Early diagnosis of breast cancer using ftir and chemometry technique

For cancer patients, early detection of the relevant cancer symptoms is critical to increase survival rates. It has been observed that postmenopausal women which correspond age of 40-55 are most affected by breast cancer in the world. Breast cancer-based deaths among women are estimated accounted for 14% of total cancer-related deaths [1,2]. Early diagnosis of breast cancer is not cheap, rapid or easy. If it is possible to reach all these features in one single case, researches will have come a long way to treatment. Nowadays, there are a few methods to detect a breast cancer which are MRI (magnetic resonance imaging), mammography, ultrasonography and histopathological examination. It was very likely that these methods would give some erroneous measurements. MRI is an expensive tool and has low precision to express clearly differences between benign and malignant tumor types. Additionaly, even it was the best cancer detection method, it has a critical risk for the advanced stage breast cancer patients because of high radiation area. Ultrasonography and mammograpy have low accuracy in comparison to MRI. On the other hand, histopathological examination can be considered the most using way to recognize the breast cancer. However, this method has some negativeness such as personal mistakes and time consuming procedures. IR spectroscopy, which has many applications in the industry for environment- or production-based applications, offers a number of advantageous solutions in many ways. In the past, this technique has been used in a wide range of applications, from the combination of gas or liquid mixtures, to the purity of gases or to the analysis of trace components for environmental analysis. The devices used can be increased in variety, from simple filter-based photometers to complex opto-mechanical devices such as Fourier Transform Infrared (FTIR) devices. In complex measurement situations, larger quantities of spectral information are required and either full-spectrum or multi-wavelength analyzers are used. Among the analytical techniques available for process analytical measurements, IR provides a versatile measurement technique where all physical forms of a sample - gas, liquid, solids, and even mixed phases - are present. In recent years, many new technologies have emerged as a result of telecommunication developments in the late 1990s. While not directly applicable at this time, many of these technologies will offer a way to improve the performance of IR measurement devices and help in the development of smaller, lower cost instruments in the future [3]. FTIR has been used as a tool in many investigations. This technique was first used to classified breast cancer tissue in 1996. Normal and cancer tissues are discriminated with phospodiester groups amide I and amide II bands which are increased in cancer tissues[4]. H. Fabian et al. were used IR imaging system with functional group mapping techniques and cluster analysis to discriminate benign and malignant tumor tissues [5]. FTIR-ATR imaging system which is a practical tool to identify chemical signatures of fluid or solid unknown structures was detected distinguish between nuclear grades [6]. FTIR was showed that second derivative spectra of breast tissues differ by specific peaks [7]. Samples which were formalin-fixation applied and paraffin-embedded (FFPE process) were used widely to analyze. FFPE process were showed small but significant changes in cancer cells after IR imaging [8]. FTIR were succesfully used with recevier operating characteristic curves, second derivative spectrums and deconvolution of protein amide I peak to distinguish normal, malignant and benign tissues [9]. Taking account all of these, IR spectroscopy can be demonstrated as the newest and accuracy tool to determine a cancer and also its grades by analyzing biochemical structures of tissue sections. In some studies, different methods have been used in history in order to interpret the data more efficiently. Chemometrics, or chemometric, is the use of mathematical and statistical methods to improve understanding of chemical information and to relate quality parameters or physical properties to analytical instrument data. The results of the studies are modeled and these models can be routinely applied to future data to estimate parameters with the same characteristics. The result of the chemometry approach gains efficiency in evaluating product quality. It can be used to develop more efficient laboratory applications or automated quality control systems. The only requirement is an appropriate tool and software to interpret the patterns in the data [13]. Chemometric method results for the analysis of spectrum data provide many effective ways to calibrate the data. The three general headings to be noted in the use of the chemometric methods are as follows: Powerful calibration, sample selection diagnostics, variable selection and statistical calculation to create reliable models; correct application of spectroscopic data preprocessing to reduce and correct parasites such as overlapping bands, baseline deviations, scattering and path length variation; model validation and integration, rigorous estimation, quality control measurement, real-time product quality and process monitoring [13,15]. The aim of this study was not only to discriminate normal and tumor tissues but also nuclear grades. 17 breast tissue samples were used including 14 normal and 17 tumor tissues. It is stated what could be the spectrum expressing making the pathology in the crude breast tissues which tumor and healthy tissues were not microscopically analyzed. It has been determined that the breast tissue showing cancer formation produces three types of spectrum depending on the fat content: strong, weak fat content and no fat content. Additionaly, there were too many paramaters but more suitable indicates were chosen to analyze which were histological grades (grade I, II and III). Samples were obtained from Acıbadem Medical School Pathology Laboratory (Istanbul/Turkey). Postoperative samples were taken to the pathology laboratory and acquired macroscopic sections (1cm thickness, cubic shape). The sections were immediately injected with formaldehyde and were not allowed to dry. The specimens were analyzed by FTIR-ATR spectroscopy and studied chemometrically. Chemical fingerprints of samples were determined using a classical FTIR spectrometer (Perkin Elmer Spectrum 100) in the range of 650-4000 cm-1 at spectral resolution of 4 cm-1 with 16 co-scan. A total of 211 normal and pathological samples taken from 17 patients were evaluated by taking measurements from 106 surfaces. The samples were evaluated according to their fat content after macroscopic FTIR-ATR measurements and at first stage the necessary tissue spectra were determined. The spectral positions of the chemical compositions of the tissues have been compared with the literature. Spectral data were treated with partial least squares (PLS) technique to reduce the size of the data. Spectral information related to the chemical nature of DNA&RNA and proteins were treated with canonical variate analysis (CVA) and soft independent modelling of class analogies (SIMCA) to predict tumor tissues on a 2D variate plane and determine the class (healthy or not) of an unknown sample. Clinical histopathologic results were compared with established models. As a result of the studies, it has been observed that normal and pathological tissue specimens can be distinguished from each other very well, especially in the fingerprint region, and similar results are obtained when grouping of the unknown chemical information is done. The chemometric methods used have been shown to achieve great success in this context and as the most important difference from previous articles, tumor tissues were attempted to classify by carsinoma stages with chemometric examination.