How Artificial Intelligence is Changing Radiology, Pathology

Images obtained by MRI machines, CT scanners, and x-rays, as well as biopsy samples, allow clinicians to see the inner workings of the human body. However, these images often contain large amounts of complex data that can be difficult and time consuming for human providers to evaluate.  AI tools can augment the workflow of radiologists and pathologists, acting as clinical decision support and enhancing care delivery.“There are so many moving pieces when it comes to artificial intelligence,” Keith Dreyer, DO, PhD, Chief Data Science Officer and Corporate Director of Enterprise Medical Imaging at Partners Healthcare,

“Imaging analytics is a good example of the complexity involved in all different disciplines, as well as the pace of progress that we’re seeing in AI development at large.”



What are the top ways AI will enhance radiology and pathology and lead to better patient care?


AI and machine learning have demonstrated great potential in supplementing and verifying the work of clinicians, particularly in the complex field of imaging analytics.

Pathologists must meticulously evaluate medical images to diagnose patients, sometimes examining hundreds of tissue slides for traces of abnormalities.

Machine learning and deep learning algorithms offer the opportunity to streamline pathologists’ decision-making, allowing them to review detailed data with improved accuracy and fewer errors.

“If the network can tell which patients have cancer and which do not, this technology can serve as triage for the pathologist, freeing their time to concentrate on the cancer patients,” said Anant Madabushi, a biomedical engineering professor at Case Western Reserve and co-author of the study.

While the researchers acknowledged that the tool showed great promise in supporting the work of pathologists, the team also noted that AI won’t be replacing clinicians anytime soon.

“The network was really good at identifying the cancers, but it will take time to get up to 20 years of practice and training of a pathologist to identify complex cases and mimics, such as adenosis,” Madabushi said.

The belief that AI can improve treatment decisions and patient care extends beyond the realm of research.  

The FDA recently cleared an AI algorithm that can detect distal radius fractures and provide clinical decision support at the point of care.



“Artificial intelligence algorithms have tremendous potential to help health care providers diagnose and treat medical conditions,” said Robert Ochs, PhD, acting deputy director for radiological health, Office of In Vitro Diagnostics and Radiological Health in the FDA’s Center for Devices and Radiological Health.

“This software can help providers detect wrist fractures more quickly and aid in the diagnosis of fractures.”

AI can provide further clinical decision support by improving diagnostic processes and enabling providers to identify diseases with greater accuracy.

In pathology, many diagnostic processes currently rely on physical tissue samples obtained through biopsies. AI can enhance radiology tools, making them accurate and detailed enough to replace physical samples.

Researchers at Colorado State University are using machine learning to develop a virtual biopsy tool that will make early detection of melanoma faster and cheaper.

“The mantra for melanoma has always been, 'when in doubt, cut it out,'” said Jesse Wilson, assistant professor in the Department of Electrical and Computer Engineering and in the School of Biomedical Engineering at Colorado State.

But removing skin lesions is an invasive and intensive process, and even more so for patients with numerous suspicious moles.

The team at Colorado State will work to simplify and virtualize melanoma detection, allowing dermatologists to treat the disease earlier and faster.

The National Institutes of Health (NIH) is also seeking to improve lesion detection with AI.

The organization recently released a dataset of more than 32,000 medical images, large enough for scientists to train a deep learning neural network and create a large-scale lesion detector with one unified framework.

Researchers will be able to identify an individual’s lesions more accurately and allow them to quickly evaluate the whole body for cancer risk.

In the future, it may also be possible to extend this lesion detector into other imaging modalities, including MRIs.


Using AI algorithms to extract meaning from medical images will also allow radiology and pathology to make significant contributions to precision medicine.

In 2016, researchers at Stanford University School of Medicine found that a machine learning algorithm was able to accurately differentiate between two types of lung cancers.

Distinguishing between cancer types can be a challenging task for pathologists and can lead to wide variations in the identification of the stage and grade of the cancer.

“Pathology as it is practiced now is very subjective,” said Michael Snyder, PhD, Professor and Chair of Genetics at Stanford University.

“Two highly skilled pathologists assessing the same slide will agree only about 60 percent of the time. This approach replaces the subjectivity with sophisticated, quantitative measurements that we feel are likely to improve patient outcomes.”

The machine learning tool was able to identify many more cancer-specific characteristics than can be observed by clinicians, offering the possibility of more personalized treatments and therapies.

Mount Sinai Health System is also interested in advancing precision medicine with imaging analytics.

The organization created an imaging research warehouse to grant researchers access to imaging and clinical data from more than 1 million patients.

Researchers developing AI and machine learning algorithms can use the warehouse data to build innovative tools and discover new ways to treat disease.


AI has also shown promise in improving predictive analytics for radiology and pathology.

The technology can help clinicians identify the onset of disease in patients earlier, allowing them to plan for long-term care needs.

Additionally, this information can help providers improve clinical trial enrollment..   

Machine learning has demonstrated its ability to accurately flag patients who are progressing into Alzheimer’s disease, for example, allowing researchers to capture individuals who may be eligible for trials around drugs that slow neurodegeneration.

Researchers from the Alzheimer's Disease Neuroimaging Initiative developed an algorithm that used advanced imaging analytics to identify individuals on the verge of developing dementia with 84 percent accuracy.

“With its high accuracy, this algorithm has immediate applications for population enrichment in clinical trials designed to test disease-modifying therapies aiming to mitigate the progression to Alzheimer's disease dementia,” the team stated.

Machine learning has also proven to be adept at predicting how long a kidney will function adequately in patients with chronic kidney damage.

A research team at Boston University used renal biopsies to train deep learning and neural networks to predict kidney function. Chronic kidney disease frequently shows few symptoms until it is very advanced, creating the need for accurate identification for how the disease is progressing.

“Chronic kidney damage is routinely assessed semi-quantitatively by scoring the amount of fibrosis and tubular atrophy in a renal biopsy sample,” the team said.

“Although the trained eyes of expert pathologists are able to gauge the severity of disease and to detect nuances of histopathology with remarkable accuracy, such expertise is not available in all locations, especially at a global level.”

AI has the potential to serve as an aid to radiologists and pathologists tasked with making informed clinical decisions and choosing effective treatments.

With AI and machine learning, the field of imaging analytics can allow clinicians to improve care delivery and patient outcomes.