Site internet : http://www.lifexsoft.org

Introduction

Texture analysis is gaining considerable interest in medical imaging, in particular to identify parameters that might characterize tumor heterogeneity. We developed an easy-to-use freeware enabling calculation of a broad range of conventional, textural and shape indices from PET, MR and CT images.

The software is written in Java and does not rely on any commercial libraries. LIFEx is dedicated to researchers, radiologists, nuclear medicine physicians and oncologists and includes two operation modes:

1) “validated”: simplified user interface displaying conventional indices (e.g. SUV, Metabolic Volume and Total Lesion Glycolysis in FDG PET) as well as 6 textural indices calculated with default settings that have been thoroughly validated and proved as robust in previous publications;
2) “advanced”, giving access to 37 histogram, textural and shape indices in addition to conventional indices and where users can change calculation options (e.g. resampling method and number of grey levels for textural analysis).

Features

LIFEx reads DICOM images locally or over a network using a DICOM browser, is compatible with Osirix and includes a powerful 3D reconstruction-based slice viewer. Volumes of interest (VOI) can be either imported from previously created files or drawn and manipulated using LIFEx. Results are exported in Excel format files. LIFEx runs on Windows, MacOs and Linux. It is distributed with examples and includes a tutorial. User support is available for the “validated” mode. Users can optionally contribute to the gathering of index values measured in different tissue types and different images as a public data bank of reference values is currently being built and integrated in the software for assisting the users with the interpretation of their results.

LIFEx has already been distributed to research labs, nuclear medicine departments and radiology departments for investigating different tumor types (gliomas, cervix, lung, breast, and colorectal tumors), and has been very positively received. The intuitive interface associated with the tutorial made it fast to master for staff, and allowed us to start building databases of normal textural values in brain (white and grey matter), breast, liver, lung, fat, and muscles for various imaging equipment and protocols in PET and CT, while MR data are currently being collected and processed. Such data enable an accurate characterisation of the variability of different textural indices in a given imaging modality as a function of the scanner and the imaging protocol. New indices are being implemented based on users requests as the software is intended to evolve over time based on advances in the field.