Automatic Cell Detection in Bright-Field Microscope Images Using SIFT, Random Forests, and Hierarchical Clustering

Abstract

Abstract-We present a novel machine learning-based system for unstained cell detection in bright-field microscope images. The system is fully automatic since it requires no manual parameter tuning. It is also highly invariant with respect to illumination conditions and to the size and orientation of cells. Images from two adherent cell lines and one suspension cell line were used in the evaluation for a total number of more than 3500 cells. Besides real images, simulated images were also used in the evaluation. The detection error was between approximately zero and 15.5% which is a significantly superior performance compared to baseline approaches.

Improving joint learning of suspended and adherent cell detection using low-pass monogenic phase and transport of intensity equation

“…We use the automatic detection pipeline of [7]: SIFT keypoints [20] are extracted from a defocused image. These keypoints are classified using a random forest into background or cell keypoints.…”

“…The problem of cell detection in microscope images was addressed by several research papers in the last few years [1][2][3][4][5][6][7][8][9]. The difficulty of the problem is inherently related to image modality.…”

“…In [7] and [9], defocused images were directly used for image processing and/or feature extraction. On the other hand, in [10], the defocused images were used for solving the TIE and obtaining a quantitative phase map.…”

“…Recently, [7] obtained high detection accuracy on defocused bright-field images. However, the detection error when the algorithm was trained using both suspended and adherent cell lines was higher compared to the case when the algorithm was trained separately for each cell line.…”

Deep Learning Based Automatic Immune Cell Detection for Immunohistochemistry Images

“…Since the most popular immune cell markers, such as CD3 and CD8 for universal T-cells and cytotoxic T-cells respectively, are membrane markers, the stain appears as a ring instead of a blob. Another machine learning based system using SIFT, Random Forests, and Hierarchical Clustering was developed by Mualla et al in [5] for unstained cell imaging which has the properties of maintaining sufficient contrast of cell boundaries. In this work, the SIFT key-points are classified into cells and backgrounds, and all the key-points within each cell are linked together using hierarchical clustering.…”