Automatic Classification of Non Hodgkin‘s Lymphoma using Histological Images: Recent Advances and Directions
Pavani Battula, Shanu Sharma · 2018 International Conference on Advances in Computing, Communication Control and Networking (ICACCCN) · 2018
Lymphoma is a type of blood cancer, whose around 80,000 new cases are diagnosed every year. It is generated in body's immune system cells like lymph nodes, spleen, bone marrow and other parts of the body. The timely detection of the exact type of lymphoma is necessary for the early treatment of patients and for their prognosis. The lymphoma can bediagnosed by performing lymph node biopsy followed by histopathological analysis and immunohistochemistry methods. The histopathologic analysis is the most important diagnostic criteria which analyses the morphologic features of tumor under microscope using hematoxylin and eosinstained slides. Despite the analysis is carried out by expert hematologists and pathologist, the diagnosis is difficult due to various factors like requirement of human expertise, uncertainties, difference in staining of slides etc. The combination of image processing and machine learning techniques are emerging as the essential tool in various cancer detections tasks. These techniques have been employed by the researchers in classifying sub types of lymphoma CLL, MCL, FLL, whereas due to the complex features of these subtypes, the developed systems are not so efficient and progress is still going on to optimize different steps of lymphoma detection and classification. In this paper the existing work in past ten years has been analyzed and methodologies used in different steps are summarized. Paper is concluded with discussing the challenges and future directions for further improvement.