Hematological analysis is based on assessing changes in the numbers of different blood cells and their morphological, molecular, and cytogenetic properties via a complete blood count. It is integral to diagnose and monitor a range of blood conditions and diseases, ranging from allergies and infections to different types of cancers. The conventional approach to hematology analysis requires time-consuming protocols, multiple expensive chemical reagents, complex equipment, and highly trained personnel for operation, and presents a significant burden to patients and healthcare systems. There is a need for simple, fast, low-cost alternatives such as label-free techniques that eliminate the need for staining or exogenous labels. We recently demonstrated label-free hematology analysis using deep-ultraviolet (UV) microscopy, a high-resolution imaging technique that yields quantitative molecular and structural information from biological samples. In this work, we present a fast, automated analysis pipeline to classify and count the different blood cell types in single-channel UV microscopy images using a low-cost, compact deep-UV microscope. Our previous work focused primarily on white blood cells; here, we further add platelets and red blood cells. We train a YOLOv7-style network to identify and count different blood cells in smear images acquired from a deep-UV microscopy system. Our deep-UV microscope in an LED-based, compact, and portable configuration and single-step analysis pipeline could be further combined with UV-transparent PDMS-based microfluidic devices to develop a fully automated, low-cost, label-free hematology analyzer.
Deep Ultraviolet (UV) microscopy enables high-resolution, molecular imaging and typically yields 2D images representing the axial projection of a sample’s 3D absorption onto a plane. In this work, we present a tomographic imaging approach based on multispectral UV microscopy, to visualize complex 3D structural features in samples. We aim to employ through-focus intensity images captured with varying partially coherent and asymmetric illumination patterns to extract the 3D absorption and refractive index (RI) distributions of the sample at distinct UV wavelengths. The recovery procedure relies on solving the inverse scattering problem using the 3D optical transfer function of our microscope.
Quantitative Phase Imaging (QPI) has become a mainstay imaging technique in the biomedical sciences to study cells and other biological processes. Traditional QPI techniques are transmission-based and, thus, limited to thin samples. Over the past few years, multiple 3D QPI tools have emerged attempting to overcome this limitation and provide cross-sectional phase information of thicker samples. However, most of these techniques remain transmission-based, which constrains their ability to image samples thicker than a few mean free scattering lengths. Recently, we have developed quantitative oblique back-illumination microscopy (qOBM) as an epimode technique that enables label-free quantitative phase imaging of thick samples with tomographic crosssectioning. Like in most 3D QPI instances, qOBM requires multiple captures to render a quantitative phase image. Specifically, qOBM requires four raw captures, obtained by illuminating the sample obliquely from four different directions, to reconstruct the quantitative phase. This muti-capture scheme hinders qOBM’s ability to investigate valuable fast dynamic processes, such as blood flow, as well as its usability in some in-vivo applications. Here, we present a deep-learning enabled single-capture version of qOBM that quadruples the system’s imaging speed and prevents motion artifacts. To this end, we have trained a U-Net GAN to learn the qOBM reconstruction from a single capture obtained with oblique illumination. We show the capabilities and limitations of this approach, as well as some of the novel applications that this system enables, such as in-vivo high-resolution non-invasive blood flow quantitative phase imaging.
Deep Ultraviolet (UV) Microscopy enables high-resolution, quantitative, and label-free imaging of biomolecules. Recently, we have shown that UV microscopy can be used as a tool for simple and fast hematology analysis by providing diagnostically relevant information on morphological and cytogenic properties of various blood cells. We have demonstrated the ability to classify and segment red blood cells, white blood cell subtypes, and platelets via deep learning frameworks and have applied this technique for diagnosis of blood disorders. In this work, we present a compact, low-cost deep-UV microscope capable of performing a rapid complete blood count (CBC). CBCs, one of the most commonly performed medical tests in the United States, provide critical counts of blood components used to monitor and diagnose blood disorders. This device can serve as a point-of-care alternative to modern hematology analyzers by leveraging endogenous biomolecular contrast from UV light to perform label-free hematology analysis. Here we discuss our approach of using simple, low-cost optics and hardware to enable fast and accurate imaging of blood samples. We demonstrate the capability of this system to scan and capture images of whole blood on blood smears and in custom microfluidic devices. We also show that these images can be used to segment, classify, and colorize blood cell subtypes. Lastly, we evaluate the efficacy of our stage translation system by assessing its positioning and translation accuracy.
Significance: The morphological properties and hemoglobin (Hb) content of red blood cells (RBCs) are essential biomarkers to diagnose or monitor various types of hematological disorders. Label-free mass mapping approaches enable accurate Hb quantification from individual cells, serving as promising alternatives to conventional hematology analyzers. Deep ultraviolet (UV) microscopy is one such technique that allows high-resolution, molecular imaging, and absorption-based mass mapping.
Aim: To compare UV absorption-based mass mapping at four UV wavelengths and understand variations across wavelengths and any assumptions necessary for accurate Hb quantification.
Approach: Whole blood smears are imaged with a multispectral UV microscopy system, and the RBCs’ dry masses are computed. This approach is compared to quantitative phase imaging-based mass mapping using data from an interferometric UV imaging system.
Results: Consistent Hb mass and mean corpuscular Hb values are obtained at all wavelengths, with the precision of the single-cell mass measurements being nearly identical at 220, 260, and 280 nm but slightly lower at 300 nm.
Conclusions: A full hematological analysis (including white blood cell identification and characterization, and Hb quantification) may be achieved using a single UV illumination wavelength, thereby improving the speed and cost-effectiveness.
Deep ultraviolet microscopy (UV) enables high-resolution, label-free imaging of biological samples and yields diagnostically relevant quantitative molecular and structural information. We recently demonstrated that deep UV microscopy can serve as a simple, fast, and low-cost alternative to modern hematology analyzers that assess variations in the morphological, molecular, and cytogenetic properties of blood cells to monitor and diagnose blood disorders. We also introduced a pseudocolorization scheme that uses multi-spectral UV images (acquired at three different wavelengths) to generate images whose colors accurately recapitulate those produced by conventional Giemsa staining, and can thus be used for visual hematological analysis. Here, we present a deeplearning framework to virtually stain single-channel UV images acquired at 260 nm, providing a factor of three improvement in imaging speed without sacrificing accuracy. We train a generative adversarial network (GAN) using image pairs consisting of single-channel UV images of blood smears and their corresponding pseudocolorized images to generate realistic, virtually stained images. The virtual stained images are post-processed to improve contrast and yield consistent background colors. We quantify the performance of our framework in terms of the structural similarity index (SSIM) for each color channel. Our virtual staining scheme is the first step towards a completely automated hematological analysis pipeline that includes segmentation and classification of different blood cell types to compute metrics of diagnostic value. Our method eliminates the need to acquire images at different wavelengths and could potentially lead to the development of a faster and more compact label-free, point-of-care hematology analyzer.
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.