Download Techniques to Improve 3D Optical Imaging

Techniques to Improve 3D Optical
Imaging Quantification and Sensitivity
Tamara Troy, Ali Akin, Ed Lim, Konnie Urban, Ning Zhang, Jae-Beom Kim, Jay Whalen,
Chaincy Kuo, Heng Xu and Brad Rice, Caliper Life Sciences, Alameda, CA USA
Experimental Results (cont)
Abstract
Optical tomography of bioluminescent and fluorescent reporters in pre-clinical animal models is an
important technology for the study of disease and drug development. However, the quality of 3D reconstructions is limited by the sensitivity of detection and these 3D results typically lack quantification
with biologically relevant units. Therefore, new methods to improve detection and to quantify the results in terms of absolute cell number or dye molecules will be discussed.
3D Bioluminescent CT Co-Registration
A) Uncompressed
In order to enhance signal levels, a mouse bed has been developed which can slightly compress an
animal thereby reducing the amount of tissue light propagates through. A direct comparison showing
that compression increases signal levels is shown. Reconstructions for a compressed and uncompressed animal models will be shown co-registered to μCT data for cross validation.
Sensitivity of fluorescence detection can further be improved through a calibration method called normalized transmission fluorescence (NTF) efficiency. Since the sensitivity of fluorescent imaging is
limited by autofluorescence, transillumination is used in order to trap the autofluroescent signal on the
opposite side from the detector. However for transillumination imaging, the pattern of the emission
light includes contributions from tissue heterogeneity. To reduce this contribution from the detected
signal, the fluorescent emission image is normalized by a transfer function comprised of a transmission image measured with the same emission filter and open excitation filter. Eliminating the excitation light contribution from the measurement improves the robustness of the algorithm by reducing artifacts and giving better signal localization. Improvements to the fluorescent signal using this
algorithm will be shown.
Number of Cells
5.8E+05
Actual
5.0E+05
B) Slightly Compressed
Lastly, we describe a method to absolutely quantify 3D images in terms of biological activity using a
well plate calibration technique. In this technique measurements of known serial dilutions of luminescent or fluorescent cells, or fluorescent dye molecules are used in order to generate quantification databases of photons per second per cell or extinction coefficients per cell or per molecule. These databases are then used to determine the number of cells or dye molecules from the tomographic sources
reconstructed inside an animal. We present validation of this technique with the direct implantation of
known cell numbers and dye molecules.
Instrumentation
Reconstructed
Number of Cells
Reconstructed
5.1E+05
Actual
5.0E+05
Figure 5. 3D reconstruction of the A) uncompressed and B) slightly compressed mouse co-registered to CT data showing
bioluminescent cells in the right lung. The 3D reconstruction used the well plate quantification to calculate the cell concentration
(the actual concentration is 5.0E+05 cells).
Fluorescent and Bioluminescent Optical Imaging System
Normalized Transmission Fluorescent, NTF Efficiency
Fluorescence- Raster Scan
Normalized Transmission
Fluorescence
Raster Scan Pattern
Transmission-Raster Scan
Structured Light Projector for Surface Topography
Structured Light Image
Height Map
Smoothed Surface
Figure 6. Pictorial diagram showing the process involved in obtaining a NTF image where the fluorescent image is divided by a
transmission image taken at the emission wavelength (no excitation filter). NTF Efficiency reduces the geometric contribution and
effects of tissue heterogeneity thereby giving better signal localization.
2D Transillumination Fluorescent Image
A) Uncompressed: Height 25.3mm
Mouse Imaging Shuttle and Docking Stations
A) MIS
B) Optical Docking Station
B) Slightly Compressed: Height 13.4mm
C) CT Docking Station
Figure 1. The Mouse Imaging Shuttle (MIS) 1.) is used to transfer subjects from one imaging platform to another without disrupting
its position for eventual co-registration between optical and CT measurements and 2.) has the ability to slightly compress the
subject in order to enhance the optical signal levels. A) CAD drawing of the MIS showing subject placement. B) MIS docking
station showing inlet and outlet anesthesia supply. C) CT docking station showing anesthesia connectors and triangular position
encoded fiducial used for automatic co-registration.*
* Please see poster P0323B in this session for further details on optical/CT co-registration.
Figure 7. Comparison of the A) uncompressed and B) slightly compressed transillumination fluorescent images. The animal was
compressed 11.9 mm yielding a 2.3 increase in surface signal level. The data above was taken from a CF750 pillow implant in the
left peritoneum.
3D NTFLIT CT Co-Registration
A) Uncompressed
Low Dose μCT
Imaging Conditions
Power: 90kV and 160μA
Number of Dye Molecules
FOV: 60mm
Time: 17-34sec
# Views: 514-1028
Reconstructed
0.005 pmole
Actual
0.010 pmole
Dose: 13-26 mGy
Figure 2. A μCT was used for cross validation. The Mouse Imaging Shuttle,
MIS is transported between the optical in vivo imaging system and CT imaging
system via docking stations in order to acquire precise animal positioning.
B) Slightly Compressed
Experimental Results
Number of Dye Molecules
Well Plate Quantification
Figure 3. Serial dilution (2:1) of human colorectal carcinoma
HTC116-luc2 bioluminescent cells with a starting concentration of
2x105 cells. Well plate data is used to determine the Total Flux per
cell and is used by the 3D tomography to extrapolate the total
number of cells from the reconstructed signal.
Reconstructed
0.006 pmole
Actual
0.010 pmole
Figure 8. 3D reconstruction of the A) uncompressed and B) slightly compressed mouse co-registered to CT data showing the fluorescent CF750 pillow implant in the left peritoneum. The 3D reconstruction used a well plate quantification to calculate the number
of dye molecules (the actual number is 0.010 pmole).
2D Bioluminescent Images
A) Uncompressed: Height 21.5mm
B) Slightly Compressed: Height 13.4mm
Conclusions
• The bioluminescent well plate quantification was able to reconstruction the absolute cell number with a
16% error for the uncompressed mouse and 3% error for the compressed mouse.
• Compressing the animal by 8 mm increased the bioluminescent signal by 1.7.
• The fluorescent well plate quantification was able to reconstruction the number of dye molecules within a
factor of 2 for both the uncompressed and the compressed mouse.
• Compressing the animal by 11.9 mm increased the fluorescent signal by 2.1.
• Co-registration of the optical reconstruction to CT data shows exact signal location.
• NTF efficiency measurements improves signal localization by reducing excitation light artifacts.
Figure 4. Comparison of the A) uncompressed and B) slightly compressed bioluminescent images from the ventral side of a
Balb/c male mouse (25g in weight) with a direct injection of 5x105 HCT 116-luc2 cell into the right lung. The animal was
compressed 8.1 mm yielding a 1.7 increase in surface signal level. The animal was IP injected with 150 μL of D-luciferin
solution 10 minutes prior to imaging.
References
1. C. Kuo, O. Coquoz, T.L. Troy, H. Xu and B.W. Rice, “Three-Dimensional Reconstruction of In Vivo
Bioluminescent Sources Based on Multi-Spectral Imaging,” J. Biomed. Opt. 12, 024007 (2007).
2. C. Kuo and B.W. Rice, “Bayesian Techniques in Fluorescent Tomography,” Biomedical Optics
Annual Meeting, OSA Technical Digest, BSuE46 (2008).
3. H. Xu, A. Christensen, C. Kuo, D. Nilson, E. Lim, J. Whalen, J. Kim, N. Zhang, R. Singh,
S.J. Oldfield, T.L. Troy and B. Rice, “Co-registration of In Vivo Optical Tomography and μCT
for Longitudinal Studies,” Word Molecular Imaging Congress, 0323B (2010).