A micromachined millimeter-wave medical sensor for skin-cancer diagnosis
Fritzi Töpfer1 , Sergey Dudorov1 , Dragos Dancila2 , Robin Augustine2 , Xin Hu2 , Lennart Emtestam3 , Kari Gustafsson4 , Lars Tenerz5 , Anders Rydberg2 and Joachim Oberhammer1 1 KTH Royal Institute of Technology, 2 Uppsala University, 3 Karolinska Institutet, 4 ADAKT AB, 5 OPTIGA AB
1. Medical motivation
Skin cancer is the most common cancer in the age group of 15-45 years. Incidents are increasing by an average of 3-6% per year since the 1960s, and melanoma are among the most deadliest cancer types with 15-20% mortality. In Sweden, there are about 2500 melanoma and 1200 precursors diagnosed per year, resulting in almost 500 deaths in 2011. A huge screening effort is needed to find skin cancer: about 50-250 screenings have to be done for finding a single melanoma. There is no established sensor technology available, diagnosis is so far been done by highly-trained dermatologists using dermatoscopes, and after extraction verified by histological analysis. In Sweden alone, about 100 000 preventive skin surgeries are performed every year, most of which are unnecessary and could be avoided if a diagnosis tool with high sensitivity/specificity would be available. Specialized dermatology departments are increasingly becoming a bottleneck.
2. Method: a drastically miniaturized, 100-GHz millimeter-wave sensor for skin-cancer diagnosis
Electromagnetic waves at microwave frequencies are known to be able to distinguish healthy from tumour cells, as energy absorption above 1 GHz is significantly higher due to increased free and bound water content of fast and uncontrolled growing tissue . The discrimination of healthy/cancer tissue by microwaves has also been shown for skin tumours, incl. BCC  and melanoma . Macroscopic microwave sensors, available up to 40 GHz, have a too large interaction volume and are therefore not suitable for analysing shallow and inhomogeneous skin cancer tissue at an early stage. This paper presents the results of a micromachined millimeter-wave probe operating at 100 GHz, which has a probe tip with only 0.9% of the contact area of a conventional microwave probe and with a penetration depth matched to shallow skin cancer tissue layers, and uniquely combines high sensitivity and high resolution. Two different probe designs, a broadband design by KTH and a resonance concept by UU, have been implemented by micromachining and characterized.
3. Results: technical verification, first measurements on skin
Probe prototype 1 (KTH design) has successfully been technically verified, resulting in a responsivity of S11/εr=0.47dB in the permittivity range of cancer/healthy tissue at 100GHz, a reproducibility better than 1.4% (1σ), a long-term stability better than 0.6% (1σ) for 8h, and a lateral scanning resolution of about 100μm for a 600x300μm2 probe tip. First tests on human skin revealed that the probe can clearly distinguish skin tissue on different positions of the body (back of hand, palm, arm), can detect skin burns and can monitor the healing process of skin burns (discrimination between burnt skin, newly grown skin, normal skin), and can distinguish benign skin neoplasms from normal skin and subdermal (junctional) nevi. Probe prototype 2 (UU design) is a resonant probe design which has been tested on phantom materials composed of agar, polyethylene powder and TX-15. The probe can successfully distinguish 50%, 75%, and 100% phantom materials.
4. Conclusions, outlook, acknowledgments
We presented a new type of microwave wave medical sensor, enabled by micromachining. The probes were optimized for skin cancer, and provide a unique responsivity/resolution performance as compared to conventional microwave probes. The sensor prototypes have been successfully technically characterized and measurements on different skin types and skin anomalies have clearly verified the sensor function. Currently, an application for ethical approval is being filed for continuing the work towards evaluating cancer tissue. The VINNOVA Framtidens Hälsa programme is acknowledged for funding.
 Mariya Lazebnik et al., Phys. Med. Biol. 52 (2007) 6093–6115.
 R. M. Woodwardet al. , J. Invest. Dermatol., vol. 120, pp. 72-78, 2003.
 A. Ito et al., Cancer Sci., vol 94, no. 3, pp. 308-313, Mar. 2003.
Figure 1: Mmicromachined microwave skin-cancer sensor prototypes: (a) broadband sensor, (b) resonant type sensor; (c) responsivity analysis; (d) measurements on skin, discrimination of burnt/new-grown/normal skin; (e) measurements on skin, benign skin neoplasm/subdermal mole/normal skin.
The clinical literature is refer to "A micromachined millimeter-wave medical sensor for skin-cancer diagnosis" by KTH Royal Institute of Technology, 2 Uppsala University, 3 Karolinska Institutet, 4 ADAKT AB, 5 OPTIGA AB