| 研究実績の概要 |
Blood, a naturally multi-component bio-fluid, is commonly employed in pathology for various diagnostic purposes. It contains cellular components (red blood cells, white blood cells, and platelets), proteins, ions, and other coagulant agents. The blood can be separated into plasma (the fluid without cellular components) and serum (the plasma fluid without fibrinogen). Current clinical quantification methods are expensive. Recently, there have been considerable developments in infection detection methods using paper-based microfluidics; however, these have severe reproducibility and stability problems. In addition, current methods involve biochemical analysis requiring specific reagents that hinder application as a POC setup at home. To solve these problems, we propose a drying droplet method that characterizes the physical features of the drying pattern of bio-fluid droplets (blood, plasma, serum, and urine), different microscopy techniques, followed by image quantification and implementing machine learning (ML) techniques on the acquired time series data. This allows biofluid samples to be classified according to healthy/unhealthy states and disease stages. Our study investigates unhealthy samples, such as people with Type II diabetes who have low, medium, and high sugar levels, and we want to find out why their patterns change because of this drying process.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
We are investigating the drying evolution of biofluids (blood, plasma, serum, and urine), for healthy and diabetic patients. We quantified the biophysical composition of these bio-fluids by measuring the rheology and surface tension measurements before drying. The dried residue patterns are quantified at different length scales using profilometry, measuring the relation of thickness/height to mechanical stress and crack formation and establishing scaling laws for stress release through cracks or its suppression. SEM (Scanning Electron Microscopy) is used to provide micro-structural information about component aggregation and deposition of the components within the droplet. The drying evolution is monitored using optical microscopy, and this is captured via sequential time-dependent images. Machine learning (ML) is also implemented for the entire drying process, serving as a unique fingerprint for a pattern recognition tool. This includes two approaches: Approach 1 involves convolutional neural network (CNN) based strategies that depend on images directly, while Approach 2 incorporates MLs which use numerical data as a feature vector quantified from images from different techniques during the drying process, demonstrating the robustness of this bio-physical drying droplet method, to be used as a direct, rapid, and accurate screening method for detecting any abnormalities in these bio-fluids.
|
| 今後の研究の推進方策 |
To understand pattern formation in drying bio-mimetic fluids, control samples will be prepared that mimic (i) healthy and (ii) diabetic bio-fluids to investigate how different blood components affect the evaporation-induced flow fields that lead to dynamic pattern formation. The internal flows that occur during drying and the final residue patterns will be monitored while systematically varying the composition of these control samples. Diabetic biofluid will be prepared by adding different glucose levels, mimicking the different diabetic stages. We will also use the fluorescently labeled glucose so their redistribution within the droplet can be tracked quantitatively during the drying process via epifluorescence microscopy. The initial concentration of the sugar will be varied to induce a specific level of RBC aggregation and its effects in the blood; in contrast, the addition of sugar will give us insight into the fluid, such as plasma and serum by mapping the visco-elastic to elastic transition. In blood, these will provide insight into the flow fields within the droplets and the associated redistribution of RBCs at the initial, middle, and final stages of the drying evolution demonstrating how concentration and aggregation of RBCs in bio-mimetic fluids influence morphological pattern formation. Furthermore, the quantitative measurements will give us more information about this pattern during the drying process.
|
