Deep Learning in Label-free Cell Classification

Abstract

Label-free cell analysis is essential to personalized genomics, cancer diagnostics, and drug development as it avoids adverse effects of staining reagents on cellular viability and cell signaling. However, currently available label-free cell assays mostly rely only on a single feature and lack sufficient differentiation. Also, the sample size analyzed by these assays is limited due to their low throughput. Here, we integrate feature extraction and deep learning with high-throughput quantitative imaging enabled by photonic time stretch, achieving record high accuracy in label-free cell classification. Our system captures quantitative optical phase and intensity images and extracts multiple biophysical features of individual cells. These biophysical measurements form a hyperdimensional feature space in which supervised learning is performed for cell classification. We compare various learning algorithms including artificial neural network, support vector machine, logistic regression, and a novel deep learning pipeline, which adopts global optimization of receiver operating characteristics. As a validation of the enhanced sensitivity and specificity of our system, we show classification of white blood T-cells against colon cancer cells, as well as lipid accumulating algal strains for biofuel production. This system opens up a new path to data-driven phenotypic diagnosis and better understanding of the heterogeneous gene expressions in cells.

Review: imaging technologies for flow cytometry

“…For example, due to cell-to-cell and cell-to-substrate adhesion, strategies for image segmentation such as threshold, watershed and edge-detection have been under development for decades, yet still a bottleneck of the automated image analysis in microscopy. This problem is of little effect in IFC where the cells are put in suspension and interrogated on a single-cell basis 63–67 . For blood cells, bone marrow cells, and many cancer cells flowing in the blood vessels, IFC is the most promising approach to study their morphological changes.…”

Leveraging uncertainty information from deep neural networks for disease detection

“…By stacking layers of linear convolutions with appropriate non-linearities 4 , abstract concepts can be learnt from high-dimensional input alleviating the challenging and time-consuming task of hand-crafting algorithms. Such DNNs are quickly entering the field of medical imaging and diagnosis [5][6][7][8][9][10][11][12][13][14][15] , outperforming state-of-the-art methods at disease detection or allowing one to tackle problems that had previously been out of reach. Applied at scale, such systems could considerably alleviate the workload of physicians by detecting patients at risk from a prescreening examination.…”

Development of full‐field optical spatial coherence tomography system for automated identification of malaria using the multilevel ensemble classifier

“…To reduce the speckle and spatial coherence, we generate the multiple light sources by the combination of rotating diffuser and vibrating multimode fiber bundle (MMFB). To overcome the complications for assessing the high-dimensional data and minimizing the diagnostic errors, machine learning automation systems is used [13,[22][23][24][25][26]. Also, the present system is based on slightly off-axis, which need only single interferogram to extract the phase information that will be very helpful for dynamic process such as RBC [20,21].…”