Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Unsupervised high-throughput segmentation of cells and cell nuclei in quantitative phase images

Julia Sistermanns, Ellen Emken, Gregor Weirich, Oliver Hayden, Wolfgang Utschick

TL;DR

This work proposes an unsupervised multistage method that segments correctly without confusing noise or reflections with cells and without missing cells that also includes the detection of relevant inner structures, especially the cell nucleus in the unstained cell.

Abstract

In the effort to aid cytologic diagnostics by establishing automatic single cell screening using high throughput digital holographic microscopy for clinical studies thousands of images and millions of cells are captured. The bottleneck lies in an automatic, fast, and unsupervised segmentation technique that does not limit the types of cells which might occur. We propose an unsupervised multistage method that segments correctly without confusing noise or reflections with cells and without missing cells that also includes the detection of relevant inner structures, especially the cell nucleus in the unstained cell. In an effort to make the information reasonable and interpretable for cytopathologists, we also introduce new cytoplasmic and nuclear features of potential help for cytologic diagnoses which exploit the quantitative phase information inherent to the measurement scheme. We show that the segmentation provides consistently good results over many experiments on patient samples in a reasonable per cell analysis time.

Unsupervised high-throughput segmentation of cells and cell nuclei in quantitative phase images

TL;DR

This work proposes an unsupervised multistage method that segments correctly without confusing noise or reflections with cells and without missing cells that also includes the detection of relevant inner structures, especially the cell nucleus in the unstained cell.

Abstract

In the effort to aid cytologic diagnostics by establishing automatic single cell screening using high throughput digital holographic microscopy for clinical studies thousands of images and millions of cells are captured. The bottleneck lies in an automatic, fast, and unsupervised segmentation technique that does not limit the types of cells which might occur. We propose an unsupervised multistage method that segments correctly without confusing noise or reflections with cells and without missing cells that also includes the detection of relevant inner structures, especially the cell nucleus in the unstained cell. In an effort to make the information reasonable and interpretable for cytopathologists, we also introduce new cytoplasmic and nuclear features of potential help for cytologic diagnoses which exploit the quantitative phase information inherent to the measurement scheme. We show that the segmentation provides consistently good results over many experiments on patient samples in a reasonable per cell analysis time.
Paper Structure (14 sections, 5 equations, 3 figures, 1 table)

This paper contains 14 sections, 5 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Data
  3. Cytologic Samples and Diagnostic Process
  4. Optical Setup
  5. Segmentation Method
  6. Statistical Analysis of Phase Values
  7. Threshold Stage
  8. Checking Mechanisms
  9. Detection of Internal Structures
  10. New Features
  11. Shape
  12. Nuclear and Cytoplasmic Features
  13. Results
  14. Conclusion

Figures (3)

  • Figure 1: Two Examples of the proposed cell detection process (section \ref{['sec:statistics']}- \ref{['sec:check']}). From left to right: (1) quantitative phase image, (2) image after the threshold segmentation (step 1$\&$2), (3) plausibility check (step 3) which discards contours which are too small, lie inside another contour or are noise or debris; (4) To better highlight the detected cells the enscribing circle around the contours and the bounding box are highlighted.
  • Figure 2: Two examples of the internal detection and feature extraction. Features: threshold $t$, cell diameter $d_c$, roundness $\gamma_c$, circularity $\chi_c$, polygonality $\psi_c$, roundness nucleus $\gamma_n$, circularity nucleus $\chi_n$, nuclear-cytoplasmic ratio $\Delta_{n,c}$, volumetric ratio $\nu_{n,c}$, position of nucleus $\delta_{n,c}$, number possible nuclei $N_{n,c}$, number internal structures in nucleus $N_{n,i}$, maximum and mean optical density nucleus $\rho_{o,n}$ and $\bar{\rho}_{o,n}$
  • Figure 3: Examples of occuring errors. The occurring errors were categorized into the following classes, adapted from Loewke:2018aa to include the detection of inner structures: (1) missed cell (false negative), (2) not-a-cell (false positive), (3) poor cell boundary, (4) missed internal structure, (5) not-a-nucleus: chosen nucleus red, possible nuclei yellow, (6) poor nucleus boundary.