Software solutions have long been an important consideration for purchasers of medical imaging equipment. The initial focus has typically been on software supporting storage and communication of digital images through ‘digital imaging and communications in medicine’ (DICOM) capabilities and ‘picture archiving and communication systems’ (PACS). Widespread adoption of electronic medical records (EMRs) has helped to drive recent developments in software, which have expanded interest in artificial intelligence, image reconstruction algorithms, and computer-aided detection and diagnosis. Ultimately, these systems have the same aim: to recognise clinical problems and abnormalities in digital images to assist diagnosis and improve results.
Particular attention is being placed on how these software developments may increase the accuracy of medical imaging, improve patient outcomes, and steer healthcare professionals towards enhanced treatment of patients. Medical imaging professionals are hoping for algorithms to recognise and detect signs of disease; artificial intelligence programmes to provide an additional opinion to support healthcare professionals in their decision making; and for intelligence systems that will enable comparison of new medical images with existing ones from similar cases.
The benefits of such software developments are still being determined; however, it is largely accepted that, when used correctly, such software can enhance a radiologist’s ability to identify and appropriately diagnose abnormalities in medical images, therefore increasing patient confidence in the result and reducing mistakes. The future target for software systems, such as artificial intelligence, is to become a tool used to make healthcare professional more efficient, to provide a clearer view of the case, and provide more robust information for diagnosis and referral.
Because of the data-focused nature of medical imaging, radiology in particular, it is well-suited to the opportunities presented through artificial intelligence and associated analytics. Such opportunities could lead to transformations in value-based care, radiology workflows, and patient outcomes. The healthcare environment is preparing for shifts towards value-based practice, incentivizing healthcare providers to avoid unnecessary procedures, and encouraging collaborative workflows and interoperability. Developments in software tools, such as artificial intelligence and algorithms, will assist in the route to true value-based healthcare.
Despite the perceived benefits of these software tools, there are considerable challenges to overcome before adoption will spread. Initial challenges will be from regulatory barriers, such as gaining approval from the Food and Drug Administration (FDA) for the use of artificial intelligence tools. There will also be practical challenges such as ensuring sufficient training of staff in the use of these technologies. Perhaps the most difficult challenge to overcome comes from within the industry itself, with concern that radiologists’ jobs will be replaced by new technologies. There is also worry surrounding accuracy and mis-diagnosis through using these tools, which can lead to negative perceptions of patients. In reality, these systems will be used to complement the healthcare professional, by advising and by supporting decisions. Medical imaging experts will continue to be required to make final diagnostic decisions and override the intelligence tools if necessary.
In addition to overcoming the regulatory and cultural challenges, further progress will be required to create more advanced algorithms, improving their capability of not only identifying objects within medical images, but also recognising why such an abnormality may occur. These developments are required to ensure that these tools are useful in clinical practice.
The impact of artificial intelligence and diagnostic algorithms on the medical imaging industry has been limited to date. More widespread use of these technologies is expected over the next 5–10 years, resulting in further developments as associated challenges and teething issues are ironed-out. The vast potential of these technologies including reducing diagnostic errors, improving patient care, and enhancing radiology workflows will help to drive adoption forward. Over time, raised awareness of the benefits of these tools on the healthcare sector will further encourage uptake.
Ultimately, artificial intelligence tools will not substitute the radiologists’ brainpower, but will instead enhance it as part of a symbiotic collaboration been human and artificial intelligence.