High dynamic range implementation on acute leukemia slide images

Abstract

Conventional diagnosis of acute leukemia is based on examining the morphology of the blood and bone marrow smear under the microscope and the process can be tedious, time intensive and highly skilled resources dependent. Over recent years, features fusion techniques base on statistical and morphological features had been explored in computer vision studies to enhance the capability of acute leukemia diagnosis task. However, microscopic image capture from the light microscope usually has poor quality due to the limited dynamic range of the camera. Thus, image regions with intensity levels outside the dynamic range captured by the camera sensor suffer from lack of details, appearing either underexposed or overexposed. This paper proposed a High Dynamic Range (HDR) imaging techniques to solve the problem of limited dynamic range and enhance the morphological features of blast cells. The proposed method consists of two main parts: implementing HDR techniques on acute leukemia slide images and comparing the dynamic range between HDR image and original images with different exposure time based on the intensity histograms obtained. The results presented showed that the HDR implementation has enhanced the morphological features of blast cells and increase the dynamic range, hence may benefit in the feature extraction and classification process of acute leukemia.