Differential diagnosis of lung carcinoma with three-dimensional quantitative molecular vibrational imaging

Liang Gao, Ahmad A. Hammoudi, Fuhai Li, Michael J. Thrall, Philip T. Cagle, Yuanxin Chen, Jian Yang, Xiaofeng Xia, Yubo Fan, Yehia Massoud, Zhiyong Wang, Stephen T.C. Wong

Research output: Contribution to journalArticle

11 Scopus citations

Abstract

The advent of molecularly targeted therapies requires effective identification of the various cell types of non-small cell lung carcinomas (NSCLC). Currently, cell type diagnosis is performed using small biopsies or cytology specimens that are often insufficient for molecular testing after morphologic analysis. Thus, the ability to rapidly recognize different cancer cell types, with minimal tissue consumption, would accelerate diagnosis and preserve tissue samples for subsequent molecular testing in targeted therapy. We report a label-free molecular vibrational imaging framework enabling three-dimensional (3-D) image acquisition and quantitative analysis of cellular structures for identification of NSCLC cell types. This diagnostic imaging system employs superpixel-based 3-D nuclear segmentation for extracting such disease-related features as nuclear shape, volume, and cell-cell distance. These features are used to characterize cancer cell types using machine learning. Using fresh unstained tissue samples derived from cell lines grown in a mouse model, the platform showed greater than 97% accuracy for diagnosis of NSCLC cell types within a few minutes. As an adjunct to subsequent histology tests, our novel system would allow fast delineation of cancer cell types with minimum tissue consumption, potentially facilitating onthe- spot diagnosis, while preserving specimens for additional tests. Furthermore, 3-D measurements of cellular structure permit evaluation closer to the native state of cells, creating an alternative to traditional 2-D histology specimen evaluation, potentially increasing accuracy in diagnosing cell type of lung carcinomas.

Original languageEnglish
Article number066017
JournalJournal of biomedical optics
Volume17
Issue number6
DOIs
StatePublished - Jun 1 2012
Externally publishedYes

Keywords

  • Artificial intelligence
  • Lung cancer
  • Microscopy
  • Non-linear diagnostic imaging
  • Non-near optics

Fingerprint Dive into the research topics of 'Differential diagnosis of lung carcinoma with three-dimensional quantitative molecular vibrational imaging'. Together they form a unique fingerprint.

  • Cite this

    Gao, L., Hammoudi, A. A., Li, F., Thrall, M. J., Cagle, P. T., Chen, Y., Yang, J., Xia, X., Fan, Y., Massoud, Y., Wang, Z., & Wong, S. T. C. (2012). Differential diagnosis of lung carcinoma with three-dimensional quantitative molecular vibrational imaging. Journal of biomedical optics, 17(6), [066017]. https://doi.org/10.1117/1.JBO.17.6.066017