Improving Classification of Head and Neck Squamous Cell Carcinoma using Biomarker Based Features
msra
摘要
We present a biomarker based computer assisted grading system for Squamous Cell Carcinoma of Head and Neck (SCCHN) immuno-histochemical (IHC) images. This system quantitatively evaluates biomarker expression in addition to cancerous and textural features. The stepwise grading approach includes: 1) image preprocessing to rectify variations due to illumination and acquisition conditions, 2) biomarker quantifi- cation using stain color references, 3) marking of cancerous and non-cancerous region of interest (ROI), and 4) grade evaluation based on the percentage of folate receptors (FR) expressed in the cancerous region along with nuclear and textural features. We analyze the grading efficacy of different feature types (biomarker, cancerous and textural) for computer assisted grading of SCCHN. Our methodology encompasses image based morphological tech- niques to provide new quantitative measures for typical cancer attributes (nuclear atypia, pleomorphism and necrosis) and allows usage of these attributes along with biomarker expression characteristics. We intend our system to assist pathologists in a clinical setting by correlating the FR expression to cancer grade. We obtain grading accuracy up to 94% for a heterogeneous image dataset and our results show that biomarker (FR) features result in considerable improvement in computer assisted grading of SCCHN as compared to cancerous and textural features.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络