Cancer Breakthroughs, a new review session at ASTRO’s 61st annual meeting held Sept. 15-18 in Chicago, highlighted important breakthroughs and practice-changing trials in radiation oncology from the year’s largest US cancer research meetings.
Kristy K. Brock, PhD, (pictured left) a professor in the Department of Imaging Physics at the University of Texas MD Anderson Cancer Center in Houston, discussed two technological innovations presented at the 2019 annual meeting of the American Association of Physicists in Medicine (AAPM): the first of two whole-body positron emission technology (PET) scanners developed by the EXPLORER Consortium, and exciting research in the use of breast magnetic resonance imaging (MRI) radiomics to distinguish between malignant and benign tumors.
uExplorer: The first ultra-high-resolution total-body scanner FDA-cleared for clinical PET imaging
The development of the first ultra-high-resolution total body PET scanner has the potential to make an incredible impact on diagnostic imaging. University of California-Davis researchers, led by Simon R. Cherry, PhD, professor of biomedical engineering, and Ramsey Badawi, PhD, professor of biomedical engineering and chief of nuclear medicine, designed a PET scanner with a total axial field of view (FOV) of 195 cm using existing technology from state-of-the-art PET/CT scanners. The scanner can acquire a three-dimensional (3D) image of the human body in a single position, unlike current generation PET/CT scanners with axial FOVs of 15 to 30 cm.
How this technology will revolutionize the clinical use of PET is as unknown today as when the first PET modality was introduced in 1977. In her presentation, Dr. Brock described the system and the first human images acquired, and raised many questions for ASTRO attendees to consider as this technology proliferates in research and clinical use.
The uExplorer PET/CT scanner, commercialized by United Imaging Healthcare of Houston and which received FDA 510(k) clearance in January 2019, can produce a whole-body diagnostic scan in as little as 20 to 30 seconds compared to 10 to 20 minutes with a conventional PET scanner. It can also scan up to 40 times faster or use up to 40 times less radiation dose. A 40-fold increase in dynamic range enables the imaging of radiotracers by more than five half lives, enabling repeat imaging of a patient without the need for additional injection of radiotracers.1,2
The total body PET allows maximal detection of the radiation from injected radiotracers emitted from the body. The sensitivity of the scanner can be increased by a factor of about 40 for total body imaging or a factor of about 4 to 5 for imaging a single organ by increasing the geometric coverage to include the entire body. This sensitivity can be used to increase the signal-to-noise ratio (SNR) in reconstructed images, supporting reconstructed images at higher spatial resolution, and potentially allowing the detection of smaller or lower contrast lesions.
“This technology will allow considerable power of the tracer kinetic principle to be applied," said Dr. Brock. "The increased sensitivity and dynamic range will allow imaging at high SNR at much later times after tracer injection. As tumor contrast typically increases with time as tracers clear from other tissues, later imaging may reveal a different picture about the extent of disease and allow smaller or less tracer-avid lesions to be seen.”
Dr. Brock posed several questions to encourage discussion on harnessing the potential of this new system: “How will the multi-organ nature of whole-body PET change the paradigm of imaging to detect metastatic disease? What additional tracer development is needed? How will the ability to obtain kinetic information aid in the development, optimization, and characterization of new drug development? What additional research needs to be performed to improve detector efficiency and timing resolutions? What opportunities exist for whole-body PET/MR or eliminating the need for CT?”
Detecting malignant breast cancer lesions with multiparametric MRI
Multiparametric MRI helps improve the diagnostic accuracy of breast cancer. Computer-aided (CADx) radiomics, the rapid extraction of data from diagnostic images, combined with machine learning and artificial intelligence (AI), are transforming image interpretation into a quantitative process.
Researchers throughout the world are working to develop predictive models from quantitative image data that will help personalize patient care. Dr. Brock said that at the University of Chicago’s Giger Lab, researchers are working to develop multiparametric breast MRI radiomics methods that integrate information from dynamic contrast-enhanced (DCE) perfusion, T2-weighted (T2W), and diffusion-weighted imaging (DWI) sequences and compare them with single-parametric methods that use only one sequence in the task of distinguishing benign from malignant breast lesions. Sequences such as T2W and DWI provide information that complements dynamic sequences that are proven to show high sensitivity but limited specificity. The researchers’ objective is to create predictive models for treatment planning and assess breast cancer risk recurrence.
The current benchmark in breast MRI CADx uses DCE alone, explained Dr. Brock. Isabelle Qiyuan Hu, a medical physics graduate student, and colleagues, presented research relating to the algorithmic model they are developing to identify malignant breast lesions using data from a University of Chicago database of 1,011 breast MRI exams.
Data extracted included seven geometric features, 43 DCE features, 19 T2W features, and six DWI features. The researchers utilized two radiomics techniques, combining components for feature and classifier fusion, and aggregating the output from four single parametric classifiers.
“Each of the two multiparametric MRI radiomics methods—feature fusion and classifier fusion—significantly outperformed all four single-parametric classifiers in distinguishing between benign and malignant breast lesions,” said Dr. Brock. “While a significant amount of future work is needed, these findings have very significant potential impact. They demonstrate the powerful potential of computational analysis to harness additional data within diagnostic images that we cannot see to create a more powerful prediction."
“Machine learning and AI provide a powerful set of tools to address the increasingly complex decision-making process in cancer treatment," she added. "But how do we move from using machine learning and AI to be ‘descriptive’ and ‘diagnostic,’ determining what is happening to our patients and why, to being ‘predictive’ and ‘descriptive,’ determining what is likely to happen and what we need to do?”
Dr. Brock then asked session attendees to consider how these types of results can translate into clinical practice, how to collect and manage data to ensure that robust tools are developed that translate between clinics, how different specialties and organizations should work together to achieve this goal, and how to develop sound guidelines and clear validation.
Editor’s note: Part 1, ASTRO '19: Cancer Breakthroughs session showcases top research from leading cancer meetings, can be found at https://www.appliedradiationoncology.com/articles/astro-19-cancer-breakthroughs-session-showcases-top-research-from-leading-cancer-meetings-part-1.Back To Top
ASTRO ’19: Cancer Breakthroughs session discusses total body PET, breast MRI radiomics (Part II). Appl Rad Oncol.