Recent advances and future perspectives of machine learning techniques offer promising applications in medical imaging. Machine learning has the potential to improve different steps of the radiology ...workflow including order scheduling and triage, clinical decision support systems, detection and interpretation of findings, postprocessing and dose estimation, examination quality control, and radiology reporting. In this article, the authors review examples of current applications of machine learning and artificial intelligence techniques in diagnostic radiology. In addition, the future impact and natural extension of these techniques in radiology practice are discussed.
Abstract
Human immunodeficiency virus (HIV)–associated neurocognitive disorders (HAND) criteria are frequently used to describe cognitive impairment in persons living with HIV (PLWH) across diverse ...populations globally. These criteria typically find 20–60% of PLWH meet criteria for HAND, which does not tally with clinical observations in the modern era that cognitive disorders present relatively infrequently. Most with HAND have asymptomatic neurocognitive impairment; however, the significance of low cognitive test performance without symptoms is uncertain. Methods underlying HAND criteria carry a false-positive rate that can exceed 20%. Comorbidities, education, and complex socioeconomic factors can influence cognitive test performance, further increasing the potential for misclassification. We propose a new framework to characterize cognitive impairment in PLWH that requires a clinical history and acknowledges the multifactorial nature of low cognitive test performance. This framework is intended to be applicable across diverse populations globally, be more aligned with clinical observations, and more closely represent HIV brain pathology.
Abstract Multiple theories have been proposed for sensation seeking and vulnerability to impulse-control disorders Zuckerman M, Kuhlman DM (2000) Personality and risk-taking: Common biosocial ...factors. J Pers 68:999–1029, and many of these rely on a dopamine system deficit. Available animal models reproduce only some behavioral symptoms and seem devoid of construct validity. We used lentivirus tools for over-expressing or silencing the dopamine transporter (DAT) and we evaluated the resulting behavioral profiles in terms of motivation and self-control. Wistar adult rats received stereotaxic inoculation of a lentivirus that allowed localized intra-accumbens delivery of a DAT gene enhancer/silencer, or the green fluorescent protein, GFP. These animals were studied for intolerance to delay, risk proneness and novelty seeking. As expected, controls shifted their demanding from a large reward toward a small one when the delivery of the former was increasingly delayed (or uncertain). Interestingly, in the absence of general locomotor effects, DAT over-expressing rats showed increased impulsivity (i.e. a more marked shift of demanding from the large/delayed toward the small/soon reward), and increased risk proneness (i.e. a less marked shift from the large/uncertain toward the small/sure reward), compared with controls. Rats with enhanced or silenced DAT expression did not show any significant preference for a novel environment. In summary, consistent with literature on comorbidity between attention-deficit/hyperactivity disorder and pathological gambling, we demonstrate that DAT over-expression in rats' nucleus accumbens leads to impulsive and risk prone phenotype. Thus, a reduced dopaminergic tone following altered accumbal DAT function may subserve a sensation-seeker phenotype and the vulnerability to impulse-control disorders.
The prognostic role of persistence of circulating tumor cells (CTC) after upfront tumor surgery for outcome of adjuvant (chemo)radiation in locally advanced squamous cell carcinoma of the head and ...neck (LASCCHN) was evaluated.
In this prospective study, peripheral blood samples from 144 patients with LASCCHN presenting after tumor resection for adjuvant treatment were analyzed for CTC. Their detection was correlated with tumor site, clinical risk factors, disease-free (DFS) and overall survival (OS).
CTC were detected in 42 of 144 patients (29%). CTC detection was higher in cases with nodal involvement and in carcinomas located at the tonsil or base of tongue but was not influenced by age, smoking history, T stage, extracapsular lymph node extension, surgical margins or the human papillomavirus status. Overall, the presence of CTC was not predictive for OS or DFS. However, while in oropharyngeal carcinomas (OPC, n = 63), the detection of CTC was associated per trend with improved DFS CTC+ versus CTC- (% of patients without evidence of disease at 2 years): 100% versus 79%; log rank: P = 0.059; the reverse was observed for carcinomas from other sites (non-OPC, n = 81; CTC+ versus CTC-: 29% versus 75%; P = 0.001). In multivariate analysis, CTC remained an independent prognostic marker for DFS hazard ratio (HR) 4.3, 95% confidence interval (CI) 1.7–10.9, P = 0.002 and OS (HR 2.7, 95% CI 1.2–6.3, P = 0.016) in non-OPC.
Assessment of CTC in non-OPC should prove useful for identification of patients who benefit from treatment intensification. The basis for the good prognostic value of CTC in OPC has to be elucidated in future studies.
To assess key trends, strengths, and gaps in validation studies of the Food and Drug Administration (FDA)-regulated imaging-based artificial intelligence/machine learning (AI/ML) algorithms.
We ...audited publicly available details of regulated AI/ML algorithms in imaging from 2008 until April 2021. We reviewed 127 regulated software (118 AI/ML) to classify information related to their parent company, subspecialty, body area and specific anatomy type, imaging modality, date of FDA clearance, indications for use, target pathology (such as trauma) and findings (such as fracture), technique (CAD triage, CAD detection and/or characterization, CAD acquisition or improvement, and image processing/quantification), product performance, presence, type, strength and availability of clinical validation data. Pertaining to validation data, where available, we recorded the number of patients or studies included, sensitivity, specificity, accuracy, and/or receiver operating characteristic area under the curve, along with information on ground-truthing of use-cases. Data were analyzed with pivot tables and charts for descriptive statistics and trends.
We noted an increasing number of FDA-regulated AI/ML from 2008 to 2021. Seventeen (17/118) regulated AI/ML algorithms posted no validation claims or data. Just 9/118 reviewed AI/ML algorithms had a validation dataset sizes of over 1000 patients. The most common type of AI/ML included image processing/quantification (IPQ; n = 59/118), and triage (CADt; n = 27/118). Brain, breast, and lungs dominated the targeted body regions of interest.
Insufficient public information on validation datasets in several FDA-regulated AI/ML algorithms makes it difficult to justify clinical applications since their generalizability and presence of bias cannot be inferred.
Background Human papillomavirus (HPV)-based primary screening guidelines are based on screening test performance and prevalence data generated in high-resource areas with low HIV infection rates. ...There is an urgent need for local data on infection and disease prevalence, as well as screening test performance, among both HIV-positive and HIV-negative South African (SA) women, in order to inform updated screening guidelines.Objectives This study describes the baseline characteristics of participants in the cross-sectional phase of the multicentric DIAgnosis in Vaccine And Cervical Cancer Screen (DiaVACCS) screening trial. The objective was to determine the prevalence of positive screening and pre-invasive disease using different tests and strategies in the SA HIV-positive and HIV-negative population.Methods A total of 1 104 women aged 25 - 65 years and eligible for screening were included, 465 HIV positive and 639 HIV negative. Visual inspection and molecular and cytological screening tests were done on self-sampled and healthcare worker-collected specimens. All participants who screened positive and 49.1% of those who screened negative were invited for colposcopy and biopsy, and those qualifying for treatment were recalled for large loop excision of the transformation zone as part of the trial. The worst histology result for each participant was used, and for untested women, multiple imputation was used to estimate verification bias-adjusted histology values.Results Visual inspection was positive in 50.4% of HIV-positive v. 20.9% of HIV-negative women, cytology (atypical squamous cells of undetermined significance) in 39.9% v. 17.0%, and high-risk HPV DNA in 41.2% v. 19.6%. Overall, high-grade squamous intraepithelial lesion-positive cytology peaked in the age group 30 - 39 years at 16.7%. After adjustment for verification bias, histological diagnosis of cervical intraepithelial neoplasia (CIN)2+ was suspected in 44.7% v. 23.5% and CIN3+ in 23.3% v. 10.2% of HIV-positive and negative women, respectively. Invasive cancer was diagnosed in 15 women (1.95% of histological studies performed), and verification bias adjustment suggested 20 cases (1.8% of the study population).Conclusion The baseline findings from the DiaVACCS trial confirm a high prevalence of HPV-related cervical pathology in the SA HIV-negative screening population, showing a clear need to reach these women with a screening programme. Among HIV-positive women, prevalence values were almost doubled. The prevalence of existing invasive cervical cancer was 1 - 2% of all women. Further analysis of the performance of single and multiple screening tests between the two subgroups will contribute to the choice of the most effective strategies to identify women at risk of developing invasive cancer.
As the role of artificial intelligence (AI) in clinical practice evolves, governance structures oversee the implementation, maintenance, and monitoring of clinical AI algorithms to enhance quality, ...manage resources, and ensure patient safety. In this article, a framework is established for the infrastructure required for clinical AI implementation and presents a road map for governance. The road map answers four key questions: Who decides which tools to implement? What factors should be considered when assessing an application for implementation? How should applications be implemented in clinical practice? Finally, how should tools be monitored and maintained after clinical implementation? Among the many challenges for the implementation of AI in clinical practice, devising flexible governance structures that can quickly adapt to a changing environment will be essential to ensure quality patient care and practice improvement objectives.
Worldwide over 40% of patients receiving renal replacement therapy (RRT) are aged 65 years or older, a number that is still increasing. Renal transplantation is the preferred RRT, providing ...substantial survival benefit over those remaining on dialysis, including the elderly. Only 3% of patients aged 65 years or older accepted on the waiting list actually received a kidney transplant offer within the Eurotransplant allocation region. To increase the chance for elderly to receive a timely kidney transplant, the Eurotransplant Senior Program was introduced. The ESP supports local allocation of older kidneys to older donors in order to decrease cold ischemia time, while disregarding former exchange principles based on matching for HLA antigens. As a consequence, more elderly received a kidney transplant and a relative higher incidence of acute rejection resulted in additional courses of high steroids and/or depleting antibody therapy. Since death with a functioning graft due to infections is the dominant reason of graft loss in elderly, more intense clinical immunosuppression to prevent or treat acute rejection is not a very attractive option. Therefore in elderly kidney transplant candidates, we advocate reintroduction of minimal histocompatibility criteria (i.e., HLA-DR matching) followed by age-matching with mandatory local/regional allocation to also facilitate short cold ischemia.
ABSTRACT
We investigate the role of secondary electron and ion emission from impact of gas molecules on the Cassini Langmuir probe (RPWS-LP or LP) measurements in the ionosphere of Saturn. We add a ...model of the emission currents, based on laboratory measurements and data from comet 1P/Halley, to the equations used to derive plasma parameters from LP bias voltage sweeps. Reanalysing several hundred sweeps from the Cassini Grand Finale orbits, we find reasonable explanations for three open conundrums from previous LP studies of the Saturn ionosphere. We find an explanation for the observed positive charging of the Cassini spacecraft, the possibly overestimated ionospheric electron temperatures, and the excess ion current reported. For the sweeps analysed in detail, we do not find (indirect or direct) evidence of dust having a significant charge-carrying role in Saturn’s ionosphere. We also produce an estimate of H2O number density from the last six revolutions of Cassini through Saturn’s ionosphere in greater detail than reported by the Ion and Neutral Mass Spectrometer. Our analysis reveals an ionosphere that is highly structured in latitude across all six final revolutions, with mixing ratios varying with two orders of magnitude in latitude and one order of magnitude between revolutions and altitude. The result is generally consistent with an empirical photochemistry model balancing the production of H+ ions with the H+ loss through charge transfer with e.g. H2O, CH4, and CO2, for which water vapour appears as the likeliest dominant source of the signal in terms of yield and concentration.
Abstract Background Few data exist on prevalence, morbidity, and mortality of pediatric heart failure hospitalizations. We tested the hypotheses that pediatric heart failure–related hospitalizations ...increased over time but that mortality decreased. Factors associated with mortality and length of stay were also assessed. Methods and Results A retrospective analysis of the Healthcare Cost and Utilization Project Kids' Inpatient Database was performed for pediatric (age ≤18 years) heart failure–related hospitalizations for the years 1997, 2000, 2003, and 2006. Hospitalizations did not significantly increase over time, ranging from 11,153 (95% confidence interval CI 8,898–13,409) in 2003 to 13,892 (95% CI 11,528–16,256) in 2006. Hospital length of stay increased from 1997 (mean 13.8 days, 95% CI 12.5–15.2) to 2006 (mean 19.4 days, 95% CI 18.2 to 20.6). Hospital mortality was 7.3% (95% CI 6.9–8.0) and did not vary significantly between years; however, risk-adjusted mortality was less in 2006 (odds ratio 0.70, 95% CI 0.61 to 0.80). The greatest risk of mortality occurred with extracorporeal membrane oxygenation, acute renal failure, and sepsis. Conclusions Heart failure–related hospitalizations occur in 11,000–14,000 children annually in the United States, with an overall mortality of 7%. Many comorbid conditions influenced hospital mortality.
PDF
