Radiology reporting is becoming more specialized. As a large group with more than 30 radiologists, physicians who use our services expect imaging exams to be read by subspecialists who are equipped with the additional knowledge and experience to enhance the quality of both the diagnosis and the radiology report.
At the same time, we believe that artificial intelligence (AI) and deep learning are vital tools that can be used to help make clinical decisions.
For example, AI can expedite a comparative analysis of current plaques and other abnormalities with previous studies. AI can count the number of plaques, measure the size of each plaque and provide an analysis of its growth or reduction much faster than a radiologist. However it is the radiologist who determines the significance of the findings and makes the diagnosis of a patient’s condition.
AI also plays an important role by analyzing the increase or decrease of cancerous tumors and provides radiologists with the data needed to deliver an overall diagnosis that quantifies the progress or regression for each patient.
Improvement in any field requires research and development. AI offers an important tool that aids in text analytics and changes millions of bits of unstructured raw data into meaningful and helpful results and conclusions that can lead decision-makers to success in many endeavors.
AI technology also offers the ability to reduce costs, improve operational efficiency and accelerate productivity, despite an aging population. And governments or insurance agencies can use this data as a tool to drive programs that enhance population health and increase efficacy of diagnosis and treatment.
Dose tracking is another important element in population health. The Veterans Administration is now tracking radiation dose for each individual veteran during his or her lifetime to discover the cumulative effect of radiation and define the limits that should be imposed to avoid overexposure. This addresses the need to measure and identify the value and risk of radiation-based procedures.
Ambitious Strides in Technology
These ambitious strides in technology have advanced the practice of radiology. Deep learning is an enhancement and not a threat. We view this much like the promise that computers would reduce the use of paper. Today we are using computers to generate more paper than ever before.
We continue to depend upon the human brain to interpret the diagnostic content of imaging exams. The difference is that today we have more tools than ever before to measure and evaluate a patient’s condition. We expect these decisions to be made by humans for years to come.
In fact, personalized medicine is the next step forward. Collecting and tracking data from imaging studies and other sources can help physicians make diagnostic and treatment decisions based on each individual’s health history – instead of relying upon a protocol that is applied based on generic statistics such as gender, age, clinical history and other factors.
Personalized medicine can also benefit from gains in the specialization of diagnostic providers. Today we have subspecialties in medicine, but leaders in the industry are looking at a better way to approach diagnosis by focusing on specific anatomy instead of a subspecialty branch. For example, musculoskeletal radiologists or orthopaedic surgeons may specialize in a single joint, such as the wrist, hip, knee, ankle or shoulder.
Experienced radiologists have seen thousands or even hundreds of thousands of images, which equips them to identify abnormalities. In the future, deep learning will be deployed to identify abnormalities for review by the appropriate specialty radiologists. This will be a perfect merger of technology and human decision making. Statistical tools are valuable but they cannot replace a trained specialist.
Ultimately new technology will not only help radiologists and other healthcare providers deliver better care, it will empower patients to make educated choices about decisions related to their health.