Artificial Intelligence in Radiology in Pakistan; Pros and Cons

KEYWORDS

AI: Artificial Learning; ALS: Amyotrophic Lateral Sclerosis; DL: Deep Learning; PIAIC: President Initiative for Artificial Intelligence and Computing

ABBREVATIONS

Artificial intelligence; Machine learning; Deep learning; Supervised learning; Unsupervised learning; Artificial intelligence in radiology

LETTER-TO-EDITOR

Dear Editor,

Artificial Intelligence (AI) is a computer-based software to analyze data (signs, symptoms, images etc.) with the machines in a way (algorithms) that can mimic the human cognitive functions to optimize problem solving [1]. Ever since the development of the Computer Assisted Detection (CAD), AI has come a long way. Machine learning (ML: linear algorithms) and Deep Learning (DL; Artificial neural connection inspired network) are the two major subsets employed in problem solving. In radiology AI has aided in a few different diagnoses that include more accurate detection of breast cancers (from 75.3%-85.8%) [2], classifying most common posterior fossa brain tumors (wait time from 40 minutes for pathology down to 3 minutes) [3], detection of neurodegenerative diseases (ALS, Alzheimer’s, Parkinson’s) in under 10 seconds, finding out hidden fractures [4] etc. The role of DL has specially been remarkable in Radiology [5]. With the 4th industrial revolution and PIAIC initiative, Pakistan is looking forwards to deploy AI in different fields including the health sector. AI may look promising but is sure to pose challenges [6]. Health economics in a low-income country like us, must first be considered across all the departments that employ radiology, and hence the AI. We may need to find the answer to the question that if machines make decisions independently, then who is to blame for wrong decisions. Also, it raises many questions ethically [7] on using and selling the data with little or no consent. Medical decisions require understanding and empathy to reach a shared goal, the treatment plan. Weighing the pros and cons, AI should be used more as a supplemental rather than replacement tool.

AUTHORS CONTRIBUTION

All authors contributed for the manuscript.

REFERENCES

1. Miller DD, Brown EW (2018) Artificial intelligence in medical practice: The question to the answer? Am J Med 131(2): 129-133.
2. Kim HE, Kim HH, Han BK, Kim KH, Han K, et al. (2020) Changes in cancer detection and false-positive recall in mammography using artificial intelligence: A retrospective, multireader study. Lancet Public Health 2(3): E138-E148.
3. Hollon TC, Pandian B, Adapa AR, Urias E, Save AV, et al. (2020) Near real-time intraoperative brain tumor diagnosis using stimulated Raman histology and deep neural networks. Nat Med 26(1): 52-58.
4. Voelker R (2018) Diagnosing fractures with AI. JAMA 320(1): 23.
5. Hosny A, Parmar C, Quackenbush J, Schwartz LH, Aerts HJWL (2018) Artificial intelligence in radiology. Nat Rev Cancer 18(8): 500-510.
6. Liu X, Faes L, Kale AU, Wagner SK, Fu DJ, et al. (2019) A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: A systemic review and metaanalysis. Lancet Public Health 1(6): E271-E297.
7. Murphy K, Di Ruggiero E, Upshur R, Willison DJ, Malhotra N, et al. (2021) Artificial intelligence for good health: a scoping review of the ethics literature. BMC Med Ethics 22(1): 14.