Researchers endeavor to make thermography a strong supplemental tool to mammography.
Researchers are one step closer to making thermal imaging a more successful, detailed diagnostic support tool for mammography.
While thermal imaging has been around for decades as a non-radiation method for detecting temperature changes in the skin that can alert providers to the presence of breast cancer, it has not been reliable for screening. Now, investigators from The University of Texas at Dallas, along with radiologists from the University of Texas (UT) Southwestern Medical Center, have moved the needle, creating a proof-of-concept computer model that not only reveals breast cancer’s thermal properties, but it also provides more detail about the tumors.
They published their research in Nature Research’s Scientific Reports in June.
“Infrared imaging could potentially provide useful information in a diagnostic setting to radiologists,” said Fatemeh Hassanipour, Ph.D., corresponding author and associate professor of mechanical engineering in the Erik Jonsson School of Engineering and Computer Science. “We want it to be used like a second device for monitoring tumors.”
The overall goal, she said, is to improve digital thermal imaging, making it a better tool for monitoring disease progression and treatment. The hope is that it will be strong supplemental tool for mammography.
To build their model, the team recruited 11 women from the UT Southwestern and Parkland Health & Hospital System in Dallas. The model included a high-resolution infrared camera, clinical data from the volunteers, 3D scanning, and computer-aided design. Using an infrared camera, they imaged the skin to capture temperature fluctuations that were created by the changes breast cancer creates to blood vessels and cell metabolism.
This technique details heat and blood flow patterns close to the breast surface, but it does not offer details about tumor activity deeper in the tissue, they said. Hassanipour’s team wanted to go further. By applying engineering tools to the imaging data, they made a model that could measure the thermal characteristics of breast cancer throughout the breast, revealing the detectable temperature differences between healthy and cancerous tissue. It can also pick up on any increased perfusion rates in affected breasts, they said.
According to their analysis, the team said, modeling results matched surface temperatures from the patient’s infrared imaging reasonably well. In particular, high values were characteristic of late-stage triple negative breast cancer with larger tumor sizes, potentially indicating that aggressive cancers are high heat producers. In addition, patients who had a generalized elevated temperature throughout their breast also had elevated cancer-level blood perfusion rates.
Given these results, this computational modeling does have potential clinical applications, they said. It is possible this model could be used to monitor tumor response to cancer treatment instead of other imaging modalities, as well as track tumor growth over time.
Despite their success, Hassanipour’s team cautioned that their model does have limitations. Not all breast cancer generate enough heat for thermal imaging to be useful, and this model has only been applied to triple negative breast cancer. It cannot effectively be applied to all cancers, they explained. However, the plan is to develop additional models that can provide the same level of detection and identification for other breast cancer patients.
“Our team has many great ideas moving forward,” Hassanipour said. “The subject case that we reported in the manuscript was a proof of concept. A lot of lessons were learned that will facilitate modeling other cancers.”