Worldwide, up to 40% of the breast conserving surgeries require additional operations due to positive resection margins. We propose to reduce this percentage by using hyperspectral imaging for resection margin assessment during surgery.
Spectral hypercubes were collected from 26 freshly excised breast specimens with a pushbroom camera (900-1700nm). Computer simulations of the penetration depth in breast tissue suggest a strong variation in sampling depth (~0.5-10 mm) over this wavelength range. This was confirmed with a breast tissue mimicking phantom study. Smaller penetration depths are observed in wavelength regions with high water and/or fat absorption. Consequently, tissue classification based on spectral analysis over the whole wavelength range becomes complicated. This is especially a problem in highly inhomogeneous human tissue.
We developed a method, called derivative imaging, which allows accurate tissue analysis, without the impediment of dissimilar sampling volumes. A few assumptions were made based on previous research. First, the spectra acquired with our camera from breast tissue are mainly shaped by fat and water absorption. Second, tumor tissue contains less fat and more water than healthy tissue. Third, scattering slopes of different tissue types are assumed to be alike.
In derivative imaging, the derivatives are calculated of wavelengths a few nanometers apart; ensuring similar penetration depths. The wavelength choice determines the accuracy of the method and the resolution. Preliminary results on 3 breast specimens indicate a classification accuracy of 93% when using wavelength regions characterized by water and fat absorption. The sampling depths at these regions are 1mm and 5mm.
Esther Kho, Lisanne L. de Boer, Koen K. Van de Vijver, Henricus J. C. M. Sterenborg, and Theo J. M. Ruers, "Overcoming sampling depth variations in the analysis of broadband hyperspectral images of breast tissue (Conference Presentation)," Proc. SPIE 10060, Optical Biopsy XV: Toward Real-Time Spectroscopic Imaging and Diagnosis, 1006013 (Presented at SPIE BiOS: February 01, 2017; Published: 1 May 2017); https://doi.org/10.1117/12.2250871.5409321647001.
Conference Presentations are recordings of oral presentations given at SPIE conferences and published as part of the proceedings. They include the speaker's narration with video of the slides and animations. Most include full-text papers. Interactive, searchable transcripts and closed captioning are now available for most presentations.
Search our growing collection of more than 22,000 conference presentations, including many plenaries and keynotes.