MIML: Multiplex Image Machine Learning for High Precision Cell Classification via Mechanical Traits within Microfluidic Systems DOI Creative Commons
Khayrul Islam, Ratul Paul, Shen Wang

et al.

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

Label-free cell classification is advantageous for supplying pristine cells further use or examination, yet existing techniques frequently fall short in terms of specificity and speed. In this study, we address these limitations through the development a novel machine learning framework, Multiplex Image Machine Learning (MIML). This architecture uniquely combines label-free images with biomechanical property data, harnessing vast, often underutilized morphological information intrinsic to each cell. By integrating both types our model offers more holistic understanding cellular properties, utilizing typically discarded traditional models. approach has led remarkable 98.3\% accuracy classification, substantial improvement over models that only consider single data type. MIML been proven effective classifying white blood tumor cells, potential broader application due its inherent flexibility transfer capability. It's particularly similar morphology but distinct properties. innovative significant implications across various fields, from advancing disease diagnostics behavior.

Language: Английский

Sensing area-variable electrode for wide-range droplet size detection DOI
Jaewook Ryu, Ki-Ho Han

Sensors and Actuators B Chemical, Journal Year: 2024, Volume and Issue: 417, P. 136121 - 136121

Published: June 25, 2024

Language: Английский

Citations

1
Fabricating a low-temperature synthesized graphene-cellulose acetate-sodium alginate scaffold for the generation of ovarian cancer spheriod and its drug assessment DOI Creative Commons

Pooja Suryavanshi,

Yohaan Kudtarkar,

Mangesh Chaudhari

et al.

Nanoscale Advances, Journal Year: 2023, Volume and Issue: 5(18), P. 5045 - 5053

Published: Jan. 1, 2023

3D cell culture can mimic tumor pathophysiology, which reflects cellular morphology and heterogeneity, strongly influencing gene expression, behavior, intracellular signaling. It supports cell-cell cell-matrix interaction, attachment, proliferation, resulting in rapid reliable drug screening models. We have generated an ovarian cancer spheroid interconnected porous scaffolds. The scaffold is fabricated using low-temperature synthesized graphene, cellulose acetate, sodium alginate. Graphene nanosheets enhance proliferation aggregation, aids the formation of spheroids. spheroids are assessed after day 7 14 for generation reactive oxygen species (ROS), expression hypoxia inducing factor (HIF-1⍺) vascular endothelial growth (VEGF). Production ROS was observed due to aggregated mass, enhanced production HIF-1⍺ VEGF results from a lack nutrition. Furthermore, efficacy anticancer doxorubicin at varying concentrations on by studying caspase-3/7 14. current findings imply that graphene-cellulose-alginate (GCA) generates model test drug.

Language: Английский

Citations

3
Reinforcement Learning-Optimized Personalized Cancer Treatment Strategies: A Case Study of Lung Cancer DOI Creative Commons
Chi-Chun Zhou, Zhaocong Liu, Xinhui Li

et al.

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 17, 2024

Abstract Personalized cancer treatment strategies (PCTS) tailor treatments on the basis of a patient’s health status, type, and stage. By considering evolving interactions options over time, PCTS seeks to balance suppression with minimizing harm maximizing therapeutic benefits. However, limited clinical trial resources limit ability explore optimal PCTSs fully through experimentation, presenting significant challenge their development. In this study, we introduce "digital twin" model that integrates comprehensive patient data, characteristics, individual responses employs reinforcement learning (RL) identify PCTS. Using lung as case calibrated parameters for various demographic groups, stages, options, utilizing real data from SEER dataset. The RL-optimized significantly outperformed traditional clinician decisions, leading notable improvements in survival. For example, among women aged 45--64 years stage IIIA, IIIB, IVA, IVB cancer, survival increased by 46%, 59%, 23%, 149%, respectively. Similarly, men years, improved 108%, 97%, 40%, 62%, respectively, across same stages. This study lays critical foundation use AI optimizing paves way further research applications.

Language: Английский

Citations

0
A Demethylation-Switchable Aptamer Design Enables Lag-Free Monitoring of m6A Demethylase FTO with Energy Self-Sufficient and Structurally Integrated Features DOI

Yakun Shi,

Yutian Lei, Meng Chen

et al.

Journal of the American Chemical Society, Journal Year: 2024, Volume and Issue: 146(50), P. 34638 - 34650

Published: Dec. 4, 2024

Cellular context profiling of modification effector proteins is critical for an in-depth understanding their biological roles in RNA N6-methyladenosine (m6A) regulation and function. However, challenges still remain due to the high complexities, which call a versatile toolbox accurate live-cell monitoring effectors. Here, we propose demethylation-switchable aptamer sensor engineered with site-specific m6A (DSA-m6A) lag-free demethylase FTO activity living cells. As proof concept, DNA against adenosine triphosphate (ATP) selected construct DSA-m6A model, as "universal energy currency" role ATP could guarantee equally fast spontaneous conformation change upon demethylation binding organisms, thus enabling sensitive neither time delay nor recourse extra supply substances. This ATP-driven design facilitates biomedical research, including imaging, inhibitor screening, single-cell tracking dynamic nuclear translocation starvation stimuli, characterization biomimetic heterotypic three-dimensional (3D) multicellular spheroid well first report on vivo imaging activity. strategy provides simple yet clinical diagnosis, drug discovery, therapeutic evaluation, study demethylation.

Language: Английский

Citations

0
MIML: Multiplex Image Machine Learning for High Precision Cell Classification via Mechanical Traits within Microfluidic Systems DOI Creative Commons
Khayrul Islam, Ratul Paul, Shen Wang

et al.

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

Label-free cell classification is advantageous for supplying pristine cells further use or examination, yet existing techniques frequently fall short in terms of specificity and speed. In this study, we address these limitations through the development a novel machine learning framework, Multiplex Image Machine Learning (MIML). This architecture uniquely combines label-free images with biomechanical property data, harnessing vast, often underutilized morphological information intrinsic to each cell. By integrating both types our model offers more holistic understanding cellular properties, utilizing typically discarded traditional models. approach has led remarkable 98.3\% accuracy classification, substantial improvement over models that only consider single data type. MIML been proven effective classifying white blood tumor cells, potential broader application due its inherent flexibility transfer capability. It's particularly similar morphology but distinct properties. innovative significant implications across various fields, from advancing disease diagnostics behavior.

Language: Английский

Citations

0