Wednesday, October 13, 2010

Be cautious with thermography

The Huffington post recently published an article by a medical doctor on the greatness of thermography as a mechanism for breast cancer detection, you can access it here.

As a breast cancer researcher, I took issue with this article because it was written by a medical doctor and provides women with some highly questionable advice (for instance: "Thermography is a better technology [than mammography]"). Dr. Northrup does not appear to be equipped to determine which technology (mammography or thermography) is better. Dr. Northrup’s main reason for believing that thermograms are the best in breast cancer detection is because of this very old study (30 years old in fact):

M. Gautherie and C. M. Gros, "Breast Thermography and Cancer Risk Prediction," Cancer, vol. 45, no. 1 (January 1, 1980), pp. 51-56.

The main point cited from the paper is that “an abnormal thermogram was 10 times more significant as a future risk indicator for breast cancer than having a history of breast cancer in your family”

This means that thermograms will give you an indication of the likelihood of you developing breast cancer at some point in your life. It does not mean that a thermogram is a good mechanism for detecting a malignant tumour from the image it generates.

Let me inform: back in the 1970s we did not have a good way to reliably detect small (2 to 3 mm) tumours. Small 2 mm tumours far from the surface of the skin are unlikely to cause a detectable difference in skin surface temperatures on a thermogram by virtue of being far from the skin surface which is where the thermogram image is acquired.

Furthermore, a more recent study has shown that “An abnormal thermogram was found in 18.6% of patients with [invasive cancer]”. So thermography only had a sensitivity of 18.6% for invasive cancers – extremely poor!

The authors conclude that “An abnormal thermogram is associated with large tumor size, high grade, and lymph node positivity”. So in other words, thermograms may help us catch tumours that are so large it may be too late to save the patient: high grade, large tumour and lymph node positive. (citation: Sterns et al., Cancer, 1996)

Another study from 1977 shows that thermography has a low sensitivity catching only 39% of tumours compared with 78% for x-ray mammography. (citation: Feig et al., Radiology)

I was surprised to see Dr. Northrup write a whole article about how great thermography is, how much better it is than mammography (which has been tested WAY more extensively) and not even mention any of the newest and highest performing breast cancer detection methods such as MRI and nuclear medicine.

I happen to be a researcher trying to further improve the already impressive breast cancer detection results obtained by using MRI technology. It has been shown that MRI can catch extremely small tumours that are far away from the surface of the patient’s skin. One of my own research papers provides an illustrative example of a small 2-3 mm tumour caught by MRI at Sunnybrook Health Sciences Centre. (the MRI lesion is available: Academic Radiology, Vol. 16(9), 2009). MRI has been compared with mammography, ultrasound and clinical breast examination and been shown to be by far the most sensitive method for detecting breast cancer from high-risk women (an example study demonstrating this can be found here).

Interestingly, I am trying to improve MRI based breast cancer detection with computer-aided diagnosis techniques (having a computer replicate the function of a radiologist medical doctor so that the computer can act as a second reader to improve the screening process). I found this interesting study looking at using computer based pattern recognition techniques to detect cancer from thermograms, however, the researchers were able to “only accurately diagnose about 61.54% of the breast cancer cases” indicating that correctly diagnosing breast cancer from thermograms is very challenging (citation: Ng et al., Journal of Medical Engineering and Technology, 2002).

Readers should be warned about relying on thermography for breast cancer detection, particularly since some studies show that it performs extremely poorly!

Jacob Levman, PhD (Medical Biophysics, University of Toronto)
Sunnybrook Research Institute, Sunnybrook Health Sciences Centre
Imaging Research, Toronto, ON, Canada