Translator Disclaimer
10 February 2011 Empirical factors affecting reconstruction of image-guided near infrared spectroscopy
Author Affiliations +
Proceedings Volume 7892, Multimodal Biomedical Imaging VI; 789204 (2011)
Event: SPIE BiOS, 2011, San Francisco, California, United States
Empirical factors in image reconstruction that can have a dramatic effect on final image quantification in image guided optical spectroscopy using MRI are examined. We present a model for image segmentation that finds that tumor region quantification is influenced by defined size of that region, and we conclude that the most effective segmentation is done using a ktrans map to define the tumor region along with guidance from a radiologist experienced in breast MRI. Data calibration also introduces another empirical factor and we weigh advantages and disadvantages of different methods of calibrating data based on imaging geometry and volume. Finally, we investigate the effects of a modified Levenburg- Marquardt reconstruction that uses a regularization parameter to make the reconstruction less ill-posed and limit the effects of image noise. We find that the regularization parameter must fall within a certain range to adequately smooth the image but if it is too large then image contrast is lost. Understanding and minimizing these factors' contribution will be increasingly important as MRI/NIRS becomes closer to clinical use to ensure that the best possible images are always produced to aid in prognostic clinical decisions.
© (2011) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Michael A. Mastanduno, Shudong Jiang, Roberta DiFlorio-Alexander, Brian W. Pogue, and Keith D. Paulsen "Empirical factors affecting reconstruction of image-guided near infrared spectroscopy", Proc. SPIE 7892, Multimodal Biomedical Imaging VI, 789204 (10 February 2011);

Back to Top