Datasets corresponding to "Real-time intelligent classification of COVID-19 and thrombosis via massive image-based analysis of platelet aggregates"

Datasets corresponding to "Real-time intelligent classification of COVID-19 and thrombosis via massive image-based analysis of platelet aggregates" Please find below an explanation for the <strong>files </strong>in this repository: <br> <br> <strong>DiseaseClassifPaper_Dataset_01.7z, DiseaseClassifPaper_Dataset_02.7z</strong> Experimental data. To reproduce the analyses, unzip both files and put the content into a folder called "Dataset" <strong>02_CNN_PhenotypeClassif.7z</strong> CNN Phenotype classification. Model was trained using AIDeveloper. using manually labelled data. Labelled Data is contained in folder "03_GatedData". The AIDeveloper session file in "02_Model\M10_Nitta6l_32pix_8class_meta.xlsx" shows, which files correspond to which subpopulation. The final model "M10_Nitta6l_32pix_8class_448.model" and corresponding .pb files are also located in that folder. <strong>03_ExampleMeasurement.zip</strong> One measurement file and a corresponding scatterplot <strong>04_Dataset_load.zip</strong> The python script "03_ExtractFeatures.py" loads the list of available experiment files (01_Dataset_Table_v02.csv). The experiment files are contained in DiseaseClassifPaper_Dataset_01.7z, DiseaseClassifPaper_Dataset_02.7z. The scrip then evaluates each experiment file to obtain distribution parameters for Area and Solidity. These values are written to new "01_Dataset_Table_v03.csv". <strong>05_RF_training</strong> Scripts to train and evaluate the Random Forest model (using features contained in "01_Dataset_Table_v03.csv"). <strong>07_pytranskit</strong> Scripts for training and evaluating CDT-PLDA classifier本数据集对应论文《基于血小板聚集物海量图像分析实现新冠肺炎与血栓形成的实时智能分类》。下文为本仓库中各文件的说明： **DiseaseClassifPaper_Dataset_01.7z、DiseaseClassifPaper_Dataset_02.7z**：实验数据。如需复现本研究的分析流程，请解压这两个压缩包，并将其中内容放入名为"Dataset"的文件夹中。 **02_CNN_PhenotypeClassif.7z**：表型分类卷积神经网络（Convolutional Neural Network, CNN）相关文件。本模型依托AIDeveloper平台，使用人工标注数据训练完成。标注数据存于"03_GatedData"文件夹中。"02_ModelM10_Nitta6l_32pix_8class_meta.xlsx"内的AIDeveloper会话文件可指明各文件对应的亚群。最终模型"M10_Nitta6l_32pix_8class_448.model"及其配套的.pb文件均存放于该文件夹内。 **03_ExampleMeasurement.zip**：一份测量文件及其对应的散点图。 **04_Dataset_load.zip**：Python脚本"03_ExtractFeatures.py"可加载可用实验文件列表（01_Dataset_Table_v02.csv）。实验文件均包含于DiseaseClassifPaper_Dataset_01.7z与DiseaseClassifPaper_Dataset_02.7z中。该脚本会逐一解析各实验文件，以获取面积（Area）与致密性（Solidity）的分布参数，并将计算结果写入新的"01_Dataset_Table_v03.csv"文件。 **05_RF_training**：用于训练与评估随机森林（Random Forest, RF）模型的脚本，所用特征提取自"01_Dataset_Table_v03.csv"。 **07_pytranskit**：用于训练与评估CDT-PLDA分类器的脚本。