Sociodemographic Variables Reporting in Human Radiology Artificial Intelligence (AI) Research

PURPOSE

Artificial intelligence (AI) is rapidly reshaping how radiology is practiced. Its susceptibility to biases, however, is a primary concern as more AI algorithms become available for widespread use. So far there has been limited evaluation of how sociodemographic variables are reported in radiology AI research. This study aims to evaluate the presence and extent of sociodemographic reporting in human subjects radiology AI original research.

METHODS

All human subjects original radiology artificial intelligence articles published from January to December 2020 in the top 6 US radiology journals, as determined by impact factor, were reviewed. Reporting of any sociodemographic variables (age, gender, and race/ethnicity), as well as any sociodemographic-based results were extracted.

RESULTS

Of the 160 included articles, 54% reported at least one sociodemographic variable, 53% reported age, 47% gender, and 4% race/ethnicity. 6% reported any sociodemographic-based results. There was significant variation in reporting of at least one sociodemographic variable by journal, ranging from 33-100%.

CONCLUSIONS

Reporting of sociodemographic variables in human subjects original radiology artificial intelligence research remains poor, putting the results and subsequent algorithms at increased risk of biases.

First 12-month follow-up of bioresorbable synthetic heart valves implanted in the aortic position in sheep

Background

Replacement of diseased heart valves with the currently available valve prostheses has serious drawbacks. The use of bioresorbable synthetic “in situ tissue engineered” heart valve prostheses has been proposed to overcome the limitations of traditional heart valve prostheses. Such bioresorbable synthetic heart valve prostheses have been successfully tested as pulmonary valve replacements in preclinical studies, but data on aortic valve replacement is lacking. Here, we present the first in-vivo study on the long-term functionality of bioresorbable synthetic heart valves in the high-pressure circulation.

Methods

Approval for the animal studies was obtained by the Amsterdam University Medical Centres Animal Care Ethics Committee (AVD1180020197705) and are in agreement with the current Dutch law on animal experiments (WOD). We surgically implanted bioresorbable synthetic aortic valve prostheses in 20 female Swifter sheep in orthotopic position. The scheduled follow-up times were 1, 3, 6, 9 and 12 months.

Results

Fifteen sheep (75%) recovered well from the surgical valve implantation procedure and were included in this analysis. No sheep died due to valve failure. All valves remained free from active infectious endocarditis, thrombotic complications, and pathological calcification. A total of 10 valves (67%) were intact, thin and pliable and remained free from leaflet thickening, retraction and degradation up to 12 months after implantation. In most valves (67%), the scaffold remained sparsely- or unpopulated by cells during 12 months follow up. In some valves (33%), colonization of the valve scaffold was observed, however this was most often associated with partial degradation of the leaflet and leaflet thickening. In general, the degradation of the scaffold fibres was limited throughout the follow-up period.

Conclusion

We are the first to evaluate in vivo bioresorbable synthetic heart valves in aortic position. While valves remained functional, the study also serves as a starting point to further optimize scaffold and neotissue development for heart valve replacement in the high-pressure environment.

Funding Acknowledgement

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): 1. Cardiovasculair Onderzoek Nederland (grant number CVON2012-01)2. Netherlands Organization for Scientific Research (024.003.013)

Machine learning from quantitative coronary computed tomography angiography predicts ischemia and impaired myocardial blood flow

Background

Atherosclerotic plaque characteristics influence the hemodynamic consequences of coronary lesions. This study sought to assess the performance of a machine learning (ML) score integrating coronary computed tomography angiography (CCTA)-based quantitative plaque features for the prediction of ischemia by invasive fractional flow reserve (FFR) and impaired myocardial blood flow (MBF) by [15O]H2O positron emission tomography (PET).

Methods

This post-hoc analysis of the PACIFIC (Prospective Comparison of Cardiac PET/CT, SPECT/CT Perfusion Imaging and CT Coronary Angiography With Invasive Coronary Angiography) trial included 208 patients with suspected coronary artery disease who underwent CCTA, [15O]H2O PET, and 3-vessel invasive FFR. Plaque quantification from CCTA was performed using semiautomated software. A boosted ensemble ML algorithm (XGBoost) trained on data from the NXT (Analysis of Coronary Blood Flow using CT Angiography: Next Steps) trial was used to develop a ML score for the prediction of per-vessel ischemia (invasive FFR ≤0.80). The performance of the ML score was evaluated in 551 vessels from the PACIFIC trial for external validation. Thereafter, we assessed the discriminative ability of the ML score for per-vessel impaired hyperemic MBF (≤2.30 mL/min/g).

Results

In total, 138 (25.0%) vessels had ischemia and 195 (35.4%) vessels had impaired hyperemic MBF. CCTA-derived quantitative percent diameter stenosis and low-density noncalcified plaque (LDNCP) volume were higher in ischemic vessels compared with non-ischemic vessels (60.8% vs. 19.9%; and 42.3 mm3 vs. 9.1 mm3; both p<0.001). The ML score demonstrated a significantly higher area under the receiver-operating characteristic curve (AUC) for predicting ischemia (0.92, 95% confidence interval [CI] 0.89–0.94) compared with visual stenosis grade (0.84, 95% CI 0.80–0.87; p<0.001). Overall, quantitative percent diameter stenosis and LDNCP volume had greatest feature importance for ML, followed by percent area stenosis, minimum luminal diameter, and contrast density drop (Figure 1). An individualized explanation of ML ischemia prediction is shown in Figure 2. When applied for impaired MBF discrimination, the ML score exhibited an AUC of 0.82 (95% CI 0.78–0.85) and was superior to visual stenosis grade (AUC 0.76, 95% CI 0.72–0.80; p=0.03).

Conclusions

An externally validated ML score integrating CCTA-based quantitative plaque features accurately predicts FFR-defined ischemia and abnormal MBF by PET, outperforming standard visual CCTA interpretation.

Funding Acknowledgement

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): National Heart, Lung, and Blood Institute, United States Performance of the ML scoreIndividual explanation of ML prediction

Colorectal Cancer Prevention in Lung Transplant Recipients: The Need for an Enhanced Surveillance Protocol

BACKGROUND

Solid organ transplant recipients are at increased risk for noncutaneous neoplasms, including colorectal cancer (CRC). We evaluated precancerous lesions detected by post-transplant surveillance colonoscopy to infer the rate at which new adenomas develop in this population.

STUDY DESIGN

We reviewed all patients who underwent lung transplant between January 2013 and August 2017 at our institution. Those with post-transplant survival <1 year, personal history of CRC, previous lung transplant, and lack of pretransplant colonoscopy were excluded.

RESULTS

During the study period, 411 patients underwent lung transplant; 237 met inclusion criteria. Median age at transplant was 63.6 (interquartile range [IQR] 59.2-68.3) years. Most recipients were immunosuppressed with a combination of prednisone, tacrolimus, and mycophenolate mofetil. At least 1 adenoma was found in 92 patients (38.8%) pretransplant and in 118 patients (49.8%) from 1 to 5 years post-transplant, with 68.6% identified at 1 year. Most adenomas were identified proximal to the splenic flexure. Multiple (≥3) adenomas were found in 31.4% of positive colonoscopies. Within 5 years after transplant, patients with a positive pretransplant colonoscopy had significantly more positive post-transplant colonoscopies than patients with a negative pretransplant colonoscopy (63.0% vs 41.4%, p < 0.001). No de novo CRC was identified.

CONCLUSIONS

Lung transplant recipients have a significantly higher risk of adenoma formation than average-risk adults (25%-30% national detection rate). This increase occurs in the early post-transplant period (within 3 years). An enhanced CRC surveillance protocol for lung transplant recipients is needed.

A Comprehensive Review of the Diagnosis, Treatment, and Management of Urologic Chronic Pelvic Pain Syndrome

PURPOSE OF REVIEW

Urologic chronic pelvic pain syndrome (UCPPS) is a chronic, noncyclic pain condition which can lead to significant patient morbidity and disability. It is defined by pain in the pelvic region, lasting for greater than 3 to 6 months, with no readily identifiable disease process. The aim of this review is to provide a comprehensive update of diagnosis and treatment of UCPPS.

RECENT FINDINGS

UCPPS encompasses chronic pelvic pain syndrome or chronic prostatitis (CP/CPPS) in men and interstitial cystitis or painful bladder syndrome (IC/PBS) in women. Underlying inflammatory, immunologic, and neuropathic components have been implicated in the pathogenesis of UCPPS. For optimal patient management, an individualized and multimodal approach is recommended. Medical management and physical therapy are the mainstays of treatment. Injection therapy may offer additional relief in medically refractory patients. Further minimally invasive management may include spinal cord and peripheral nerve stimulation, though evidence supporting efficacy is limited.

Current mismatch repair deficiency tumor testing practices and capabilities: A survey of Australian pathology providers

AIM & METHODS

An electronic survey of the Royal College of Pathologists of Australasia accredited pathology services was conducted to assess Lynch syndrome tumor screening practices and to identify barriers and capabilities to screen newly diagnosed colorectal and endometrial tumors in Australia.

RESULTS

Australia lacks a national policy for universal mismatch repair-deficient (dMMR) testing of incident colorectal and endometrial tumors cases. Routine Lynch syndrome tumor screening program for colorectal and/or endometrial tumors was applied by 95% (37/39) of laboratories. Tumor dMMR screening methods varied; MMR protein immunohistochemistry (IHC) alone was undertaken by 77% of 39 laboratories, 18% performed both IHC and microsatellite instability testing, 5% did not have the capacity to perform in-house testing. For colorectal tumors, 47% (17/36) reported following a universal approach without age limit, 30% (11/36) tested only "red flag" cases; 6% (3/36) on clinician request only. For endometrial tumors, 37% (12/33) reported clinician request generated testing, 27% (9/33) were screening only "red flag" cases, and 12% (4/33) carried out universal screening without an age criteria. BRAF V600E mutation testing of colorectal tumors demonstrating aberrant MLH1 protein expression by IHC was the most common secondary tumor test, with 53% of laboratories performing the test; 15% of laboratories also applied the BRAF V600E test to endometrial tumors with aberrant MLH1 expression despite no evidence for its utility. Tumor testing for MLH1 promoter methylation was performed by less than 15% laboratories.

CONCLUSION

Although use of tumor screening for evidence of dMMR is widely available, protocols for its use in Australia vary widely. This national survey provides a snapshot of the current availability and practice of tumor dMMR screening and identifies the need for a uniform national testing policy.

Development of the PEA-PODS (Perceptions of the Environment and Patterns of Diet at School) Survey for Students

INTRODUCTION

Few instruments assess key outcomes of school-based obesity interventions, including student perceptions of school environments and school-specific dietary intake patterns. This study describes development of PEA-PODS (Perceptions of the Environment and Patterns of Diet at School), a 2-part survey to measure these outcomes.

METHODS

Part 1 (PEA) assessed student perceptions of policies, physical environment, and practices related to healthy eating and physical activity at school. Part 2 (PODS) assessed usual intake (ie, frequency, location obtained, and foods consumed) of breakfast and lunch. Foods consumed were presented by MyPlate categories (eg, Fruits, Grains). Students in grades 3, 6, and 9 participated in 2 phases: cognitive pre-testing (n = 10) and reliability/validation testing (n = 58). Both surveys were administered 1 week apart to assess test-retest reliability and 5-day food records validated PODS. Analyses included percent agreement (70% = acceptable), Pearson correlations, and Cronbach α.

RESULTS

Cognitive pre-testing provided feedback on content, length, and age-appropriateness. Percent agreements were acceptable for test-retest reliability of PEA (71%-96%). The final version included 34 items with Likert-type responses in 4 subscales (α ≥0.78). For PODS, agreement for breakfast and lunch location was ≥75% for both reliability and validation. For foods consumed at breakfast, reliability agreement ranged from 74% to 93%, and validation agreement from 68% to 91%. For foods consumed at lunch, agreement ranges were 76% to 95% and 73% to 88%, respectively.

CONCLUSION

Both parts of the instrument demonstrate acceptable reliability, and PODS demonstrates acceptable validity. This demonstrates appropriateness for assessing perceptions of the environment and usual dietary intake patterns for school-based obesity prevention initiatives.

Structure determination of two metal-organic complexes from high-resolution synchrotron powder diffraction data

The crystal structures of [1,2-bis(2,6-diisopropylphenylimino)acenaphthene-N,N']carbonylchlororhodium(I) (1) and [N,N'-ethylene-bis(3-methylsalicylideneiminato)-O,N,N',O'](tetrahydrofurfuryl)-cobalt(II) (2) have been determined from high-resolution synchrotron X-ray powder diffraction data. Compound 1 is the first neutral Rh complex, in contrast with findings in the literature, containing a bidentate nitrogen ligand, and compound 2 is the first three-dimensional structure of a (five-coordinated) tetrahydrofurfurylcobalt(III) complex. Grid-search and Rietveld refinement have been used to determine and refine the structures, respectively. Crystals of 1 are orthorhombic, space group Pbca, Z = 8, with cell parameters a = 21.729 (2), b = 27.376 (3), c = 11.580 (1) A. Crystals of 2 are monoclinic, space group P2(1)/n, Z = 4, a = 16.6701 (6), b = 9.4170 (4), c = 13.7088 (7) A and beta = 96.520 (3) degrees. Chemical diagrams for the two compounds are given. Soft restraints were applied during Rietveld refinement; for 1 converging to R(p) = 8.4%, R(w) = 11.0%, GoF = 2.3, and for 2 converging to R(p) = 8.5%, R(w) = 11.4%, GoF = 7.6.

Structure of CnCn+2Cn-type (n = even) beta'-triacylglycerols

The crystal structures of the beta' phase of CLC (1, 3-didecanoyl-2-dodecanoylglycerol) and MPM (1, 3-ditetradecanoyl-2-hexadecanoylglycerol) have been determined from single-crystal X-ray diffraction and high-resolution X-ray powder diffraction data, respectively. Both these crystals are orthorhombic with space group Iba2 and Z = 8. The unit-cell parameters of beta'-CLC are a = 57.368 (6), b = 22.783 (2) and c = 5.6945 (6) A and the final R value is 0.175. The unit-cell parameters of beta'-MPM are a = 76.21 (4), b = 22.63 (1) and c = 5.673 (2) A and the final R(p) value is 0.057. Both the beta'-CLC and beta'-MPM molecules are crystallized in a chair conformation, having a bend at the glycerol moiety. The zigzag planes of the acyl chains are orthogonally packed, as is typical for a beta' phase. Furthermore, unit-cell parameters of some other members of the C(n)C(n+2)C(n)-type triacylglycerol series have been refined on their high-resolution X-ray powder diffraction pattern. Finally, the crystal structures are compared with the currently known structures and models of triacylglycerols.