 0.325). In multivariable analysis that included initial aneurysm diameter and AAA Contrast Inhomogeneity, both factors were significantly associated with rapid aneurysm growth (initial diameter: p = 0.029 and 0.011, and, AAA Contrast Inhomogeneity: p = 0.045 and 0.048 for growth >0.4 cm/year and >1 cm/year respectively).

Conclusions

AAA Contrast Inhomogeneity is a common observation in first-pass CTA. It is associated with rapid aneurysm growth, independent of aneurysm diameter.

Coronary computed tomography angiography (CCTA) contrast opacification gradients, or transluminal attenuation gradients (TAG) offer incremental value to predict functionally significant lesions. This study introduces and evaluates an automated gradients software package that can potentially supplant current, labor-intensive manual TAG calculation methods. All 60 major coronary arteries in 20 patients who underwent a clinically indicated single heart beat 320 × 0.5 mm detector row CCTA were retrospectively evaluated by two readers using a previously validated manual measurement approach and two additional readers who used the new automated gradient software. Accuracy of the automated method against the manual measurements, considered the reference standard, was assessed via linear regression and Bland-Altman analyses. Inter- and intra-observer reproducibility and factors that can affect accuracy or reproducibility of both manual and automated TAG measurements, including CAD severity and iterative reconstruction, were also assessed. Analysis time was reduced by 68% when compared to manual TAG measurement. There was excellent correlation between automated TAG and the reference standard manual TAG. Bland-Altman analyses indicated low mean differences (1 HU/cm) and narrower inter- and intra-observer limits of agreement for automated compared to manual measurements (25 and 36% reduction with automated software, respectively). Among patient and technical factors assessed, none affected agreement of manual and automated TAG measurement. Automated 320 × 0.5 mm detector row gradient software reduces computation time by 68% with high accuracy and reproducibility.