Premium
SU‐FF‐J‐117: Integrated Software Tools for Multi‐Modality Functional Images in Cancer Clinical Trials
Author(s) -
Cao Y,
Shen Z
Publication year - 2007
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2760622
Subject(s) - computer science , medical imaging , voxel , data mining , cluster analysis , modality (human–computer interaction) , population , software , medical physics , artificial intelligence , medicine , environmental health , programming language
Purpose: Many evolving concepts of quantitative, functional, cancer imaging for treatment planning and assessment involve a combination of data acquisition and complex post‐processing to permit spatial, physiological, and structural assessment of the tumor and normal tissue environment. An infrastructure is needed for rapid prototyping and routine processing of image and ancillary data spanning various extents of space, time, and multiple imaging modalities. Methods: A suite of functional image analysis tools (FIAT) has been developed. This infrastructure operates across variable ranges of input data, temporal scales (single samples, dynamic contrast series, samples during and post therapy), and information sources (MRI, CT, nuclear medicine images, dose distributions, etc.). It has been designed to rapidly prototype a complex functional data analysis series, and then to apply prototyped methodologies to large batches of data, combining the outputs for population trend analysis. Several groups of quantitative analysis tools have designed and developed to allow rapid, intuitive, initial trials at data analysis. Interfacing with statistical analysis packages has been integrated into the design. Advanced segmentation via fuzzy c‐means clustering is provided. Validation of analysis paradigms via simulations is supported. Results: While FIAT is constantly evolving, three years of development has generated a highly flexible suite of analysis tools, supporting a variety of different trials including early assessment of liver damage from radiation, neuronal disruption from cancer and treatment, variations in regional and temporal contract uptake, dynamic susceptibility changes, and temporal aspects of methionine uptake. Regional analyses cross scales from individual voxels through regions of arbitrary shape or size. Conclusion: The infrastructure under development with FIAT has been a very powerful investigational tool for early quantitative cancer imaging studies. As this analysis environment matures, it should become a critical component of multi‐center trials involving imaging and cancer treatment assessment.