A review on computer aided detection and classification of leukemia

Leukemia is a non-tumor type of cancer and its early diagnosis is important in the treatment and prognosis. Image based diagnosis is quick and easy compared to the conventional methods. This study aims to review the related works on computer aided diagnosis of leukemia by categorizing the works into Machine Learning (ML) and Deep Learning (DL) based technologies. The related works are identified by searching databases of eminent publishers such as ScienceDirect, Springer, IEEE Xplore, Wiley etc. over the years 2005 to 2022. The works are then grouped into those used ML and DL based classifiers. The study identified that under ML based works the SVM was used in majority (50%) of the studies reviewed and under DL category CNN was widely used in about 69% of the works considered. There are only a few works available for the classification of chronic leukemia under the ML category and no works available under the DL category. The proposed review analyzed the works based on the classifier, datasets, features, and application area involved in the studies. There is further scope of research for developing public datasets of leukemia, especially for chronic leukemia, and for developing new automated diagnostic methods for the classification of different types of leukemia. There is a shift towards DL based studies for computer aided diagnosis of leukemia from the year 2019 onwards and not much reviews which classify the related works into ML and DL based techniques are available in the literature.