Timothy P. Day, Khaled Mosharraf Mukut, Luke Klacik, Ryan O’Donnell, James Wasilewski, Somesh P. Roy
{"title":"SAGE:一种用于烟尘聚集体TEM图像中初级粒子分割的机器学习模型","authors":"Timothy P. Day, Khaled Mosharraf Mukut, Luke Klacik, Ryan O’Donnell, James Wasilewski, Somesh P. Roy","doi":"10.1016/j.proci.2025.105821","DOIUrl":null,"url":null,"abstract":"<div><div>Accurate characterization of the morphology of soot is essential for our understanding and better modeling of the physical and chemical properties of soot. The morphological characteristics of soot are traditionally explored experimentally via Transmission Electron Microscopy (TEM), usually by investigating the images via manual segmentation, which is highly labor intensive. To improve this process, a novel model for the automatic segmentation of primary particles in TEM images of soot is presented in this work. The goal of the model is to identify and isolate each primary particle from a TEM image of a soot aggregate. The model, titled Soot Aggregate Geometry Extraction (SAGE) employs a two-stage training process using a convolutional neural network: an initial training on synthetically-generated TEM images followed by a refinement training by using manually segmented real TEM images. The model was tested against a dataset of real TEM images that included images from sources different from the training data (i.e., different instruments and different researchers). When tested against this real TEM image dataset of soot, SAGE shows good performance with an F<span><math><msub><mrow></mrow><mrow><mn>1</mn></mrow></msub></math></span> score of 67.7%, indicating its ability to correctly identify primary particles while achieving a balanced trade off between missing true particles and detecting false ones. SAGE is able to detect more primary particles with better shape and size alignments with the ground truth data than traditional methods such as circular Hough transform or Euclidean distance mapping methods, leading to a much higher mean Intersection over Union score of 62.2%. Unlike most existing approaches that produce circular segmentations and require image-by-image tuning, SAGE effectively captures irregular particle boundaries without additional adjustments. The particle size distribution obtained from SAGE matches well with the ground truth. The median errors of predictions obtained from SAGE fall below 5% and 1%, respectively, for radius of gyration and fractal dimension of particles.</div></div>","PeriodicalId":408,"journal":{"name":"Proceedings of the Combustion Institute","volume":"41 ","pages":"Article 105821"},"PeriodicalIF":5.2000,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"SAGE: A machine learning model for primary particle segmentation in TEM images of soot aggregates\",\"authors\":\"Timothy P. Day, Khaled Mosharraf Mukut, Luke Klacik, Ryan O’Donnell, James Wasilewski, Somesh P. Roy\",\"doi\":\"10.1016/j.proci.2025.105821\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Accurate characterization of the morphology of soot is essential for our understanding and better modeling of the physical and chemical properties of soot. The morphological characteristics of soot are traditionally explored experimentally via Transmission Electron Microscopy (TEM), usually by investigating the images via manual segmentation, which is highly labor intensive. To improve this process, a novel model for the automatic segmentation of primary particles in TEM images of soot is presented in this work. The goal of the model is to identify and isolate each primary particle from a TEM image of a soot aggregate. The model, titled Soot Aggregate Geometry Extraction (SAGE) employs a two-stage training process using a convolutional neural network: an initial training on synthetically-generated TEM images followed by a refinement training by using manually segmented real TEM images. The model was tested against a dataset of real TEM images that included images from sources different from the training data (i.e., different instruments and different researchers). When tested against this real TEM image dataset of soot, SAGE shows good performance with an F<span><math><msub><mrow></mrow><mrow><mn>1</mn></mrow></msub></math></span> score of 67.7%, indicating its ability to correctly identify primary particles while achieving a balanced trade off between missing true particles and detecting false ones. SAGE is able to detect more primary particles with better shape and size alignments with the ground truth data than traditional methods such as circular Hough transform or Euclidean distance mapping methods, leading to a much higher mean Intersection over Union score of 62.2%. Unlike most existing approaches that produce circular segmentations and require image-by-image tuning, SAGE effectively captures irregular particle boundaries without additional adjustments. The particle size distribution obtained from SAGE matches well with the ground truth. The median errors of predictions obtained from SAGE fall below 5% and 1%, respectively, for radius of gyration and fractal dimension of particles.</div></div>\",\"PeriodicalId\":408,\"journal\":{\"name\":\"Proceedings of the Combustion Institute\",\"volume\":\"41 \",\"pages\":\"Article 105821\"},\"PeriodicalIF\":5.2000,\"publicationDate\":\"2025-01-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Proceedings of the Combustion Institute\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S1540748925000355\",\"RegionNum\":2,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ENERGY & FUELS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the Combustion Institute","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1540748925000355","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
引用次数: 0
摘要
准确表征烟灰的形态是必不可少的,我们的理解和更好的建模烟灰的物理和化学性质。传统的实验方法是通过透射电子显微镜(TEM)来研究煤烟的形态特征,通常是通过人工分割来研究图像,这是一种高度劳动密集型的方法。为了改进这一过程,本文提出了一种新的烟尘TEM图像中初级粒子的自动分割模型。该模型的目标是从煤烟集合体的TEM图像中识别和分离每个初级颗粒。该模型名为Soot Aggregate Geometry Extraction (SAGE),采用卷积神经网络的两阶段训练过程:首先对合成的TEM图像进行初始训练,然后使用手动分割的真实TEM图像进行细化训练。该模型在真实TEM图像数据集上进行了测试,该数据集包括来自不同训练数据来源的图像(即不同的仪器和不同的研究人员)。当对真实的煤烟TEM图像数据集进行测试时,SAGE显示出良好的性能,F1得分为67.7%,表明它能够正确识别初级颗粒,同时在缺失真颗粒和检测假颗粒之间实现平衡。与传统方法(如圆形霍夫变换或欧几里得距离映射方法)相比,SAGE能够检测到更多具有更好形状和尺寸与地面真实数据对齐的初级粒子,从而导致更高的平均Intersection over Union得分为62.2%。与大多数产生圆形分割并需要逐图调整的现有方法不同,SAGE有效地捕获不规则粒子边界,而无需额外调整。SAGE得到的粒度分布与地面真实值吻合较好。SAGE对颗粒的旋转半径和分形维数的预测中值误差分别在5%和1%以下。
SAGE: A machine learning model for primary particle segmentation in TEM images of soot aggregates
Accurate characterization of the morphology of soot is essential for our understanding and better modeling of the physical and chemical properties of soot. The morphological characteristics of soot are traditionally explored experimentally via Transmission Electron Microscopy (TEM), usually by investigating the images via manual segmentation, which is highly labor intensive. To improve this process, a novel model for the automatic segmentation of primary particles in TEM images of soot is presented in this work. The goal of the model is to identify and isolate each primary particle from a TEM image of a soot aggregate. The model, titled Soot Aggregate Geometry Extraction (SAGE) employs a two-stage training process using a convolutional neural network: an initial training on synthetically-generated TEM images followed by a refinement training by using manually segmented real TEM images. The model was tested against a dataset of real TEM images that included images from sources different from the training data (i.e., different instruments and different researchers). When tested against this real TEM image dataset of soot, SAGE shows good performance with an F score of 67.7%, indicating its ability to correctly identify primary particles while achieving a balanced trade off between missing true particles and detecting false ones. SAGE is able to detect more primary particles with better shape and size alignments with the ground truth data than traditional methods such as circular Hough transform or Euclidean distance mapping methods, leading to a much higher mean Intersection over Union score of 62.2%. Unlike most existing approaches that produce circular segmentations and require image-by-image tuning, SAGE effectively captures irregular particle boundaries without additional adjustments. The particle size distribution obtained from SAGE matches well with the ground truth. The median errors of predictions obtained from SAGE fall below 5% and 1%, respectively, for radius of gyration and fractal dimension of particles.
期刊介绍:
The Proceedings of the Combustion Institute contains forefront contributions in fundamentals and applications of combustion science. For more than 50 years, the Combustion Institute has served as the peak international society for dissemination of scientific and technical research in the combustion field. In addition to author submissions, the Proceedings of the Combustion Institute includes the Institute''s prestigious invited strategic and topical reviews that represent indispensable resources for emergent research in the field. All papers are subjected to rigorous peer review.
Research papers and invited topical reviews; Reaction Kinetics; Soot, PAH, and other large molecules; Diagnostics; Laminar Flames; Turbulent Flames; Heterogeneous Combustion; Spray and Droplet Combustion; Detonations, Explosions & Supersonic Combustion; Fire Research; Stationary Combustion Systems; IC Engine and Gas Turbine Combustion; New Technology Concepts
The electronic version of Proceedings of the Combustion Institute contains supplemental material such as reaction mechanisms, illustrating movies, and other data.