Morphological profiling by high-throughput single-cell biophysical fractometry

Abstract

Complex and irregular cell architecture is known to statistically exhibit fractal geometry, i.e., a pattern resembles a smaller part of itself. Although fractal variations in cells are proven to be closely associated with the disease-related phenotypes that are otherwise obscured in the standard cell-based assays, fractal analysis with single-cell precision remains largely unexplored. To close this gap, here we develop an image-based approach that quantifies a multitude of single-cell biophysical fractal-related properties at subcellular resolution. Taking together with its high-throughput single-cell imaging performance (~10,000 cells/sec), this technique, termed single-cell biophysical fractometry, offers sufficient statistical power for delineating the cellular heterogeneity, in the context of lung-cancer cell subtype classification, drug response assays and cell-cycle progression tracking. Further correlative fractal analysis shows that single-cell biophysical fractometry can enrich the standard morphological profiling depth and spearhead systematic fractal analysis of how cell morphology encodes cellular health and pathological conditions.

Information-Distilled Generative Label-Free Morphological Profiling Encodes Cellular Heterogeneity

“…As the no-CytoMAD-profile is derived from the input image and subsequently utilized in reconstructing the output image, it effectively encodes information from both BF and QPI. Here single cell images are captured by a multi-modal imaging flow cytometer, called multi-ATOM [2, 4, 7, 27], operating at a high imaging throughput of ∼10,000’s cells/sec (See the system details in Methods ).…”

“…Specifically, we correlated the CytoMAD-profile with the hand-crafted biophysical phenotypes of cells from the CytoMAD output QPI (i.e., CytoMAD-images) and input BF images. These hand-crafted biophysical phenotypes of cells (a total of 84) features (Supplementary Note S1) were extracted based on a hierarchical morphological feature extraction approach [4, 7], which has recently shown promises in label-free single-cell morphological profiling. We identified the important CytoMAD-profile in classifying seven lung cancer cell lines based on feature importance and correlated them with hand-crafted biophysical phenotypes, which are categorized into 3 groups (i.e., bulk, global texture and local texture of biophysical morphology) ( Supplementary Figure S2a ).…”

“…dry mass density, attenuation density) and local phenotypes (e.g. BF entropy, phase entropy) [4, 7]. We thus could quantify the mean values of each biophysical feature within each batch of samples.…”

“…Recent breakthroughs in label-free single-cell imaging have highlighted the significance of biophysical cytometry in understanding functional cellular heterogeneity in complex biological systems [1]. Supercharged by their increasing throughput and content, systematic profiling of biophysical cell morphology, which was once inconceivable, is now feasible to probe subtle differences in cell mass, shape, size, and biophysical/mechanical properties across different cell types/states or responses to various chemical and genetic perturbations [2][3][4][5][6][7][8]. These imaging technologies are gaining traction in pharmaceutical industries and life science laboratories, providing critical mechanistic insights into cellular functions that may be concealed in molecular assays [9,10].…”

“…As the no-CytoMADprofile is derived from the input image and subsequently utilized in reconstructing the output image, it effectively encodes information from both BF and QPI. Here single cell images are captured by a multi-modal imaging flow cytometer, called multi-ATOM[2,4,7,27], operating at a high imaging throughput of ~10,000's cells/sec (See the system details in Methods).CytoMAD distinguishes itselffrom the common generative model by integrating a selfsupervised morphology distillator. This component consists of a set of classification networks, including batch classifiers and cell type/state classifiers (Figure 1c).…”