The use of Artificial Intelligence (AI) is expanding as new applications are being identified. Healthcare has many areas that are already utilizing AI and new use cases are continually being identified and researched. The area of medical imaging is quickly becoming an area of research and adoption for AI tumor detection and monitoring.
In a study published in Diagnostics, “Redefining Radiology: A Review of Artificial Intelligence Integration in Medical Imaging”, Najjar (2022) asserts that “In tumor detection and classification, AI is increasingly being used to discern between benign and malignant lesions and various tumor types, particularly in diagnostics for breast, lung, and prostate cancers”. Najjar also discusses the use of AI in tumor monitoring and its ability to assess changes in tumor size and metabolic activity.
In a similar article found in PLoS ONE journal, “The use of artificial intelligence tools in cancer detection compared to the traditional diagnostic imaging methods: An overview of the systematic reviews”, Silva et al. (2023), performed a study comparing AI technologies used to identify and monitor tumors in cancer patients. Silva et. al postulates that “the help of AI seems feasible and accurate”.
As the use of AI in tumor detection and monitoring continues to expand, there are questions posed about the accuracy of AI tumor detection and monitoring that could affect patient treatment and outcomes. If AI image processing technology is proven to be accurate, it has the potential to provide more consistent diagnoses and better determine treatment efficacy in cancer patients.
Methods to Determine Accuracy
As AI technology methods advance in medical imaging automation, there are a number of disparate AI methods being developed at any given time. This range of technologies are not consistently analyzed and compared with competing methods to determine which have the most accurate tumor identification and monitoring.
In the study PLoS ONE journal, “The use of artificial intelligence tools in cancer detection compared to the traditional diagnostic imaging methods: An overview of the systematic reviews”, Silva et al. (2023) outlines a comprehensive systematic comparison of 382 studies performed using a range of AI methods to identify and monitor tumors in patients. The analysis of 32 of the most rigorously documented studies were performed independently by the authors and then collaborated on their findings to determine the best and most accurate methods. Silva et al. (2023) found that “several AI approaches are promising in terms of specificity, sensitivity, and diagnostic accuracy”.
The article concludes that as AI methods continue to be studied for accuracy, studies should follow the systematic review process defined in the article. This consistent and systematic approach will enable medical imaging to identify the most accurate AI methods as clinical practice adopts the technologies.
AI Uses in Oncology
Today practical applications of AI are being implemented in oncology because of the fundamentally digital nature of oncological imaging. An article published in Diagnostics journal, “Redefining Radiology: A Review of Artificial Intelligence Integration in Medical Imaging”, Najjar (2022) discusses that AI “technologies are being actively explored and adopted in cancer imaging” because of the ability to identify tumors and discern the difference between benign and malignant tumors. The author continues on to address tumor monitoring, saying “AI algorithms provide an objective, consistent means of assessing changes in tumour size or metabolic activity changes”.
The article draws a clear conclusion that the use of AI technology in detection and monitoring of tumors in cancer patients is more accurate and consistent than traditional methods. As confidence in AI technology in tumor imaging continues to grow, the technology will become the imaging reading method used by many practicing oncology radiologists.
Implementation of AI in Radiology
For many practicing radiologists, the leap from manual reading of imaging studies to implementing AI to automatically complete diagnoses is hinged on their confidence of the accuracy of AI methods. As AI imaging results are confirmed and research continues in AI imaging technology, the AI adoption rates by practicing physicians will increase.
In a clinical implementation video, Radiologist vs Artificial Intelligence (AI): Am I worried. youtube.com. Gill, J. (May, 2022), a radiologist who specializes in AI consulting, speaks to radiologists who are considering adopting AI. He directly addresses AI technology accuracy and how AI will change their daily workflow. He states that “by embracing AI and adapting to these changing times we will see our jobs transform and an overall increase in patient care”. Gill is part of a larger conversation between radiologists that will continue to raise the confidence and boost education in the AI technologies.
Conclusion
Proving the accuracy of AI methods in oncology radiology is only half of the challenge facing AI in its aim to become a widely accepted practice. The second half of the challenge is gaining the confidence of the practicing radiologists.
As research continues, confidence of the use of AI in imaging and tumor identification will increase within the imaging and radiology fields. This confidence will expand the acceptance and adoption of AI technologies providing a more consistent and effective tumor identification and monitoring.
References
Gill, J. (May, 2022). Radiologist vs Artificial Intelligence (AI): Am I worried. youtube.com. https://www.youtube.com/watch?v=2trSP-9ezDY
Najjar, R. (2023). Redefining Radiology: A Review of Artificial Intelligence Integration in Medical Imaging. Diagnostics (2075–4418), 13(17), 2760. https://doi.org/10.3390/diagnostics13172760
Silva, H. E. C. da, Santos, G. N. M., Leite, A. F., Mesquita, C. R. M., Figueiredo, P. T. de S., Stefani, C. M., & de Melo, N. S. (2023). The use of artificial intelligence tools in cancer detection compared to the traditional diagnostic imaging methods: An overview of the systematic reviews. PLoS ONE, 18(10), 1–29. https://doi.org/10.1371/journal.pone.0292063
Comments
Post a Comment