Automatic Detection of Micrometastases in Lymph Nodes by Infrared Micro-Spectral Imaging
H Ashby-Richardson, NM Laver, SP Naber, BL Bird, M Diem. Tufts Medical Center, Boston, MA; Northeastern University, Boston, MA
Background: Infrared micro-spectral imaging (IRMSI) is a novel, optical technique that can provide a rapid measurement of sample biochemistry and identify variations that occur between healthy and abnormal tissue. The advantage of this method is that it is objective and provides reproducible results, independent of fatigue, experience and inter-observer variability. This study explores the correlation between the spectral and conventional histopathology of axillary lymph nodes containing breast micrometastases (MMT).
Design: Twenty axillary lymph nodes, known to have contained breast MMT's, were selected from the Tufts Medical Center tissue archive. Unstained tissue sections were cut for investigation using IRMSI. The tissue section was interrogated by a beam of IR light that sampled pixels (6.25 m x 6.25 m in size) and consecutive 1 mm x 1 mm IR images were recorded from regions that previously showed MMT's. Each IR image consisted of 25,600 complete IR spectra that describe the tissue's discrete biochemistry at each pixel coordinate. By use of Hierarchical Cluster Analysis (HCA), IR spectra were sorted into groups dependent upon their spectral similarity and pseudo-color images were constructed to compare against conventional histopathology following H&E staining of the tissue. Correlating the tissue's spectral and histopathological features, a diagnostic computer algorithm was generated.
Results: More than 30 IR images from axillary lymph nodes were recorded for a total of 750,000 IR spectra. Multivariate analyses of the data revealed that each tissue type displayed its own distinct spectral pattern. Adipose tissue showed prominent IR bands associated with lipids, whereas the lymph node capsule showed strong IR bands associated with collagen. Spectral differences between lymphocytes and MMT's were also readily discernible, with distinct variations occurring to both the protein and phosphate composition of abnormal tissue. More importantly, small micrometastases (<150m), as well as small clusters of individual cancer cells (<25m) were correctly identified using this technique.
Conclusions: IR micro-spectral imaging can identify small micrometastases within excised tissue. The utilization of optimized IR imaging instruments may allow the acquisition and diagnosis of entire lymph nodes within a few minutes.
Monday, March 9, 2009 9:30 AM
Poster Session I Stowell-Orbison/Autopsy Award # 21, Monday Morning