Quantitative imaging workshop XIX: Utilizing quantitative thoracic imaging to optimize population health final summary

Lung cancer screening involves the use of thoracic CT for both detection and measurements of suspicious lung nodules to guide the screening management. Since lung cancer screening eligibility typically requires age over 50 years along with >20 pack‐year tobacco exposure, thoracic CT scans also frequently reveal evidence for pulmonary emphysema as well as coronary artery calcification. These three thoracic diseases are collectively three of the leading causes of premature death across the world. Screening for the major thoracic diseases in this heavily tobacco‐exposed cohort is broadening the focus of lung cancer screening to a more comprehensive health evaluation including discussing the relevance of screen‐detected findings of the heart and lung parenchyma. The status and implications of these emerging issues were reviewed in a multidisciplinary workshop focused on the process of quantitative imaging in the lung cancer screening setting to guide the evolution of this important new area of public health.


| INTRODUCTION
The Prevent Cancer Foundation and American Lung Association cosponsored the 19th annual Quantitative Imaging Workshop (QIW) on 3-4 November 2022.The specific goal of this 19th annual Workshop was advancing the use of quantitative CT imaging to detect not only lung cancer but other major thoracic, smoking-related diseases occurring in older individuals.While the QIW has focused on computed tomography (CT) imaging related to lung cancer screening, this activity also finds additional tobacco-related disease information.Interest in lung screening is fueled by new reports demonstrating a higher portion of early-stage lung cancer cases with a lower portion of advanced (stages 3/4) cases compared to standard detection approaches. 1,2The United States Preventative Services Task Force (USPSTF) expanded their recommended eligibility for lung cancer screening to include younger people (50 and over) and lower tobacco exposure (with 20 pack years of tobacco use). 3The USPSTF's revision reflects the publication of confirmatory finding of significant lung cancer mortality reduction associated with thoracic CT detection from The NELSON trial. 4The NELSON trial was the first major screening trial to use quantitative imaging to detect suspected lung cancers, which resulted in a more efficient lung cancer case detection in a lower-risk, screen-eligible cohort. 4Improved results with quantitative CT imaging have allowed screening management optimization. 5Nations are now working to assemble relevant multidisciplinary teams to responsibly deliver lung screening services.Advances in imaging technologies are also expanding the scope of clinical information available through the quantitative assessment of thoracic screening CT scans.
The QIW explores issues with the possible applications of quantitative imaging with early tobacco-related disease as well as illuminating barriers to implementing such innovative approaches to accelerate public health benefit in this space.Accordingly addressing these complex challenges requires a broad array of health professionals who have roles in the optimal implementation of high-quality lung cancer screening.By design, QIW functions as "roll-up the sleeves" meetings where participants can discuss critical barriers to screening success and develop promising solutions.Unfortunately, the pace of lung cancer screening uptake is slow. 6The COVID-19 pandemic was a barrier to screening participation and the pandemic was associated with higherthan-expected rates of late-stage lung cancers. 7The potential of lung cancer screening to save lives has not yet been realized.QIW XIX was held in 2022 in a hybrid format with some faculty participating on-site participants from a wide range of countries linked into the Workshop proceedings via video conferencing technology.With the permission of the presenting faculty, QIW presentations and discussions are freely available on the Prevent Cancer Foundation website. 8bacco use is associated with the development of lung cancer, emphysema and coronary artery disease, and these three thoracic diseases are collectively the three leading causes of premature death across the world. 9Since lung cancer screening eligibility requires age over 50 years along with >20 pack-year tobacco exposure, evaluation of the screening thoracic CT imaging routinely detects other major co-morbid conditions including evidence of emphysema and/or coronary calcium more frequently than lung cancer.Screening for the major thoracic diseases in this heavily tobacco-exposed cohort is providing a public health opportunity for enhancing the benefit of lung cancer screening with a more comprehensive health evaluation including discussing the relevance of screen-detected findings of the heart and lung parenchyma.
An interest of the Workshop has been to ensure equitable access to lung cancer screening services for all relevant at-risk communities.
Cost is a major concern for screening participants from getting time-off from work to paying for transportation to a screening center. 9Therefore, getting more important health information from the thoracic CT obtained for lung cancer screening may enhance the value of this process.The QIW Workshop participants recently published a perspective about the opportunity to detect and manage emphysema, highlighting this example of additional health benefits that can emerge by using more complete imaging information routinely available from for the lung cancer screening process. 10This is a remarkable opportunity in leveraging the monetary investment in lung cancer screening which requires the performance of a low dose CT, since that screening image can also be reviewed for evidence of pulmonary emphysema.This additional information can be used to personalize the recommendations for existing preventive interventions such as with smoking cessation if relevant, or with other existing recommendations for pneumococcal and flu vaccination as well as physical activity.These are underutilized and yet economical preventive services could be of particular benefit for the even higher risk subset of screening participants who are discovered to have emphysema for the first time on their screening CT.
Additional thoracic CT findings during lung cancer screening such as with detecting major chronic diseases such as coronary calcium also can potentially benefit from secondary preventive interventions.We have recently called for urgent research to explore the actual benefit of preventive interventions such as with smoking cessation and increased physical activity as thoracic CT screening routinely detects more case of emphysema and coronary calcium than cases of lung cancer. 10,11The QIW series has focused on the quality-control issues to acquire a more optimized thoracic CT scan for reliable quantitative assessment for lung cancer now will include robust quality measurements for these three major tobacco-related diseases.This more tobacco-exposure inclusive strategy could convey important health benefits at acceptable cost especially if preventive interventions based on quantitative imaging findings from analysis of annual thoracic CT scans are refined for optimal health benefit in this cohort of older, tobacco-exposed individuals.
The takeaway from this discussion is that objective evidence will be required to demonstrate the impact of reanalysis of a thoracic CT to improve health outcomes.This included defining any new clinical activities required for managing the additional thoracic CT information to achieve the health benefit along with a favorable cost/benefit profile for the new screening management.

| RELEVANCE OF POPULATION HEALTH-REIMBURSEMENT FOR AI, ENSURING QUALITY AND THE ROLES OF PAYERS: PAYING FOR AI AND QUALITY
A real-world consideration with extracting more information from thoracic CT scans performed for lung cancer screenings is whether payers will provide additional compensation for performing additional image analysis.This is particularly important in the United States, but the evolution of thoracic CT screening will have important cost implications in other national settings as well.The purpose of this panel was to define emerging trends in payer reimbursement for AI services-and physician AI services.Richard Frank, MD, report the new taxonomy for several distinct kinds of AI services that will be the basis for Medicare physician fee schedule's approach to AI services, including machine learning and AI analysis of the thoracic CT scan acquired for lung cancer screening. 12Zeke Silva, MD, a radiologist, and chair of American Medical Association's Digital Medicine Payment Advisory Group, provided an explanation of the components of the resource-based relative value scale, which is used to set the relative value of different CPT© codes, and how that is relevant to emerging AI services.The discussion included the relevance of these codes for lung cancer screening, 9 and how changes in the Medicare physician fee schedule operates on a zerosum basis-increases in some fees are offset by decreases in other fees.In the United States as in the rest of the world, expanding the scope of lung cancer screening to include other tobacco-related diseases routinely detected on annual thoracic CT screening will have to be developed by demonstrating objective evidence of improved health outcomes using health interventions that can be provided within tight cost constraints.

| Innovation and population health: Roles of payers and radiology benefit managers
The purpose of this panel was to explain the roles of Radiology Benefit Managers (RBMs) and payers in delivering CT imaging to tens of millions of Americans, and to gain insight into their capabilities to help advance lung cancer screening implementation.RBMs are organizations that contract with payers to manage the radiology benefits of the payers' covered lives.The RBM's delegated functions may include establishing or maintaining a network of radiology providers, establishing provider reimbursement, medical management of the radiology benefit including prior authorization, and quality controls.
RBMs and payers will be key to scaling lung cancer screening nationally.Ideally, RBMs-or others like them-will be able to contribute to both increasing the rate of screening and its quality.Panelists were from two large RBMs, one Blue Cross Blue Shield plan, one integrated delivery system and a specialty hospital.Discussions included the potential roles of RBMs in helping a member's prior screening results be available to the provider of their current screening, measuring provider quality including radiation dose, and shifting patient imaging services to lower-cost community sites rather than more expensive hospital sites.These trends have implications for national screening infrastructure as the ambulatory screening in many settings has not been set-up to fluidly store and enable comparisons of full thoracic CT imaging data required for annual quantitative screening analysis.However, in considering solutions for this CT screening implementation challenge, an additional, positive outcome of more robust thoracic CT image storage capability, could be a more efficient infrastructure to aggregate screening thoracic CT cases along with clinical outcomes.Such a CT image data/ clinical outcomes resource not only could anchor screening management but also enable more rapid tool validation research for optimizing thoracic CT screening.

| CLINICAL IMPLICATIONS OF REPORTING EMPHYSEMA ON LOW-DOSE CT SCANS FOR CANCER SCREENING
This session further explored issues related to the ability to extract additional imaging information about other tobacco-related diseases such as emphysema on annual thoracic CT screenings performed for lung cancer. 10Presenters including leading pulmonologists and radiologists explored the ability of widely available quantitative imaging technology and the importance of image quality considerations to routinely detect evidence of early emphysema in asymptomatic screening participants. 13It is therefore appropriate to consider what next steps could be beneficial for screening subjects found to have emphysema.At the current time, since there are so few articles on this subject, so Dr. Rizzo, Chief Medical Officer of the American Lung Association volunteered to contact and convene representatives from the American Lung Association, American Thoracic Society, American College of Chest Physicians and the COPD Foundation and the American College of Radiology to address this opportunity to define best medical advice of screen-detected emphysema in asymptomatic individuals.Given the growth in lung cancer screening participation, developing a guidance document outlining potential screening management options for emphysema findings along with research measures for responsibly implementing quantitative emphysema imaging analysis is an urgent priority.
Further discussion focused on how to propose activities which may spur progress with lung cancer screening uptake and then improve the efficiency of screening process workflow.However, there was also discussion about how to ensure that the benefits of these powerful new informatics resources are used equitably for all potential beneficiaries including members of underserved populations.

| Proposed action plans
Concrete activities over the next year to spur progress with screening implementation include: (1) A demonstration project with screening program navigators to evaluate if additional emphysema information enhances adherence with annual screening, especially for underserved populations; (2) A demonstration pilot to evaluate the impact of physical activity on imaging-related monitoring of annual lung cancer/ emphysema status to determine if this intervention for lung cancer screening participants improves health or screening compliance outcomes.

| INTERSECTION OF HEALTH AND WELLNESS: QUANTITATIVE IMAGING FOR EARLY DISEASE MANAGEMENT
This session included a description of what people are currently hearing about screening and competing interpretations regarding the potential benefit.The most important benefit is that screen-detected cancers are typically asymptomatic, localized cancers and can frequently be cured.With symptom-detected lung cancer, the disease is typically found to be already metastatic and therefore not curable.The session's focus was how to inform the public about the progress with early detection to emphasize the goal to find early lung cancer which can frequently be cured with limited surgical procedure, pre-empting the progression to lethal metastatic lung cancer. 5,14Panelists discussed an interesting recent study that analyzed survey results for screening participants and their physicians.The analysis found that screening participants were more optimistic than clinicians relative to perceptions of screening benefit.Yet clinician recommendations were still influential in the decision of at-risk individuals to engage in lung cancer screening.
During the panels, the possibility that the earlier detection of major, non-lung-cancer, co-morbidities could emerge as a motivation for eligible screening participants to remain adherent with annual follow-up imaging.Therefore, attention to defining the specific components of harms and benefits of ongoing annual screening participation merit further consideration.Further, panelists discussed concerns that the frequently the available material informing about the balance of screening benefit vs. harms has been presented in an unduly pessimistic fashion, which is not reflective of the current screening experience data.For example, the dose of medical radiation currently used to obtain a screening CT image continues to fall and so the actual harm of medical radiation exposure in this screening setting has correspondingly receded especially given the context of the older, tobaccoexposed individuals.Yet the unlikely occurrence of medical radiation harm continues to be presented as a major harm associated with annual lung cancer screening participation.A consensus among Workshop participants emerged that recalibrating the harms/benefit discussion to reflect the growing evidence of screening benefit along with improved screening management may influence more individuals to participate in recommended lung cancer screening measure.This discussion was followed by a provocative presentation from Voroni Health on the many different body parts that can be measured using CT scans of the chest including measuring of the body content of muscle and fat within the thorax.Additional work is required in considering body fat composition to define in what settings fat distributions are associated with various illnesses including metabolic syndrome and heart disease.An example of one such imaging technology was presented demonstrating the utility of STYKU, a 3D technique using cameras that produce images based on the tissue composition of the body. 15Here various circumferences from multiple body sites are measured automatically and compared to people of comparable age as well as to a group of young normal volunteers.This technique is currently being used in physician offices as well as in health clubs.
The panel discussion focused on the importance of measuring body composition and tracking it, and the need to better develop standards for normal and disease states.Fat distribution and quality along with muscle distribution and quality were considered to have potential for predicting a variety of important health issues including response to treatment.Further research is required to guide quantitative assessments of tissue compartments within the thoracic screening setting.

| EXPLORING NEW CT IMAGING AND AI TECHNICAL OPPORTUNITIES
A key example of a new imaging platform is the emergence of photon counting CTs with exquisite spatial resolution capabilities even with exceptionally low radiation dose.This tool could have tangible impact on lung cancer screening and nodule management related to its advances in in-plane resolution (0.14 mm) and minimum slice thickness (0.2 mm).Such improvements in resolution could improve the ability to measure change in the volume of small lung nodules over shorter time intervals could accelerate the time required for the diagnostic work-up of a suspicious nodule detected during lung cancer screening.The utility of this capability is underscored by a recent report demonstrating that protracted diagnostic work-up intervals to evaluate pulmonary nodules even in the range around 5 mm in diameter can decrease the probability of a curative surgical resection. 16is panel also focused on the continuing and rapid advancements of conventional CT technology and how it may impact CT lung cancer screening in opening new opportunities to better detect and manage early lung cancer.This is especially relevant due to the emergence of various artificial intelligence methods used within the CT image formation process and their potential to improve image quality.
A consensus resolution from the QIW XIX Workshop participants is to work with the wider CT imaging community to aggregate informative thoracic CT images, especially in settings where long-term outcomes of screening participants can be determined.One recommendation is to encourage performing CT image quality experiments on the new CT photon counting scanners using collaborative open-imaging research approach to address developmental issues more rapidly.In a similar fashion, a second recommendation is to use large thoracic CT image collections with known clinical outcome to evaluate the performance on new AI tools both to characterize the specific utility as well as potential limitations of these new tools more rapidly in the thoracic CT screening context.
Of particular interest in the thoracic screening setting, both photon counting scanners and AI tools could help in more precisely determining the optimal time-interval to obtain significant follow-up scans to ensure that measured changes in nodule volume indicate the clinical aggressiveness of suspicious lung nodules rather than measurement biases.Finally, the QIW participants affirmed that a transcendent issue is the need to coordinate and integrate these new processes into a fluid clinical workflow stream so that the complex array of healthcare workers engaged in delivering screening services can efficiently accomplish their shared goal.

| Integrating AI into the lung health workflow
One new research opportunity for more precise real-world CT calibration to support inter-operable CT imaging performance when a screening participants' serial thoracic CT scans are acquired on different CT scanners.One proposed solution to this challenge involved using 3D printed models of fully characterized benign and malignant lung nodules.Innovative approaches are required across the broad array of lung cancer screening sites, many of which are ambulatory imaging sites without easy access to sophisticated medical physics support.A scalable and economical option to address this standardization of quantitative imaging is to use additive manufacturing to locally generate reference objects to compare relevant CT scanners regarding image acquisition and post processing.Precise, small homogeneous objects can be fabricated with widely available, inexpensive printers with precise delivery of UV-sensitive resins in the range of 20-50 μ, well below the limits of detection of 100 μ with the best available photon counting CT devices.If such an approach can be validated, inexpensive quantitative imaging phantoms can be available for evaluation across the world enabling the performance evaluation for CT scanners used for lung cancer screening.This approach would complement existing image quality control approaches when discrimination of small volume changes are of clinical relevance such as with annual lung cancer screening management.This approach does not address other critical issues with validation of AI tools for use in disease screening, but providing reference objects to compare quantitative imaging algorithm may result in greater precision in CT imaging follow-up times.This improvement could be a critical step toward real-world precision management of early lung cancer and enhance screening participants' confidence in the process.The session ended with expectations that AI detection and quantitative measurement will continue to improve and be adopted for the management of early lung cancer.
In the final panel, the faculty considered the long-desired goal to leverage AI tools to improve the performance and productivity of radiologists CT lung cancer screening readings.While AI holds promise, it could diminish real-world screening performance unless robust, economical AI methods can be integrated into the lung screening clinical workflow.Thus, it is important that CT lung cancer screening AI system integrates with the complex reality of existing lung cancer screening management systems and hospital Electronic Health Records (EHR).This integration and interoperability will require consensusharmonized and standardized processes to allow quantitative information integration into screening management protocols as well as to be reported in the EHR.Pilot implementation efforts are currently performed at major medical centers or with initiative-taking early research adopters.There is an explicit need to integrate AI processes into the clinical workflows at routine screening sites to ensure broad access to accurate screening imaging detection and management.The major bottleneck that was noted in many presentations across the Workshop is the intractable problems of gaining access for AI developers to large, high-quality thoracic CT image databases which include mature clinical outcome data.

| Breakout sessions-Getting serious about the public health impact of thoracic CT imaging
In the clinical discussions, a major issue emerged regarding how to advance engagement with the public and the medical community about the benefit of lung cancer screening as a public health tool.As the focus on the COVID pandemic is receding, how do we best frame the message that lung cancer screening is an internationally validated life-saving cancer screening service that could save many more lives, especially if this approach is extended to other major co-morbid, tobacco-related conditions detected on thoracic CT including emphysema and coronary calcium.From the Workshop discussions, it was evident that there are two key audiences that impact enthusiasm for lung cancer screening including potential screening participants from the public as well as primary care providers who deliver this service.
Communications approaches for these two different audiences are clearly different.The public is growing more engaged with the potential of lung cancer screening to extend health, but issues of access, cost and compliances are critical.For clinician providers clarity regarding the value proposition are still of concern.There is consistent focus on the burden of screening harms despite favorable emerging information about improved diagnostic work-up efficiency and improved long-term outcomes for critical issues such as low rates of over diagnosis. 17Ongoing communication efforts to both screening participants and health professionals must continue to achieve greater screening participation.Specifically, better ways to communicate the pragmatic importance of AI tools are needed to motivate much greater public image/data donation so equitable AI tool validation can occur.Since access to thoracic CT images is such a challenge, we have worked with lung cancer professionals and advocacy organizations to facilitate access to large numbers of screening images since this can now be done by leveraging cost and accessibility of cloud computing. 18We recently reported that evaluation of this cloud resource for lung cancer screening cases was feasible and that quantitative tools including AI tools could be interrogated in this General Data Protection Regulation-compliant environment. 19The image resource from this cloud environment could also be used to accelerate tool development for research and care for emphysema, coronary calcium as well as lung cancer.
The growing expectation for finding other tobacco-related diseases during lung cancer CT screening increases the public health context for lung cancer screening.Thoracic CT in a population of tobacco-exposed individuals will detect lung and heart disease in addition to lung cancer, so we need a strategy to ensure incentives are in place to develop carefully validated computational tools to responsibly advance the use of this important public health approach.