Evaluating Feature Extraction in Ovarian Cancer Cell Line Co-Cultures Using Deep Neural Networks

This dataset provides detailed imaging data from various co-culture assays of ovarian cancer and fibroblast cell lines, treated with a wide range of drugs. The structured organization and comprehensive naming conventions allow for easy navigation and analysis of the data. The images are treated with 528 drugs from FIMM Oncology Library to study the drug effect on cancer cell morphology in presence of fibroblasts.

The dataset comprises images from 2D coculture high-content screening data in .tiff format. The images were acquired using the Opera Phenix at a 10x magnification. It includes a total of 245,760 raw images (including 4 field of views), each with a resolution of 1080x1080 pixels. For initial analysis, the images were read directly into CellProfiler, a software platform designed for high-throughput image analysis. To facilitate neural network processing, each image was converted into a NumPy array using Python 3 and the Python Imaging Library (PIL).

The data set is available for download in five separate ZIP archives, Kuramochi_BjhTERT.zip (93.68 GB), Kuramochi_WI38.zip (93.50 GB), MH_BjhTERT.zip (62.22 GB), OvCar3_BjhTERT.zip (84.98 GB), OvCar8_WI38.zip (91.89 GB).

For a description on the file structure, see associated documentation file Dataset_Description.pdf.