ALK positive lung cancer identification and targeted drugs evaluation using microscopic hyperspectral imaging technique

It is important to distinguish ALK positive lung cancer from ALK negative lung cancer and to monitor the efficacy of targeted drugs. This paper applies a microscopic hyperspectral imaging technique to identify ALK positive lung cancer cells and evaluate the therapeutic effect. The hyperspectral images of five groups of lung tissues are captured by the home-made microscopic hyperspectral imaging system. A preprocessing algorithm is proposed to reduce obvious banding noise and noise particles from original data. Then a segmentation algorithm, which assembles Supported Vector Machine (SVM) and Majority analysis together with the Clumping processing, is presented. The ALK positive and negative lung cancers can be distinguished by both the fluctuation of spectral curves and the relative proportion between cytoplasm and cell nucleus. In addition, the treatment efficacy can be surveyed in the same way. The experimental results show that the relative proportion of cytoplasm in Group ALK-neg is 77.3%, while that in Group ALK-pos is 40.6%. It is obvious that there exist differences in contents and spectra between Group ALK-neg and Group ALK-pos. Moreover, the experimental results of quantitative analysis and spectral curves analysis show that ALK positive lung cancer cells treated with the low concentration of targeted drugs will be developed towards the ALK negative lung cancer. This paper shows the potential of using hyperspectral images to detect the ALK positive lung cancer cells and evaluate the therapeutic effect of targeted drugs.