Biomedical Image Analysis

ICIP 2015

"A novel feature descriptor based on microscopy image statistics"

Neslihan Bayramoglu, Juho Kannala, Malin Akerfelt, Mika Kaakinen, Lauri Eklund,Matthias Nees, Janne Heikkilä

We propose a novel feature description algorithm based on image statistics. The pipeline first performs independent component analysis on training image patches to obtain basis vectors (filters) for a lower dimensional representation. Then for a given image, a set of filter responses at each pixel is computed. Finally, a histogram representation, which considers the signs and magnitudes of the responses as well as the number of filters, is applied on local image patches. We propose to apply this idea to a microscopy image pixel identification system based on a learning framework.

ICPR 2014

"Detection of tumor cell spheroids from co-cultures using phase contrast images and machine learning approach"

Bayramoglu, Neslihan, Mika Kaakinen, Lauri Eklund, Malin Akerfelt , Matthias Nees, Juho Kannala, Janne Heikkila

In this work, we propose a method for detecting tumour cell spheroids in phase contrast (PC) images of co-cultures of fibroblasts and tumor cells in an experimental 3D model.
PC images are first filtered to remove noise. Superpixel segmentation is performed to reduce complexity and to work on compact regions that follow image boundaries. A Random Forest classifier is trained on features that are extracted from both positive and negative samples. During test time similar process is followed and superpixels are assigned to probabilistic outputs based on the average decision generated by the trees. Final decision is made by simple thresholding.

Data Download

A sample image from our database. (Left) Fluorescent image. Green Fluorescent Protein (GFP) is used to label fibroblast cells. (Middle Left) Phase contrast image of a 3D culture containing tumor and fibroblast cells. (Middle Right) Phase contrast and fluorescent images are superimposed. (Right) Tumor spheroid mask image.

You can download phase contrast images and tumor masks.

Download "Col" data

Download "Mat" data (Download splitted train and test images for "Mat" data)

Download "Mat+Col" data (Download splitted train and test images for "Mat+Col" data)

Please cite the publications below to refer to the datasets:

Related Publications

Bayramoglu, Neslihan, Mika Kaakinen, Lauri Eklund, Malin Akerfelt , Matthias Nees, Juho Kannala, Janne Heikkilä, "Detection of tumor cell spheroids from co-cultures using phase contrast images and machine learning approach", In 22nd International Conference on Pattern Recognition (ICPR) 2014.

Bayramoglu, Neslihan, Juho Kannala, Malin Akerfelt, Mika Kaakinen, Lauri Eklund,Matthias Nees, Janne Heikkilä, "A novel feature descriptor based on microscopy image statistics", International Conference on Image Processing (ICIP) 2015.