Datasets used in "3-D quantification of filopodia in motile cancer cells" IEEE Transactions on Medical Imaging 38(3), 862-872 (2019)

Citation Author(s):
Carlos
Castilla
Center for Applied Medical Research, University of Navarra, Pamplona, Spain
Martin
Maska
Centre for Biomedical Image Analysis Faculty of Informatics, Masaryk University, Brno, Czech Republic
Dmitry
Sorokin
Centre for Biomedical Image Analysis Faculty of Informatics, Masaryk University, Brno, Czech Republic
Erik
Meijering
Departments of Medical Informatics and Radiology, Erasmus University Medical Center Rotterdam, The Netherlands
Carlos
Ortiz-de-Solorzano
Center for Applied Medical Research, University of Navarra, Pamplona, Spain
Submitted by:
Carlos Ortiz De...
Last updated:
Tue, 05/17/2022 - 22:17
DOI:
10.21227/H26W9K
Research Article Link:
Links:
License:
602 Views
Categories:
0
0 ratings - Please login to submit your rating.

Abstract 

These datasets were used to produce the results of the following TMI paper: "3D Quantification of Filopodia in Motile Cancer Cells", Castilla C., et al. (2019). IEEE Transactions on Medical Imaging 38(3):862,872.

 

3D+t image real and synthetic  sequences of cells of the A549 lung adenocarcinoma cancer cell line, displaying three different phenotypes of CRMP-2, a protein involved in the assembly and disassembly of actin filaments. All real videos were acquired on aUltraviewERS (Perkin Elmer, Inc., Waltham, MA, USA), spinning disk confocal microscope, using the 488 nm line of an Ar/Kr laser to image the sample through a Plan-Apochromatic 63x 1.20 NA water immersion objective lens (Carl Zeiss, AG., Wetzlar, Germany). The videos contained one cell, imaged every two minutes duringone hour. The original image voxel size of 0.126x0.126x1.0 µm was resampled in the axial direction using cubic spline interpolation to obtain isotropic image data with a voxel size of 0.12x0.126x0.126 µm. The synthetic videos were produced using our recent cell simulator [1], to reproduce as closely as possible the image properties and cell phenotypes of the real videos.

 

Along with the datasets, we provide AVI videos displaying the results of maximum intensity projections (MIP) of the segmentation of the cells using two segmenation methods, one based on the minimization of the Chan-Vese model (CVS) and a 3D Convolutional Neural Network (CNN)

 

The file names encode the following information:

 

CRMP2 phenotype and video type:

- WTR: Wild type, real video

- OER: Over expressing, real video

- PDR: Phospho-defective, real video

- WTS: Wild type, synthetic video

- OES: Over expressing, synthetic video

- PDS: Phospho-defective, synthetic video

 

Segmentation method:

- CVS: Minimization of the Chan-Vese models

- CNN: Convolutional Neural Network

 

Video number:
- 01

- 02

- 03

 

[1]    D. V. Sorokin, I. Peterlík, V. Ulman, D. Svoboda and M. Maška, “Model-based generation of synthetic 3D time-lapse sequences of motile cells with growing filopodia,” In IEEE International Symposium on Biomedical Imaging, pp. 822–826, 2017

 

Instructions: 

Each zip file contains a set of 3D image frames.

Dataset Files

LOGIN TO ACCESS DATASET FILES