Automated abdominal CT contrast phase detection using an interpretable and open-source artificial intelligence algorithm

OBJECTIVES

To develop and validate an open-source artificial intelligence (AI) algorithm to accurately detect contrast phases in abdominal CT scans.

MATERIALS AND METHODS

Retrospective study aimed to develop an AI algorithm trained on 739 abdominal CT exams from 2016 to 2021, from 200 unique patients, covering 1545 axial series. We performed segmentation of five key anatomic structures-aorta, portal vein, inferior vena cava, renal parenchyma, and renal pelvis-using TotalSegmentator, a deep learning-based tool for multi-organ segmentation, and a rule-based approach to extract the renal pelvis. Radiomics features were extracted from the anatomical structures for use in a gradient-boosting classifier to identify four contrast phases: non-contrast, arterial, venous, and delayed. Internal and external validation was performed using the F1 score and other classification metrics, on the external dataset "VinDr-Multiphase CT".

RESULTS

The training dataset consisted of 172 patients (mean age, 70 years ± 8, 22% women), and the internal test set included 28 patients (mean age, 68 years ± 8, 14% women). In internal validation, the classifier achieved an accuracy of 92.3%, with an average F1 score of 90.7%. During external validation, the algorithm maintained an accuracy of 90.1%, with an average F1 score of 82.6%. Shapley feature attribution analysis indicated that renal and vascular radiodensity values were the most important for phase classification.

CONCLUSION

An open-source and interpretable AI algorithm accurately detects contrast phases in abdominal CT scans, with high accuracy and F1 scores in internal and external validation, confirming its generalization capability.

CLINICAL RELEVANCE STATEMENT

Contrast phase detection in abdominal CT scans is a critical step for downstream AI applications, deploying algorithms in the clinical setting, and for quantifying imaging biomarkers, ultimately allowing for better diagnostics and increased access to diagnostic imaging.

KEY POINTS

Digital Imaging and Communications in Medicine labels are inaccurate for determining the abdominal CT scan phase. AI provides great help in accurately discriminating the contrast phase. Accurate contrast phase determination aids downstream AI applications and biomarker quantification.

Abdominal CT metrics in 17,646 patients reveal associations between myopenia, myosteatosis, and medical phenotypes: a phenome-wide association study

BACKGROUND

Deep learning facilitates large-scale automated imaging evaluation of body composition. However, associations of body composition biomarkers with medical phenotypes have been underexplored. Phenome-wide association study (PheWAS) techniques search for medical phenotypes associated with biomarkers. A PheWAS integrating large-scale analysis of imaging biomarkers and electronic health record (EHR) data could discover previously unreported associations and validate expected associations. Here we use PheWAS methodology to determine the association of abdominal CT-based skeletal muscle metrics with medical phenotypes in a large North American cohort.

METHODS

An automated deep learning pipeline was used to measure skeletal muscle index (SMI; biomarker of myopenia) and skeletal muscle density (SMD; biomarker of myosteatosis) from abdominal CT scans of adults between 2012 and 2018. A PheWAS was performed with logistic regression using patient sex and age as covariates to assess for associations between CT-derived muscle metrics and 611 common EHR-derived medical phenotypes. PheWAS P values were considered significant at a Bonferroni corrected threshold (α = 0.05/1222).

FINDINGS

17,646 adults (mean age, 56 years ± 19 [SD]; 57.5% women) were included. CT-derived SMI was significantly associated with 268 medical phenotypes; SMD with 340 medical phenotypes. Previously unreported associations with the highest magnitude of significance included higher SMI with decreased cardiac dysrhythmias (OR [95% CI], 0.59 [0.55-0.64]; P < 0.0001), decreased epilepsy (OR, 0.59 [0.50-0.70]; P < 0.0001), and increased elevated prostate-specific antigen (OR, 1.84 [1.47-2.31]; P < 0.0001), and higher SMD with decreased decubitus ulcers (OR, 0.36 [0.31-0.42]; P < 0.0001), sleep disorders (OR, 0.39 [0.32-0.47]; P < 0.0001), and osteomyelitis (OR, 0.43 [0.36-0.52]; P < 0.0001).

INTERPRETATION

PheWAS methodology reveals previously unreported associations between CT-derived biomarkers of myopenia and myosteatosis and EHR medical phenotypes. The high-throughput PheWAS technique applied on a population scale can generate research hypotheses related to myopenia and myosteatosis and can be adapted to research possible associations of other imaging biomarkers with hundreds of EHR medical phenotypes.

FUNDING

National Institutes of Health, Stanford AIMI-HAI pilot grant, Stanford Precision Health and Integrated Diagnostics, Stanford Cardiovascular Institute, Stanford Center for Digital Health, and Stanford Knight-Hennessy Scholars.

Beyond the Binary: Moving the Radiology Workforce Toward Gender Inclusion, From the AJR Special Series on DEI

Gender representation in radiology has traditionally been evaluated and reported through binary models, accompanied by advocacy efforts focused on increasing the number of women in radiology. A paucity of data exists to understand the entire gender composition of the radiology workforce, including representation of people who are transgender and gender diverse. Further, little information exists on how to provide a supportive work environment for radiologists and support staff who identify as belonging to an underrepresented gender minority group. Intentional efforts to comprehensively understand the gender representation of the radiology workforce can help to establish a diverse workforce that is more representative of the patient populations that we serve, while promoting high-quality inclusive health care. Moving beyond gender binary thought and practices can help foster a culture of inclusion and belonging in radiology. This article provides practical steps that radiology practices can take to understand and support gender diversity beyond the binary in the radiology workforce, including providing definitions and inclusive language, understanding limitations of historical methods of gender data collection in radiology and relevant published literature, establishing best practices for future data collection, and developing a strategic vision with action items to create a more inclusive work environment.

An Upstream Reparative Justice Framework for Improving Diversity in Radiology

Healthcare workforce diversity is vital in combating health disparities. Despite much recent attention to downstream strategies to improve diversity in radiology, such as increased recruitment efforts and holistic application review, workforce diversity has not tangibly improved in recent decades. Yet, little discussion has been devoted to defining the obstacles which might delay, complicate, or altogether prevent persons from groups that have been traditionally marginalized and minoritized from a career in radiology. Refocusing attention to upstream barriers to medical education is vital to develop sustainable workforce diversity efforts in radiology. The purpose of this manuscript is to highlight the varied obstacles students and trainees from historically under-represented communities may face along the radiology career pathway and to provide concrete corollary programmatic solutions. Utilizing a reparative justice framework, which encourages race- and gender-conscious repair of historical injustices, and the socio-ecological model, which recognizes an individual's choices are informed by historical and ongoing systems of power, this manuscript advocates for tailored programs to improve justice, equity, diversity, and inclusion (JEDI) in radiology.

Osteopathic Versus Allopathic Radiologist Workforce Characteristics: A Medicare Administrative and Claims Data Analysis

PURPOSE

Radiologist medical school pathways have received little attention in recent workforce investigations. With osteopathic enrollment increasing, we assessed the osteopathic versus allopathic composition of the radiologist workforce.

METHODS

Linking separate Medicare Doctors and Clinicians Initiative databases and Physician and Other Supplier Files from 2014 through 2019, we assessed (descriptively and using multivariate panel logistic regression modeling) individual and practice characteristics of radiologists who self-reported medical degrees.

RESULTS

Between 2014 and 2019, as the number of osteopathic radiologists increased 46.0% (4.7% to 6.0% of total radiologist workforce), the number of allopathic radiologists increased 12.1% (representing a relative workforce decrease from 95.3% to 94.0%). For each year since completing training, practicing radiologists were 3.7% less likely to have osteopathic (versus allopathic) degrees (odds ratio [OR] = 0.96 per year, P < .01). Osteopathic radiologists were less likely to work in urban (versus rural) areas (OR = 0.95), and compared with the Midwest, less likely to work in the Northeast (OR = 0.96), South (OR = 0.95), and West (OR = 0.94) (all P < .01). Except for cardiothoracic imaging (OR = 0.78, P = .24), osteopathic radiologists were more likely than allopathic radiologists to practice as general (rather than subspecialty) radiologists (range OR = 0.37 for nuclear medicine to OR = 0.65 for neuroradiology, all P < .01).

CONCLUSIONS

Osteopathic physicians represent a fast-growing earlier-career component of the radiologist workforce. Compared with allopathic radiologists, they more frequently practice as generalist radiologists, in rural areas, and in the Midwest. Given recent calls for greater general and rural radiology coverage, increasing osteopathic representation in the national radiologist workforce could improve patient access.

Imaging of Sarcopenia

Sarcopenia is currently underdiagnosed and undertreated, but this is expected to change because sarcopenia is now recognized with a specific diagnosis code that can be used for billing in some countries, as well as an expanding body of research on prevention, diagnosis, and management. This article focuses on practical issues of increasing interest by highlighting 3 hot topics fundamental to understanding sarcopenia in older adults: definitions and terminology, current diagnostic imaging techniques, and the emerging role of opportunistic computed tomography.

Variables Influencing Radiology Volume Recovery During the Next Phase of the Coronavirus Disease 2019 (COVID-19) Pandemic

The coronavirus disease 2019 (COVID-19) pandemic has reduced radiology volumes across the country as providers have decreased elective care to minimize the spread of infection and free up health care delivery system capacity. After the stay-at-home order was issued in our county, imaging volumes at our institution decreased to approximately 46% of baseline volumes, similar to the experience of other radiology practices. Given the substantial differences in severity and timing of the disease in different geographic regions, estimating resumption of radiology volumes will be one of the next major challenges for radiology practices. We hypothesize that there are six major variables that will likely predict radiology volumes: (1) severity of disease in the local region, including potential subsequent "waves" of infection; (2) lifting of government social distancing restrictions; (3) patient concern regarding risk of leaving home and entering imaging facilities; (4) management of pent-up demand for imaging delayed during the acute phase of the pandemic, including institutional capacity; (5) impact of the economic downturn on health insurance and ability to pay for imaging; and (6) radiology practice profile reflecting amount of elective imaging performed, including type of patients seen by the radiology practice such as emergency, inpatient, outpatient mix and subspecialty types. We encourage radiology practice leaders to use these and other relevant variables to plan for the coming weeks and to work collaboratively with local health system and governmental leaders to help ensure that needed patient care is restored as quickly as the environment will safely permit.

Optimizing Performance by Preventing Disruptive Behavior in Radiology

Disruptive behaviors impede delivery of high-value health care by negatively impacting patient outcomes and increasing costs. Health care is brimming with potential triggers of disruptive behavior. Given omnipresent environmental and cultural factors such as constrained resources, stressful environments, commercialization, fatigue, unrealistic expectation of perfectionism, and burdensome documentation, a burnout epidemic is raging, and medical providers are understandably at tremendous risk to succumb and manifest these unprofessional behaviors. Each medical specialty has its own unique challenges. Radiology is not exempt; these issues do not respect specialty or professional boundaries. Unfortunately, preventive measures are too frequently overlooked, provider support programs rarely exist, and often organizations either tolerate or ineffectively manage the downstream disruptive behaviors. This review summarizes the background, key definitions, contributing factors, impact, prevention, and management of disruptive behavior. Every member of the health care team can gain from an improved understanding and awareness of the contributing factors and preventive measures. Application of these principles can foster a just culture of understanding, trust, support, respect, and teamwork balanced with accountability. The authors discuss these general topics along with specific issues for radiologists in the current medical environment. Patients, providers, health care organizations, and society all stand to benefit from better prevention of these behaviors. There is a strong moral, ethical, and business case to address this issue head-on. ^©RSNA, 2018.

Clinical Decision Support at the Point-of-Order Entry: An Education Simulation Pilot with Medical Students

RATIONALE AND OBJECTIVES

We have been called to reform radiology undergraduate medical education (UME) curricula. Clinically available clinical decision support provides an opportunity to improve education regarding appropriate imaging utilization, patient safety, and cost-effective care.

MATERIAL AND METHODS

We created an education simulation portal utilizing integrated clinical decision support. The portal was then piloted with 34 volunteer medical students at our institution in a blended learning environment. A program assessment was performed utilizing the results from a qualitative survey, pre-test, and post-test.

RESULTS

The large majority of medical students felt this supplemental education resource should be included in our UME curriculum (85.29%). All students perceived value in the education simulation portal. The students performed significantly better on the post-test in multiple categories (overall P <.0001), including Choosing Wisely topics (P = .0207).

CONCLUSIONS

Based on our program assessment from this pilot program, we believe this innovative educational resource has significant potential to fill curricular gaps in radiology UME curricula. This platform is scalable and can be further customized to fill needs across the continuum of medical education.

Measurement of ulnar variance from the lateral radiograph: a comparison of techniques

PURPOSE

To determine the reliability of measuring ulnar variance on lateral wrist radiographs and to compare this technique with previously described methods.

METHODS

Ulnar variance was measured in 100 normal wrist radiographs using the methods of perpendiculars, central reference point, and the lateral radiograph by 3 surgeons on 2 occasions. Intraobserver repeatability and agreement between raters and methods were assessed and compared.

RESULTS

Intra- and interobserver reliability and agreement were both excellent using all 3 methods within a ± 1.0-mm cutoff. However, there was substantial pairwise disagreement in measures of ulnar variance between all 3 methods.

CONCLUSIONS

This study demonstrates that, for measurement of ulnar variance, the methods of perpendiculars, central reference point, and lateral radiographic measurement each have clinically acceptable intraobserver repeatability and interobserver agreement. Despite their independent reliability, each method of radiographic determination of ulnar variance had considerable disagreement with the other methods, indicative of inherent inaccuracies in the techniques. The lateral radiograph uniquely allows for visualization of the amount of ulnar head protruding proximal or distal to the concave lunate facet and allows for a rapid estimation of pronosupination, which is known to affect ulnar variance.

CLINICAL RELEVANCE

Determination of ulnar variance can be an important component of surgical decision making in various pathological conditions of the hand and wrist. Traditionally, it has been measured through methods using the posteroanterior wrist radiograph, but there are potential shortcomings with these methods, and use of the lateral radiograph may provide a more clinically relevant picture of ulnar variance. This study shows that measurement from the lateral radiograph provides similar reliability to previously accepted techniques.