AI Insights: Personal Touch
By Rose Higgins and Mimi Huizinga, MD, MPH, FACP
Radiology Today
Vol. 22 No. 7 P. 8

Radiomics improves the cancer patient's journey.

The science of genomics changed cancer research and treatment by giving clinicians the ability to understand how specific genes predicted the likelihood of acquiring cancer or the success of a particular therapy.

Radiomics represents the next great advance in oncology research and treatment. Radiomics is a relatively novel technology that helps life sciences and health care organizations unlock hundreds of previously unobtainable data points from images, such as MRIs and PET scans, by using AI-powered algorithms to extract information about the biology of tumors and lesions.

Health care organizations can compare radiomic data with similar data from past images, as well as the biology of healthy organs, to gain a deeper understanding of tumor or lesion response to a specific therapy, enabling better decision making about care and treatment along every step of a cancer patient’s journey. Following is a deeper explanation of how.

Phase 1: Early Identification and Diagnosis
Traditional images present radiologists with a point-in-time view of lesions and tumors. For example, some tumors are classified by radiologists as “something to watch,” but that statement is hardly definitive without more data to support it. A second image at a later time will offer more information, such as whether the tumor has grown along the long and/or short axes, but still does not deliver definitive guidance about whether clinicians should take any action—and if so, what action. As this process unfolds, valuable time for the patient is lost.

In contrast, by adding radiomics to the picture, clinicians and researchers realize new quantitative insights into the tumor’s biology that traditional images lack. Specifically, they obtain information about volume, texture, and many other data points in a low-cost and noninvasive manner. By combining radiomic data with clinical and genomic data, oncologists acquire new insights about how to better personalize which course of treatment will create the best outcomes for individual patients.

A real-world example of radiomics’ predictive capabilities comes from research by the Moffitt Cancer Center in Tampa, Florida. Researchers discovered, when screening patients with low-dose computer topology, two biomarkers for lung cancer that can help stratify patients for five-year survival risk: Statistical Root Mean Square and Neighborhood Grey Tone Difference Matrix Busyness.

With these two biomarkers, researchers found they could accurately identify patients as high risk (0% five-year overall survival) or low risk (78% five-year overall survival). Using new insights from these predictive data, physicians may elect to recommend more “aggressive follow-up and/or adjuvant therapy to mitigate their poor outcomes,” per the Moffitt researchers. 

Phase 2: Treatment
During the treatment phase of a patient’s cancer journey, radiomics offers value by extracting information about tumor biology to define a novel set of quantifiable patterns. As in the Moffitt example, these imaging biomarkers deliver clinical signatures that are unique to each patient, helping clinicians design personalized treatment plans.

With traditional approaches, patients often must enter a “wait-and-see” period shortly after commencing treatment. During this period, the limitations associated with standard images may prevent clinicians from gaining adequate detail into whether the treatment is producing the desired results, prompting clinicians to order another biopsy. The downside of this approach is that it may increase risk to the patient, particularly for cancers such as glioblastoma that can settle deep into vital organs, creating hesitance on the part of clinicians in ordering biopsies, thereby leaving them in the dark about the treatment’s effectiveness.

However, by adding radiomics to the process, oncologists gain similar insights as they would from a biopsy but via a significantly less invasive method that still enables them to track progress of the treatment. Armed with more detailed, precise data, clinicians can better determine whether a treatment is leading to the expected response while reducing risk to the patient and saving the costs of performing a biopsy procedure. If the treatment is not producing the desired results, clinicians can examine radiomic data about tumor biology to find which key parameters, such as texture and heterogeneity, are not trending in a positive direction. They then can adjust the course of treatment based on the growing body of knowledge about cancer treatments.

In some cases, for example, oncologists may decide to discontinue chemotherapy if radiomic data indicate that the treatment is not producing the intended effect—saving patients the stress, emotional burden, and expense of chemotherapy. Alternatively, oncologists may elect to pursue a different course of therapy based on a review of the literature, in addition to radiomic data that reveal insights into the tumor’s reaction to prior treatment.

In this regard, radiomics can deliver significant value for life sciences organizations as they develop, monitor, and adjust clinical trials, particularly as the medical community identifies more imaging biomarkers that yield new information. Life sciences organizations can leverage radiomic data in the initial stages of clinical trials to improve how they identify patients who are more likely to benefit from a particular therapy and should therefore be included in the trial. Similarly, advanced imaging analytics data can be used by life sciences companies to identify patients who should be excluded from trials based on commonalities between their tumors and known outcomes associated with the same biomarkers.

Also, during the clinical trial process, radiomics can help researchers identify previously undiscoverable changes to tumor biology that indicate more precisely the efficacy of a course a treatment while eliminating the need for one or more biopsies. Accordingly, researchers gain the ability to reach faster conclusions about whether clinical trials themselves should continue, as well as whether individual participants should continue to be included in those trials. In all of these ways, radiomic data provide clinicians the ability to monitor treatment response at a deeper level and make course corrections more quickly than they otherwise would, improving their decision making and the likelihood of desired outcomes.           

Phase 3: Survivorship
During the survivor phase of the cancer patient’s journey, radiomics again has an important role to play, as survivors undergo routine medical scans that generate the images oncologists depend on to monitor recovery. By revealing tumor biology and comparing it with prior images, radiomics ensures that key facts are examined and oncologists don’t miss key clues to tumor response. 

For example, if radiomic data indicate no changes since the prior image, survivors can be reassured with much greater confidence and precision than is associated with the traditional reading of an image. Alternatively, if key radiomic parameters indicate a developing issue, oncologists can be informed of it more quickly to begin to address it earlier than they otherwise would have.

A New Weapon
Former President Richard Nixon famously declared the nation’s “War on Cancer” when his administration established the National Cancer Institute in 1971. As the war continues 50 years later, radiomics offers clinicians an important new weapon and cancer patients new hope.

— Rose Higgins is the CEO of HealthMyne.

— Mimi Huizinga, MD, MPH, FACP, serves on the HealthMyne board of directors. She is the vice president and head of US Oncology Medical for Novartis.