Recognition Of Ovarian Cancer Cells Based On Two-dimensional Light Scattering Method

Ovarian cancer is a common malignancy among women.Due to its covert properties,most patients were in advanced stages when diagnosed with ovarian cancer.Early detection of ovarian cancer is very important for its treatment.At present,early diagnostic techniques for ovarian cancer include detection of serum tumor markers,methods of genomics,and vaginal ultrasonography.However,these early screening techniques are still limited in different aspects.Therefore,more research on early detection techniques of ovarian cancer is necessary.Label-free detection techniques based on light scattering have the advantages of rapidity,accuracy,and non-invasiveness.They have been widely studied in the fields of cell sorting,counting,and analysis of cells' biological and pathological function.However,the research of using light scattering methods,especially two-dimensional(2D)light scattering methods on the analysis of ovarian cancer cell is still rare.In this thesis,we adopt 2D light scattering method for the recognition and classification of ovarian cancer cells.Our work includes the design and establishment of a single-cell 2D light scattering pattern acquisition platform,which based on single-mode fiber illumination.With this platform,a number of 2D light scattering patterns from ovarian cancer cells and normal ovarian cells were collected.In the process of this collected data,we applied Histogram of Oriented Gradients(HOG)algorithm to extract anisotropic features from 2D light scattering patterns and using Support Vector Machine(SVM)to achieve classification.By this method,we achieved an accuracy of 90.81%in the label-free differentiation between ovarian cancer cells and ovarian normal cells.The results show that it is possible to achieve label-free screening of ovarian cancer cells with their 2D light scattering patterns.It also indicates that the specific information of cancer cells and normal cells may be characterized by the gradient anisotropy of their 2D light scattering patterns with HOG algorithm,and this method may also be applied to the label-free detection of other types of cells.The work shows that 2D light scattering technology is expected to be used for clinical label-free screening of ovarian cancer cells.