Visionary Technologies: Artificial Intelligence in Lung Cancer Imaging and Diagnosis

Lung cancer is a scourge of the United States (US) population, causing at least 131,880 yearly deaths and accounting for almost 25% of all cancer mortality (American Cancer Society, 2021). While these statistics are daunting, promise exists in the form of artificial intelligence (AI), which refers to a computer algorithm performing tasks commonly associated with a human level of expertise (West, 2018). AI utilization continues to rise in medical imaging and diagnostics. A particular area of interest in these fields is deep learning (DL), a subset of AI in which an algorithm is able to learn from extensive datasets to make classifications and predictions sometimes exceeding the performance of human experts (Marr, 2018). The use of AI in imaging and diagnostics is crucial for reducing lung cancer mortality in the US. 

Early detection and diagnosis is essential to preventing death from lung cancer. While the one-year survival rate for patients diagnosed with stage I lung cancer is 81-85%, this figure falls to 15-19% for patients diagnosed when this disease has advanced to stage IV. Unfortunately, around 75% of patients have stage III or stage IV lung cancer at the time of diagnosis (Knight et al., 2017). Patients require accurate interpretation of computed tomography (CT) scans, which create detailed images of internal structures, during the early stages of disease to receive life-saving intervention. A DL model trained on a dataset of 42,290 CT cases outperformed six radiologists in diagnosing lung cancer from CT scans with an 11% decrease in false positives and a 5% decrease in false negatives (Ardila et al., 2019). Given that AI can more accurately recognize lung cancer than human diagnostic specialists, it has the potential to detect otherwise undetectable cancers and help patients receive crucial treatment early on in disease progression. 

Figure 1. The 5-year survival rate for non-small cell lung cancer decreases dramatically based on the stage of diagnosis, according to data from the American Cancer Society.

Although CT scans are a common method used to diagnose lung cancer, they are very expensive for both hospitals and uninsured patients in the US. In fact, the average cost of a CT scan for a patient in the US is $867, compared to $97 in Canada and $279 in the Netherlands (Papanicolas et al., 2018). While the cause of these exorbitant prices is still hotly contested, they pose a significant deterrent to patients seeking care. In fact, approximately 21% of American adults reported skipping a recommended medical test or treatment due to the financial burden of healthcare (Kirzinger et al., 2019). AI image analysis algorithms can reduce treatment costs by up to 50% because they are more efficient and require fewer monetary resources in terms of pay, maintenance, and legal protections than the diagnostic radiologists who would otherwise interpret CT scans (Ahuja, 2019). The increased adoption of AI in clinical settings across the US will enable patients to receive the screening they need, ultimately reducing lung cancer mortality.

Figure 2. Data from the International Federation of Health Plans show that the average cost of an abdominal CT scan is significantly higher in the US than comparable industrialized countries.

Significant disparities in health outcomes exist across the US. In rural areas, where upwards of 59 million people live, lung cancer mortality is up to 20% higher than in metropolitan areas (Coughlin et al., 2019). A potential cause of this increased mortality in rural areas is a lack of screening accessibility. Because rural hospitals are twice as likely to report insufficient interventional radiology staffing compared to nonrural hospitals, they tend to lack the specialized staff required to interpret and make diagnoses based on imaging tests such as CT scans, delaying or even preventing essential intervention for lung cancer patients (Friedberg et al., 2019). The US can look to historical precedent to address the limited resources of rural areas. In 1998, the Indian government pioneered the Early Detection and Prevention System (EDPS), which is an AI system that provides recommendations for staff in rural clinics without specialized physicians; importantly, the rate of consistency between EDPS and leading physicians is 94%, demonstrating that such an AI system can be an effective substitute for a doctor’s expertise (Guo and Li, 2018). Adopting this type of system in the US would allow rural patients to receive timely diagnosis and treatment, lowering lung cancer mortality. 

Because lung cancer continues to be a significant cause of mortality in the US, the medical field requires novel technologies to ensure that this disease no longer plagues the population. Deep Learning algorithms provide a favorable alternative to the inefficient, expensive, and inequitable methods currently used to diagnose and treat lung cancer, and their increased adoption across the US could save thousands of lives every year. Despite this, the issue of the implementation of these DL algorithms in hospitals across the US remains a constant source of debate. Future research should focus on the direct applicability of such algorithms in day-to-day clinical settings. 

Edited by Rhea Tumminkatti

References

Ahuja, A. S. (2019). The impact of artificial intelligence in medicine on the future role of the physician. PeerJ, 7(e7702). https://doi.org/10.7717/peerj.7702

Ardila, D., Kiraly, A. P., Bharadwaj, S., Choi, B., Reicher, J. J., Peng, L., Tse, D., Etemadi, M., Ye, W., Corrado, G., Naidich, D. P., & Shetty, S. (2019). End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nature Medicine, 25, 954-961. https://doi.org/10.1038/s41591-019-0447-x

Friedberg, E. B., Corn, D., Prologo, J. D., Fleishon, H., Pyatt, R., Duszak, R., & Cook, P. (2019). Access to Interventional Radiology Services in Small Hospitals and Rural Communities: An ACR Membership Intercommission Survey. J Am Coll Radiol., 16(2), 185-193. https://doi.org/10.1016/j.jacr.2018.10.002

Guo, J., & Li, B. (2018). The Application of Medical Artificial Intelligence Technology in Rural Areas of Developing Countries. Health Equity, 2(1), 174-181. https://doi.org/10.1089/heq.2018.0037

Key Statistics for Lung Cancer. (2021, January 12). American Cancer Society. Retrieved October 9, 2021, from https://www.cancer.org/cancer/lung-cancer/about/key-statistics.html

Kirzinger, A., Muñana, C., Wu, B., & Brodie, M. (2019, June 11). Data Note: Americans’ Challenges with Health Care Costs. Kaiser Family Foundation. Retrieved October 10, 2021, from https://www.kff.org/health-costs/issue-brief/data-note-americans-challenges-health-care-costs/

Knight, S. B., Crosbie, P. A., Balata, H., Chudziak, J., Hussell, T., & Dive, C. (2017). Progress and prospects of early detection in lung cancer. Open Biol, 7(9). https://doi.org/10.1098/rsob.170070

Marr, B. (2018, October 1). What Is Deep Learning AI? A Simple Guide With 8 Practical Examples. Forbes. Retrieved October 9, 2021, from https://www.forbes.com/sites/bernardmarr/2018/10/01/what-is-deep-learning-ai-a-simple-guide-with-8-practical-examples/?sh=792239a8d4ba

Papanicolas, I., Woskie, L. R., & Jha, A. K. (2018). Health Care Spending in the United States and Other High-Income Countries. JAMA, 319(10), 1024-1039. https://doi.org/doi:10.1001/jama.2018.1150

West, D. M. (2018, October 4). What is artificial intelligence? Brookings. Retrieved October 9, 2021, from https://www.brookings.edu/research/what-is-artificial-intelligence/

Image References

Kliff, S. (2015, May 31). Average cost of an abdominal CT scan [Illustration]. Vox. https://www.vox.com/2014/4/17/18076656/health-prices

Rathod, H. (2020, November 10). Non-Small Cell Lung Cancer: 5-Year Survival Rates [Illustration]. VeryWellHealth. https://www.verywellhealth.com/what-is-stage-4-lung-cancer-life-expectancy-2249420

Leave a Reply

Your email address will not be published.