Is radiology going to be replaced by AI?

Artificial intelligence (AI) is progressively becoming integral to medicine, and AI algorithms are stunningly matching the performance of medical specialists. For this, the question arises is that “is radiology going to be replaced by AI”, and, if that is the case, to what extent and when? The current consensus is that AI will never entirely replace radiology, and it merely enables radiologists to be more effective in their careers (1). Here, we explore the current standing and future role of AI and its algorithms like deep learning (DL) in radiology.

To elucidate the influence of AI on radiology, we need to know what both radiologists and patients assume about this topic. In a recent study in France, for example, a total of 670 radiologists were asked to express their opinions on AI applications in radiology. While having inadequate backgrounds in AI, the radiologists mostly welcomed technically advanced training on AI as a way to enhance their future practice and remarked that AI will diminish medical errors (with imaging) and shorten the interpretation time of each examination (2). In another study on 675 members of the European Society of Radiology (ESR), AI was found to just assist radiologists in better interacting with patients, although it was found to exert a significant effect on breast, thoracic, oncologic, and neuroimaging. Importantly, 55% of the respondents found “AI-only reporting” to be not admitted by the patient and stated that a strong patient-radiologist nexus is highly crucial in real medical settings (3). Most importantly, according to the literature, radiology experts strongly shed light on the concept of patient-centered healthcare and consider AI simply an ally that can improve their performance. However, they know that acquiring AI principles and techniques is a key prerequisite for them if they want to proceed successfully in the future (4).

On the other hand, according to the literature, despite AI making routine tasks be carried out faster and more efficacious, it will never perform as effectively as radiologists, and the tasks fulfilled by radiologists are beyond a simple image processing and interpretation (1), and this is why patients never fully trust on computer-aided diagnosis (CAD). Notably, most of the tasks performed by radiologists are not attainable by CADs, including quality improvement and assurance, educating patients (e.g., to follow advice), radiology procedures (such as preparing patients, performing protocols, working with devices, etc.), patient follow-ups, and so forth (5). However, research supports the “radiologist-AI connection” as a potent strategy to enhance the specificity and sensitivity of diagnoses. For example, the “radiologist + CAD” diagnosis has been reported to enhance sensitivity (from 79.5 to 89.1%) and specificity (from 73.1 to 78.1%) compared to diagnoses merely by the radiologist (6). Thus, radiologists (serving at the edge of the digital era in medicine) are missioned to present AI to healthcare, but they need to know that AI technologies will not cover “respecting patients’ values and preferences”, “judging medical observations”, and “offering solutions tailored to each patient with a special medical condition” (7).

Today, it is incontestable that AI has the strength to alter the landscape of radiology. Thus, although most radiology experts and students ensure that patients will demand “specialty-trained human physicians” in the future, they ascertain that they need to upgrade their knowledge about AI (e.g., DL, machine learning, ANNs, CNNs, etc.) if want to avoid dragging behind future experts with sufficient AI background (8). Likewise, radiologists must consider the advantages of AI technologies (such as the ChatGPT model) in promoting diagnostic accuracy and efficiency, improving radiology workflow, and minimizing interpretation variability (9). Furthermore, AI can heighten many steps in ordinary radiography, including optimizing scan ordering, AI-aided screening of patients to prepare them for radiological exams, optimizing scan protocols, optimizing patients scheduling and positioning, shortening scan acquisition time, and achieving advanced visualization and quantification (10).

According to what is mentioned above, the future of radiology will be strongly impacted by AI. Although patients have currently less knowledge about AI edges or flaws in radiology, future research is expected to elucidate other aspects that will assuredly alter the opinion of patients and even radiologists. However, in reply to the question of “Is radiology going to be replaced by AI”, the answer is indisputably NO, but radiologists with sufficient AI background will assuredly be at the center of radiographic services delivered in the future. Similarly, future patients will perhaps expect their physicians to be well-matched with the latest technologies in this field, as they will demand services that are delivered as rapidly as possible and at higher quality.

References

1. Mazurowski, Maciej A. “Artificial intelligence may cause a significant disruption to the radiology workforce.” Journal of the American College of Radiology 16.8 (2019): 1077-1082.
2. Waymel, Quentin, et al. “Impact of the rise of artificial intelligence in radiology: what do radiologists think?.” Diagnostic and interventional imaging 100.6 (2019): 327-336.
3. European Society of Radiology (ESR) communications@ myesr. org Codari Marina Melazzini Luca Morozov Sergey P. van Kuijk Cornelis C. Sconfienza Luca M. Sardanelli Francesco. “Impact of artificial intelligence on radiology: a EuroAIM survey among members of the European Society of Radiology.” Insights into imaging 10.1 (2019): 105.
4. Kobayashi, Yasuyuki, Maki Ishibashi, and Hitomi Kobayashi. “How will “democratization of artificial intelligence” change the future of radiologists?.” Japanese journal of radiology 37 (2019): 9-14.
5. Nawrocki, Tomer, et al. “Artificial intelligence and radiology: have rumors of the radiologist’s demise been greatly exaggerated?.” (2018): 967-972.
6. Cacciamani, Giovanni E., et al. “Is Artificial Intelligence Replacing Our Radiology Stars? Not Yet!.” European Urology Open Science 48 (2023): 14-16.
7. Pesapane, Filippo, Marina Codari, and Francesco Sardanelli. “Artificial intelligence in medical imaging: threat or opportunity? Radiologists again at the forefront of innovation in medicine.” European radiology experimental 2 (2018): 1-10.
8. Auloge, Pierre, et al. “Interventional radiology and artificial intelligence in radiology: Is it time to enhance the vision of our medical students?.” Insights into Imaging 11 (2020): 1-8.
9. Srivastav, Samriddhi, et al. “ChatGPT in radiology: the advantages and limitations of artificial intelligence for medical imaging diagnosis.” Cureus 15.7 (2023).
10. Pierre, Kevin, et al. “Applications of artificial intelligence in the radiology roundtrip: Process Streamlining, workflow optimization, and beyond.” Seminars in Roentgenology. Vol. 58. No. 2. WB Saunders, 2023.