Proliferation score prediction using novel SMHC feature using adaptive XGBoost model

Mitosis cell counting from histopathology image is one of the important process as a part of proliferative activity for cancer grading. It provides a level of progression and estimate the aggressiveness of particular diseases. Unfortunately, manual mitosis counting evaluation is a tedious, very labor intensive, and challenging task in analysing grade of a particular cancer. Since breast cancer recurrence and metastasis are intrinsically related to mortality, it is critical to predict the recurrence and metastasis risk of an individual patient, which is essential for adjutant therapy and early intervention. The aim of this study is to propose supervised model to solve the problem of mitosis detection and reduced the time lapse of process compared to the manual counting. We formulate a Low-Rank Matrix Recovery (LRMR) model to find salient regions from microscopy images and extract candidate patch sequences, which potentially contain mitosis cells. Once the patches are segmented, a novel Sparse Patch based Mitosis Handcrafted Feature (SMHCF) are extracted and based on the feature values a eXtreme Gradient Boosting (XGBoost) model are used to classify the mitosis and non-mitosis cells and count of the each cells. Our method is compared with the manual counting values and other state of the art methods. In that performances of our model achieved with better precision is 0.971, recall is 0.960, F1-score is 0.963 rates.