FRIDA An open source framework for image dataset analyis
Toby Cornish MD; The Johns Hopkins Medical Institutions; Angelo D. DeMarzo PhD; The Johns Hopkins Medical Institutions; Bora Gurel PhD; The Johns Hopkins Medical Institutions ; James Morgan BS; The Johns Hopkins Medical Institutions;
Content:
Analysis of digital photomicrographs has become routine practice in many pathology studies. Frequently, these analyses require repetitive processing of image datasets containing an arbitrary number of images. Commercial software packages exist that help automate the analysis process, but the number of open source packages designed for image dataset processing is limited. FRIDA (an acronym for FRamework for Image Dataset Analysis) is a custom software package for the analysis of RGB color image datasets, including those generated from the automated scanning of tissue microarray slides.
Technology:
FRIDA is a platform-independent, open source software package written in Java. It uses the image processing and analysis libraries from ImageJ (Wayne Rasband, NIH). It can be deployed remotely using Java Web Start allowing remote analysis of image datasets.
Design:
FRIDA implements a flexible framework for image analysis capable of processing image datasets of arbitrary size. It is currently available for download and use and is under active development (http://bui2.win.ad.jhu.edu/image_analysis/). Central to the design of FRIDA is the process of masking of images. Image masking enables the user to interactively include and exclude portions of the image or to automatically select pixels of specific colors, such as brown immunohistochemical staining or blue hematoxylin staining, using hue-saturation-brightness color space thresholding. Logical combination of image masks allows for the analysis of specific regions of interest based on morphology, pixel color, and other user-defined criteria. The software is designed for standalone image dataset analysis but has features that facilitate input of results into the Johns Hopkins TMAJ software package for tissue microarray creation and data management (http://tmaj.pathology.jhmi.edu/). Future features of FRIDA will include closer integration with TMAJ, compatibility with arbitrary multichannel image data such as fluoresence or multispectral data, integration of native ImageJ plugins for masking and pre-analysis image filtering, and the addition of analytical methods such as particle counting and measurement.
Results:
N/A
Conclusion:
FRIDA represents an alternative to commercial software for the analysis of color image datasets. Pathologists and other researchers that work with color image datasets may be interested in using or contributing to this open source project.
