3D prostate histology reconstruction: an evaluation of image-based and fiducial-based algorithms

PURPOSE

Evaluation of in vivo prostate imaging modalities for determining the spatial distribution and aggressiveness of prostate cancer ideally requires accurate registration of images to an accepted reference standard, such as histopathological examination of radical prostatectomy specimens. Three-dimensional (3D) reconstruction of prostate histology facilitates these registration-based evaluations by reintroducing 3D spatial information lost during histology processing. Because the reconstruction accuracy may constrain the clinical questions that can be answered with these data, it is important to assess the tradeoffs between minimally disruptive methods based on intrinsic image information and potentially more robust methods based on extrinsic fiducial markers.

METHODS

Ex vivo magnetic resonance (MR) images and digitized whole-mount histology images from 12 radical prostatectomy specimens were used to evaluate four 3D histology reconstruction algorithms. 3D reconstructions were computed by registering each histology image to the corresponding ex vivo MR image using one of two similarity metrics (mutual information or fiducial registration error) and one of two search domains (affine transformations or a constrained subset thereof). The algorithms were evaluated for accuracy using the mean target registration error (TRE) computed from homologous intrinsic point landmarks (3-16 per histology section; 232 total) identified on histology and MR images, and for the sensitivity of TRE to rotational, translational, and scaling initialization errors.

RESULTS

The algorithms using fiducial registration error and mutual information had mean ± standard deviation TREs of 0.7 ± 0.4 and 1.2 ± 0.7 mm, respectively, and one algorithm using fiducial registration error and affine transforms had negligible sensitivities to initialization errors. The postoptimization values of the mutual information-based metric showed evidence of errors due to both the optimizer and the similarity metric, and variation of parameters of the mutual information-based metric did not improve its performance.

CONCLUSIONS

The extrinsic fiducial-based algorithm had lower mean TRE and lower sensitivity to initialization than the intrinsic intensity-based algorithm using mutual information. A model relating statistical power to registration error for certain imaging validation study designs estimated that a reconstruction algorithm with a mean TRE of 0.7 mm would require 27% fewer subjects than the method used to initialize the algorithms (mean TRE 1.3 ± 0.7 mm), suggesting the choice of reconstruction technique can have a substantial impact on the design of imaging validation studies, and on their overall cost.

Determining the unique refractive index properties of solid polystyrene aerosol using broadband Mie scattering from optically trapped beads

A method is described to measure the refractive index dispersion with wavelength of optically trapped solid particles in air. Knowledge of the refraction properties of solid particles is critical for the study of aerosol; both in the laboratory and in the atmosphere for climate studies. Single micron-sized polystyrene beads were optically trapped in air using a vertically aligned counter-propagating configuration of focussed laser beams. Each bead was illuminated using white light from a broadband light emitting diode (LED) and elastic scattering within the bead was collected onto a spectrograph. The resulting Mie spectra were analysed to accurately determine polystyrene bead radii to ±0.4 nm and values of the refractive index to ±0.0005 over a wavelength range of 480-700 nm. We demonstrate that optical trapping combined with elastic scattering can be used to both accurately size polystyrene beads suspended in air and determine their wavelength dependent refractive index. The refractive index dispersions are in close agreement with reported values for polystyrene beads in aqueous dispersion. Our results also demonstrate a variation in the refractive index of polystyrene, from bead to bead, in a commercial sample. The measured variation highlights that care must be taken when using polystyrene beads as a calibration aerosol.

Degradation and rearrangement of a lung surfactant lipid at the air-water interface during exposure to the pollutant gas ozone

The presence of unsaturated lipids in lung surfactant is important for proper respiratory function. In this work, we have used neutron reflection and surface pressure measurements to study the reaction of the ubiquitous pollutant gas-phase ozone, O3, with pure and mixed phospholipid monolayers at the air-water interface. The results reveal that the reaction of the unsaturated lipid 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine, POPC, with ozone leads to the rapid loss of the terminal C9 portion of the oleoyl strand of POPC from the air-water interface. The loss of the C9 portion from the interface is accompanied by an increase in the surface pressure (decrease in surface tension) of the film at the air-water interface. The results suggest that the portion of the oxidized oleoyl strand that is still attached to the lipid headgroup rapidly reverses its orientation and penetrates the air-water interface alongside the original headgroup, thus increasing the surface pressure. The reaction of POPC with ozone also leads to a loss of material from the palmitoyl strand, but the loss of palmitoyl material occurs after the loss of the terminal C9 portion from the oleoyl strand of the molecule, suggesting that the palmitoyl material is lost in a secondary reaction step. Further experiments studying the reaction of mixed monolayers composed of unsaturated lipid POPC and saturated lipid dipalmitoyl-sn-glycero-3-phosphocholine, DPPC, revealed that no loss of DPPC from the air-water interface occurs, eliminating the possibility that a reactive species such as an OH radical is formed and is able to attack nearby lipid chains. The reaction of ozone with the mixed films does cause a significant change in the surface pressure of the air-water interface. Thus, the reaction of unsaturated lipids in lung surfactant changes and impairs the physical properties of the film at the air-water interface.

Optics clustered to output unique solutions: a multi-laser facility for combined single molecule and ensemble microscopy

Optics clustered to output unique solutions (OCTOPUS) is a microscopy platform that combines single molecule and ensemble imaging methodologies. A novel aspect of OCTOPUS is its laser excitation system, which consists of a central core of interlocked continuous wave and pulsed laser sources, launched into optical fibres and linked via laser combiners. Fibres are plugged into wall-mounted patch panels that reach microscopy end-stations in adjacent rooms. This allows multiple tailor-made combinations of laser colours and time characteristics to be shared by different end-stations minimising the need for laser duplications. This setup brings significant benefits in terms of cost effectiveness, ease of operation, and user safety. The modular nature of OCTOPUS also facilitates the addition of new techniques as required, allowing the use of existing lasers in new microscopes while retaining the ability to run the established parts of the facility. To date, techniques interlinked are multi-photon/multicolour confocal fluorescence lifetime imaging for several modalities of fluorescence resonance energy transfer (FRET) and time-resolved anisotropy, total internal reflection fluorescence, single molecule imaging of single pair FRET, single molecule fluorescence polarisation, particle tracking, and optical tweezers. Here, we use a well-studied system, the epidermal growth factor receptor network, to illustrate how OCTOPUS can aid in the investigation of complex biological phenomena.

Dynamic position and force measurement for multiple optically trapped particles using a high-speed active pixel sensor

A high frame rate active pixel sensor designed to track the position of up to six optically trapped objects simultaneously within the field of view of a microscope is described. The sensor comprises 520 x 520 pixels from which a flexible arrangement of six independent regions of interest is accessed at a rate of up to 20 kHz, providing the capability to measure motion in multiple micron scale objects to nanometer accuracy. The combined control of both the sensor and optical traps is performed using unique, dedicated electronics (a field programmable gate array). The ability of the sensor to measure the dynamic position and the forces between six optically trapped spheres, down to femtonewton level, is demonstrated paving the way for application in the physical and life sciences.

Classification of fixed urological cells using Raman tweezers

In this paper we report on preliminary investigations into using Raman tweezers to classify urological cell lines. This builds on earlier work within the group, whereby Raman tweezer methodologies were developed, and the application of this technique to differentiate between live prostate cancer (CaP) and bladder cells lines (PC-3 and MGH-U1 respectively) was demonstrated.In this present study we analysed chemically fixed cells using two different fixative methods; SurePath (a commercial available liquid based cytology media) and 4% v/v formalin/PBS fixatives. The study has been expanded from our previous live cell study to include the androgen sensitive CaP cell line LNCaP, primary benign prostate hyperplasia (BPH) cells as well as primary urethral cells. Raman light from the cells was collected using a 514.5 nm Ar-ion laser excitation source in back-scattering configuration mode.Principal component-linear discriminate analysis (PC-LDA) models of resulting cell spectra were generated and these were validated using a blind comparison. Sensitivities and specificities of > 72% and 90% respectively, for SurePath fixed cells, and > 93% and 98% respectively for 4% v/v formalin/PBS fixed cells was achieved. The higher prediction results for the formalin fixed cells can be attributed to a better signal-to-noise ratio for spectra obtained from these cells.Following on from this work, urological cell lines were exposed to urine for up to 12 hours to determine the effect of urine on the ability to classify these cells. Results indicate that urine has no detrimental effect on prediction results.

Raman tweezers sorting of single microbial cells

We have selectively isolated microbial cells by identifying and then manipulating cells using a combination of Raman microspectroscopy and optical trapping. The criterion for cell discrimination is based on spectral peak shifts within the Raman spectrum of individual cells. A specific shift in the phenylalanine peak position from 1001 rel. cm(-1) to 965 rel. cm(-1) is utilized to indicate the uptake of (13) C within the cell that utilized (13) C-substrate. Cells were captured and manipulated using an infrared (1064 nm) laser while Raman spectra were acquired over shorter timescales (30 s) using a co-aligned 514.5 nm laser beam. Selected cells were manoeuvred to a clean part of a capillary tube and the tubes were cleaved to physically separate the cells. The technique was tested for cell viability and cross-contamination effects using 70 single yeast cells (Saccharomyces cerevisia). Following these tests, 58 single bacterial cells (Escherichia coli DH5α, and Pseudomonas fluorescens SBW25::Km-RFP) that exhibited (13) C uptake were sorted from bacterial populations. Among those isolated cells, 11 out of 18 yeast cells and 7 out of 18 single SBW25::Km-RFP cells were recovered by incubation; 2 out of 7 sorted yeast cells and 3 out of 8 sorted bacterial cells (single SBW25::Km-RFP) were genome amplified correctly. We show that the Raman tweezers approach has the potential to open a new frontier to study unculturable microorganisms, which account for more than 99% microbes in natural environment.

Spectral discrimination of live prostate and bladder cancer cell lines using Raman optical tweezers

An investigation into the use of Raman optical tweezers to study urological cell lines is reported, with the ultimate aim of determining the presence of malignant CaP cells in urine and peripheral fluids. To this end, we trapped and analyzed live CaP cells (PC-3) and bladder cells (MGH-U1), because both prostate and bladder cells are likely to be present in urine. The laser excitation wavelength of 514.5 nm was used, with Raman light collected both in back- and forward-scattering geometric configurations. For the backscattering configuration the same laser was used for trapping and excitation, while for forward scattering a 1064 nm laser provided the trapping beam. Analysis of cell-diameter distributions for cells analyzed suggested normal distribution of cell sizes, indicating an unbiased cell-selection criterion. Principal components analysis afforded discrimination of MGH-U1 and PC-3 spectra collected in either configuration, demonstrating that it is possible to trap, analyze, and differentiate PC-3 from MGH-U1 cells using a 514.5 nm laser. By loading plot analysis, possible biomolecules responsible for discrimination in both configurations were determined. Finally, the effect of cell size on discrimination was investigated, with results indicating that separation is based predominantly on cell type rather than cell size.

Broadband Mie scattering from optically levitated aerosol droplets using a white LED

We describe a simple and efficient means of using a white LED source to illuminate an optically levitated aerosol droplet to enable study of broadband Mie scattering. The backscattered resonances are imaged through a spectrograph and CCD which show high resolution Mie scattering intensity distributions across a spectral range of 480 to 700 nm. The wide spectral range allows assignment of resonance mode numbers and mode orders using conventional Mie theory calculations. Accurate droplet sizing, within +/- 2 nm, is possible for water-based droplets with radii between 2 microm and 8 microm. We additionally demonstrate that the refractive index dispersion can be determined from a single refractive index value at known wavelength. Finally, morphological droplet dynamics are presented showing non-linear droplet evaporation behaviour at a temporal resolution of 100 milliseconds.

Controlling and characterizing the coagulation of liquid aerosol droplets

We demonstrate that optical tweezers can be used to control and characterize the coagulation and mixing state of aerosols. Liquid aerosol droplets of 2-14 mum in diameter are optically trapped and characterized by spontaneous and stimulated Raman scatterings, which together provide a unique signature of droplet size and composition. From the conventional bright field image, the size of the trapped droplet can be estimated and compared with that determined from stimulated Raman scattering, and the motion of the particle within the trapping plane can be recorded. A maximum of four droplets can be manipulated in tandem by forming multiple optical traps through rapid beam steering. The coagulation of two droplets can be studied directly by controlling two droplets. The limiting conditions under which optical forces and capillary forces dominate the aerosol coagulation event are explored by varying the relative optical trap strengths and characterizing the coagulation of different droplet sizes. Finally, we demonstrate that the coagulation of different aerosol components can be compared and the mixing state of the final coagulated droplet can be investigated. In particular, we compare the outcome of the coagulation of an aqueous sodium chloride aerosol droplet with a second aqueous droplet, with an ethanol droplet or with a decane droplet.

Optical sculpture: controlled deformation of emulsion droplets with ultralow interfacial tensions using optical tweezers

We report a technique for deforming micron-sized emulsion droplets that have ultralow interfacial tensions, by the manipulation of multiple optical trapping sites within the droplets.

A Strategy for Characterizing the Mixing State of Immiscible Aerosol Components and the Formation of Multiphase Aerosol Particles through Coagulation

We demonstrate that the coagulation of two aerosol droplets of different chemical composition can be studied directly through the unique combination of optical tweezers and Raman spectroscopy. Multiple optical traps can be established, allowing the manipulation of multiple aerosol droplets. Spontaneous Raman scattering allows the characterization of droplet composition and mixing state, permitting the phase segregation of immiscible components in multiphase aerosol to be investigated with spatial resolution. Stimulated Raman scattering allows the integrity of the droplet and uniformity of refractive index to be probed. The combination of these spectroscopic probes with optical tweezers is shown to yield unprecedented detail in studies of the coagulation of decane and water droplets.

Spectroscopy of Growing and Evaporating Water Droplets: Exploring the Variation in Equilibrium Droplet Size with Relative Humidity

We demonstrate that the thermodynamic properties of a single liquid aerosol droplet can be explored through the combination of a single-beam gradient force optical trap with Raman spectroscopy. A single aqueous droplet, 2-6 microm in radius, can be trapped in air indefinitely and the response of the particle to variations in relative humidity investigated. The Raman spectrum provides a unique fingerprint of droplet composition, temperature, and size. Spontaneous Raman scattering is shown to be consistent with that from a bulk phase sample, with the shape of the OH stretching band dependent on the concentration of sodium chloride in the aqueous phase and on the polarization of the scattered light. Stimulated Raman scattering at wavelengths commensurate with whispering gallery modes is demonstrated to provide a method for determining the size of the trapped droplet with nanometer precision and with a time resolution of 1 s. The polarization dependence of the stimulated scatter is consistent with the dependence observed for the spontaneous scatter from the droplet. By characterizing the spontaneous and stimulated Raman scattering from the droplet, we demonstrate that it is possible to measure the equilibrium size and composition of an aqueous droplet with variation in relative humidity. For this benchmark study we investigate the variation in equilibrium size with relative humidity for a simple binary sodium chloride/aqueous aerosol, a typical representative inorganic/aqueous aerosol that has been studied extensively in the literature. The measured equilibrium sizes are shown to be in excellent agreement with the predictions of Köhler theory. We suggest that this approach could provide an important new strategy for characterizing the thermodynamic properties and kinetics of transformation of aerosol particles.

Laser Tweezers Raman Study of Optically Trapped Aerosol Droplets of Seawater and Oleic Acid Reacting with Ozone: Implications for Cloud-Droplet Properties

In this communication we report the first use of the Raman laser tweezers technique to trap and hold a mixed droplet of oleic acid and water at atmospheric pressure for 30 min, oxidize the oleic acid on the droplet, follow the decay of reactants and the growth of chemical products using Raman spectroscopy, and monitor the growth in size of the droplet as it becomes more hydrophilic. We demonstrate that the oxidation of organic films on water droplets could have large climatic effects in the atmosphere. We show that cloud-droplet growth and activation of cloud condensation nuclei (to become cloud droplets) is retarded by the presence of an organic film and that chemical oxidation of this film would allow a cloud droplet to grow, reducing cloud albedo by inducing precipitation, and would allow a cloud condensation nucleus to grow to a cloud droplet, thus forming a cloud and increasing the albedo of the Earth.