Without any staining or labeling process, label-free cell imaging has less damage to cells and is suitable for longitudinal observation of cell behaviors. The lensless digital holographic microscopy, which directly uses photoelectric sensor chip to collect holographic images, has the advantages of low cost, non-invasive, large field of view and high resolution, and has been widely used in label-free cell imaging. However, when the classical angular spectrum method is used to recover the complex optical field in lensless digital holography, the twin image shows up and has a serious impact on image quality. To improve the quality of the reconstruction, a multi-wavelength lensless digital holography method is used for cell imaging in this paper. During numerical reconstruction, autofocus is performed firstly, and then the object light field is restored by the multi-wavelength phase retrieval (MWPR) iteration method combined with the compressive sensing (CS) strategy. The experiment results show that the reconstruction method based on multi-wavelength illumination combined with compressive sensing can effectively eliminate the influence of twin image and improve the reconstruction quality. Compared with the reconstruction results of single wavelength hologram, our method can not only obtain highresolution labeled cell images with large field of view, but also obtain high-quality non-labeled cell video, which can be used as an effective technique for cell imaging and dynamic continuous observation in the future.
KEYWORDS: Fluorescence, Autofluorescence, Multispectral imaging, Image segmentation, Matrices, Fluorophores, Fluorescence imaging, Detection and tracking algorithms, Signal to noise ratio, Image processing algorithms and systems
Multispectral imaging is becoming a key technique for biomedical research, but the crosstalk between autofluorescence and fluorescent material severely affects the interpretation of fluorescence images. Spectral unmixing is an effective technique for removing autofluorescence and separating fluorescent targets in multispectral fluorescence imaging. However, the effectiveness of most methods of spectral unmixing has a strong relationship with the noise in the image. In this work, we propose a multispectral fluorescence unmixing method based on a priori information to obtain the pure spectra and their corresponding abundance coefficients in the images. First, the obtained multispectral image is segmented into several superpixels using a superpixel segmentation method, and then the relative pure spectra are extracted using a spectral extraction algorithm on the superpixels. Since the autofluorescence distribution is spread over the whole body, the extracted spectra in which the autofluorescence can be considered as pure spectra are used as a priori knowledge for unmixing. The pixel spectral data that are similar to the set of relatively pure spectra are selected as the pure spectral candidate set. Then the pure spectra can be obtained using the Non-negative matrix factorization method with prior knowledge(NMF-upk). Finally, the abundance corresponding to each spectral feature can be obtained through the least square method. The proposed unmixing method is tested on simulated data and the results show that our unmixing algorithm outperforms other methods.
Avoiding adverse effects of staining reagents on cellular viability and cell signaling, label-free cell imaging and analysis is essential to personalized genomics, drug development, and cancer diagnostics. By analyzing the images of cells, imagebased cell analytic methodologies offer a relatively simple and economical way to understand the cell heterogeneities and developments. Owing to the developments in high-resolution image sensors and high-performance computation processors, the emerging lens-less digital holography techniques enable a simple and cost-effective approach to obtain label-free cell images with large field of view and microscopic spatial resolution. In this work, the lens-less digital holography technique is adopted for image-based cell analysis. The holograms of three kinds of cells which are MDA-MB231, EC-109 and MCF-10A respectively were recorded by a lens-less digital holography system composed of a laser diode, a sample holder, a sensor and a laptop computer. The acquired holograms are first high-pass filtered. Then the amplitude images were reconstructed using the angular spectrum method and the sample to sensor distance was determined using the autofocusing criteria based on the sparsity of image edges and corner points. The convolutional neural network (CNN) was used to classify the cells. The experiments show that an accuracy of 97.2% can be achieve for two type cell classification and 91.2% for three type cell classification. It is believed that the lens-less holography combining with machine learning holds great promise in the application of stainless cell imaging and classification.
Raman tomography can provide quantitative distribution of chemicals in a three-dimensional volume with a non-invasive and label-free manner. In view of the problems of existing data collection strategy, a frequency modulation and spatial encoding based Raman tomography was proposed, which aims to improve the data collection scheme and reduce the data collection time. In this scheme, the laser beam was divided into several sub-beams to use as multipoint excitation light sources. These sub-beams were first modulated with different frequencies and then incident on the different points of sample surface simultaneously. Because the excited Raman signals would carry such modulation information, the Raman signals from which excitation position can be distinguished with the demodulation process. In detection end, the Raman scattering light first passed through a spatial-encoding mask and then was directed to the single photomultiplier tube. By changing the pattern of the mask and then performing recovery with sparse reconstruction, the distribution of the Raman signals on the sample surface can be obtained based on compressive sensing theory. Preliminary results showed that our scheme can recover the Raman images to the certain extent with a better signal-to-noise ratio, demonstrating the proposed scheme is feasible.
Raman spectroscopic imaging can provide three-dimensional data set of samples, including two-dimensional spatial image and one-dimensional Raman spectral data. Currently, three strategies can be used to achieve Raman spectroscopic imaging, including point scanning, line scanning, and wide-field illumination. Point scanning method provides the best resolution but has low imaging speed. On the contrary, wide-field illumination can image fast but provides lower spatial resolution. To integrate the advantages of two methods, a new strategy for large-field Raman spectroscopic imaging was proposed, which uses the frequency modulation based spatially encoded light as the excitation. In this method, millions of single beams simultaneously illuminate on the sample to act as the wide-field illumination. Each beam illuminates on different positions of the sample, whose intensity are modulated with different frequencies. Thus, each excitation beam has its own modulation frequency and the excited Raman signal will carry the modulation information. At the detection end, a single point detector was used to collect the time series Raman signals carrying the unique modulation information. Using the sparse reconstruction based on demodulation strategy, the Raman image can be recovered effectively. The feasibility of the method was verified with numerical simulations. The results showed that it is feasible to conduct Raman spectroscopic imaging with high-resolution and high speed under the illumination of frequency modulation based spatially encoded light and the detection of single-point detector.
Bioluminescence tomography (BLT) is a promising optical imaging tool broadly used in preclinical research to observe and quantify the distribution of bioluminescent markers in small animal models. However, due to the highly scattering property of the biological tissues and the limited surface measurements, fast and precise reconstruction in BLT remains a challenging problem. Permissible source region is a cost-effective strategy to partially solve the problem. In this paper, we present a matched filtering based strategy to extract the permissible region (PSR) adaptively for bioluminescence tomography. First, a digital matched filter is formulated according to the forward weight matrix, then the surface measurements are filtered and the permissible source region is extracted according to the first several biggest outputs of the matched filter larger than a threshold value, and finally the bioluminescent source in the permissible source region is recovered. Numerical simulation experiments are performed to evaluate the performance of the proposed method. The results show that the number of unknowns can be significantly reduced even using a small threshold value and the BLT reconstruction quality can be improved with appropriate PSR.
Cerenkov fluorescence imaging (CLI) has set a bridge between optical and nuclear imaging technologies by using an optical method to detect the distribution of radiotracers. Combining the emerged CLI technique with a clinical endoscope, the Cerenkov luminescence endoscope (CLE) was developed to avoid the problem of the poor penetration depth of the Cerenkov light. However, due to low energy of the Cerenkov light and the transportation loss during endoscopic imaging, the acquisition time of CLE signal is long and the imaging results are poor, which has limited the clinical applications of CLE. There are two ways to improve the availability of the current CLE system. First is to enhance the emitted signals of the Cerenkov light at the source end by developing new kinds of imaging probes or selecting high yield radionuclides. However, this will introduce the in vivo unfriendly problem in clinical translations. The second method is to improve the detection sensitivity of CLE system by optimizing the structure of the system. Here, we customized four endoscopes with different field of view (FOV) angles of endoscope probe and different monofilament diameters of imaging fiber bundles. By comparing the results obtained by different CLE systems, we optimized the parameters of system. The CLE imaging of 18F-FDG showed that when the distance between the probe and radionuclide source was fixed, smaller angle of FOV and lager monofilament diameter will provide higher collection efficiency.
Optical projection tomography(OPT) provides an approach to recreating three-dimensional images of small biological specimens. Light traverses through a straight line to achieve a homogeneous illumination of the specimen. As the specimens in the conventional OPT could not survive or the survival time was too short, this paper proposes a new type of sample fixation method for OPT imaging. The specimen was anaesthetized in a petri dish, and the dish was fixed under the rotational stage of our homemade OPT system for imaging. This method can reduce the damage to the specimen and be more conducive to the continuous observation for in vivo OPT. However, the sample fixation causes the problem of insufficient sampling. To obtain optical projection tomographic image with insufficient samples, this paper uses the iterative reconstruction algorithm combining with the prior information to solve the inverse reconstruction problem.
Fluorescence molecular tomography (FMT) is an important imaging technique of optical imaging. The major challenge of the reconstruction method for FMT is the ill-posed and underdetermined nature of the inverse problem. In past years, various regularization methods have been employed for fluorescence target reconstruction. A comparative study between the reconstruction algorithms based on l 1 -norm and l 2 -norm for two imaging models of FMT is presented. The first imaging model is adopted by most researchers, where the fluorescent target is of small size to mimic small tissue with fluorescent substance, as demonstrated by the early detection of a tumor. The second model is the reconstruction of distribution of the fluorescent substance in organs, which is essential to drug pharmacokinetics. Apart from numerical experiments, in vivo experiments were conducted on a dual-modality FMT/micro-computed tomography imaging system. The experimental results indicated that l 1 -norm regularization is more suitable for reconstructing the small fluorescent target, while l 2 -norm regularization performs better for the reconstruction of the distribution of fluorescent substance.
As one of molecular imaging, bioluminescence tomography (BLT) aims to recover internal source from surface
measurement. Being an ill-posed inverse problem, BLT source reconstruction is usually converted to an optimization
problem through regularization. In this contribution, we build a bimodal hybrid imaging system consisting of BLT and
micro-CT, and then propose an improved source reconstruction method based on adjoint diffusion equations (ADEs).
Compared with conventional methods based on constrained minimization problem (CMP), ADEs-based method replaces
expensive iterative computation with solving a group of linear ADEs. Given surface flux density, internal source power
density and photon fluence rate can be efficiently determined in one step. Both numerical and physical experiments are
performed to evaluate the bimodal BLT/micro-CT imaging system and this novel reconstruction method. The relevant
results demonstrate the feasibility and potential of this source reconstruction method.
Gastric cancer is the second cause of cancer-related death in the world, and it remains difficult to cure because it has
been in late-stage once that is found. Early gastric cancer detection becomes an effective approach to decrease the gastric
cancer mortality. Bioluminescence tomography (BLT) has been applied to detect early liver cancer and prostate cancer
metastasis. However, the gastric cancer commonly originates from the gastric mucosa and grows outwards. The
bioluminescent light will pass through a non-scattering region constructed by gastric pouch when it transports in tissues.
Thus, the current BLT reconstruction algorithms based on the approximation model of radiative transfer equation are not
optimal to handle this problem. To address the gastric cancer specific problem, this paper presents a novel reconstruction
algorithm that uses a hybrid light transport model to describe the bioluminescent light propagation in tissues. The
radiosity theory integrated with the diffusion equation to form the hybrid light transport model is utilized to describe
light propagation in the non-scattering region. After the finite element discretization, the hybrid light transport model is
converted into a minimization problem which fuses an l1 norm based regularization term to reveal the sparsity of
bioluminescent source distribution. The performance of the reconstruction algorithm is first demonstrated with a digital
mouse based simulation with the reconstruction error less than 1mm. An in situ gastric cancer-bearing nude mouse based
experiment is then conducted. The primary result reveals the ability of the novel BLT reconstruction algorithm in early
gastric cancer detection.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.