Browsing by Author "Wang, Xi"

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    Discovering Relationships between Reservoir Properties and Production Data for CHOPS Using Data Mining Methods
    (2016-01-15) Wang, Xi; Mahinpey, Nader; Wang, Xin; Dong, Mingzhe; Chen, Shengnan (Nancy)
    Cold Heavy Oil Production with Sand (CHOPS) produces sand, and greatly contributes to primary oil recovery. It’s generally believed that wormholes, resulting from sand flow, enhance oil recovery in this process. However, due to complexity and variability, it’s difficult for wormhole models to precisely describe how wormholes develop within the formation. In this study, we regard wormholes as an integral black box. We apply data mining methods to explore how the reservoir attributes influence the CHOPS wells production. Gain ratio is used to rank and select the most important attributes for oil production. For overall oil production performance, cumulative porosity, cumulative oil saturation, effective thickness, and average shale content are the most important and relevant attributes. Decision trees constructed by C4.5 algorithm provide details of how to classify oil production instances according to reservoir attributes. All the correctly classified rates are over 55%, which is reliable accuracy in our results.
  • Thumbnail Image
    ItemOpen Access
    Sketch-Based Editing and Deformation of Cardiac Image Segmentation
    (Spinger, 2022-07-05) Wang, Xi; Ang, Kathleen; Samavati, Faramarz
    Image segmentation plays an important role in medical image analysis and diagnosis. Automatic segmentation techniques have been developed in recent years for a variety of imaging modalities and problems. Despite the robustness of these automatic segmentation algorithms, segmentation errors are still inevitable in some cases, and manual editing is needed to correct the results. Therefore, it is essential to develop efficient and effective segmentation editing tools. In this paper, we present a sketch-based editing method for cardiac MRI segmentation using subdivision surface deformation. We evaluate our method on real and synthetic datasets. The results show that with a few user interactions, the quality of the segmentation can be improved.