Bohyoung Kim

ChartSense: Interactive Data Extraction from Chart Images

Charts are commonly used to present data in digital documents such as web pages, research papers, or presentation slides. When the underlying data is not available, it is necessary to extract the data from a chart image to utilize the data for further analysis or improve the chart for more accurate perception. In this paper, we present ChartSense, an interactive chart data extraction system. ChartSense first determines the chart type of a given chart image using a deep learning based classifier, and then extracts underlying data from the chart image using semi-automatic, interactive extraction algorithms optimized for each chart type. To evaluate chart type classification accuracy, we compared ChartSense with ReVision, a system with the state-of-the-art chart type classifier. We found that ChartSense was more accurate than ReVision. In addition, to evaluate data extraction performance, we conducted a user study, comparing ChartSense with WebPlotDigitizer, one of the most effective chart data extraction tools among publicly accessible ones. Our results showed that ChartSense was better than WebPlotDigitizer in terms of task completion time, error rate, and subjective preference.

miRTarVis+: Web-based Interactive Visual Analytics Tool for MicroRNA Target Predictions.

In this paper, we present miRTarVis+, a Web-based interactive visual analytics tool for miRNA target predictions and integrative analyses of multiple prediction results. Various microRNA (miRNA) target prediction algorithms have been developed to improve sequence-based miRNA target prediction by exploiting miRNA-mRNA expression profile data. There are also a few analytics tools to help researchers predict targets of miRNAs. However, there still is a need for improving the performance for miRNA prediction algorithms and more importantly for interactive visualization tools for an integrative analysis of multiple prediction results. miRTarVis+ has an intuitive interface to support the analysis pipeline of load, filter, predict, and visualize. It can predict targets of miRNA by adopting Bayesian inference and maximal information-based nonparametric exploration (MINE) analyses as well as conventional correlation and mutual information analyses. miRTarVis+ supports an integrative analysis of multiple prediction results by providing an overview of multiple prediction results and then allowing users to examine a selected miRNA-mRNA network in an interactive treemap and node-link diagram. To evaluate the effectiveness of miRTarVis+, we conducted two case studies using miRNA-mRNA expression profile data of asthma and breast cancer patients and demonstrated that miRTarVis+ helps users more comprehensively analyze targets of miRNA from miRNA-mRNA expression profile data. miRTarVis+ is available at http://hcil.snu.ac.kr/research/mirtarvisplus.

Texture Analysis of Torn Rotator Cuff on Preoperative Magnetic Resonance Arthrography as a Predictor of Postoperative Tendon Status

Objective To evaluate texture data of the torn supraspinatus tendon (SST) on preoperative T2-weighted magnetic resonance arthrography (MRA) using the gray-level co-occurrence matrix (GLCM) for prediction of post-operative tendon state. Materials and Methods Fifty patients who underwent arthroscopic rotator cuff repair for full-thickness tears of the SST were included in this retrospective study. Based on 1-year follow-up, magnetic resonance imaging showed that 30 patients had intact SSTs, and 20 had rotator cuff retears. Using GLCM, two radiologists measured independantly the highest signal intensity area of the distal end of the torn SST on preoperative T2-weighted MRA, which were compared between two groups.The relationships with other well-known prognostic factors, including age, tear size (anteroposterior dimension), retraction size (mediolateral tear length), grade of fatty degeneration of the SST and infraspinatus tendon, and arthroscopic fixation technique (single or double row), also were evaluated. Results Of all the GLCM features, the retear group showed significantly higher entropy (p < 0.001 and p = 0.001), variance (p = 0.030 and 0.011), and contrast (p = 0.033 and 0.012), but lower angular second moment (p < 0.001 and p = 0.002) and inverse difference moment (p = 0.027 and 0.027), as well as larger tear size (p = 0.001) and retraction size (p = 0.002) than the intact group. Retraction size (odds ratio [OR] = 3.053) and entropy (OR = 17.095) were significant predictors. Conclusion Texture analysis of torn SSTs on preoperative T2-weighted MRA using the GLCM may be helpful to predict postoperative tendon state after rotator cuff repair.

LongLine: Visual Analytics System for Large-scale Audit Logs

Abstract Audit logs are different from other software logs in that they record the most primitive events (i.e., system calls) in modern operating systems. Audit logs contain a detailed trace of an operating system, and thus have received great attention from security experts and system administrators. However, the complexity and size of audit logs, which increase in real time, have hindered analysts from understanding and analyzing them. In this paper, we present a novel visual analytics system, LongLine, which enables interactive visual analyses of large-scale audit logs. LongLine lowers the interpretation barrier of audit logs by employing human-understandable representations (e.g., file paths and commands) instead of abstract indicators of operating systems (e.g., file descriptors) as well as revealing the temporal patterns of the logs in a multi-scale fashion with meaningful granularity of time in mind (e.g., hourly, daily, and weekly). LongLine also streamlines comparative analysis between interesting subsets of logs, which is essential in detecting anomalous behaviors of systems. In addition, LongLine allows analysts to monitor the system state in a streaming fashion, keeping the latency between log creation and visualization less than one minute. Finally, we evaluate our system through a case study and a scenario analysis with security experts.

Comparison of strain and shear wave elastography for qualitative and quantitative assessment of breast masses in the same population

We investigated addition of strain and shear wave elastography to conventional ultrasonography for the qualitative and quantitative assessment of breast masses; cut-off points were determined for strain ratio, elasticity ratio, and visual score for differentiating between benign and malignant masses. In all, 108 masses from 94 patients were evaluated with strain and shear wave elastography and scored for suspicion of malignancy, visual score, strain ratio, and elasticity ratio. The diagnostic performance between ultrasonography alone and ultrasonography combined with either type of elastography was compared; cut-off points were determined for strain ratio, elasticity ratio, and visual score. Of the 108 masses, 44 were malignant and 64 were benign. The areas under the curves were significantly higher for strain and shear wave elastography-supplemented ultrasonography (0.839 and 0.826, respectively; P = 0.656) than for ultrasonography alone (0.764; P = 0.018 and 0.035, respectively). The diagnostic performances of strain and elasticity ratios were similar when differentiating benign from malignant masses. Cut-off values for strain ratio, elasticity ratio, and visual scores for strain and shear wave elastography were 2.93, 4, 3, and 2, respectively. Both forms of elastography similarly improved the diagnostic performance of conventional ultrasonography in the qualitative and quantitative assessment of breast masses.

Tumor heterogeneity of pancreas head cancer assessed by CT texture analysis: association with survival outcomes after curative resection

The value of image based texture features as a powerful method to predict prognosis and assist clinical management in cancer patients has been established recently. However, texture analysis using histograms and grey-level co-occurrence matrix in pancreas cancer patients has rarely been reported. We aimed to analyze the association of survival outcomes with texture features in pancreas head cancer patients. Eighty-eight pancreas head cancer patients who underwent preoperative CT images followed by curative resection were included. Texture features using different filter values were obtained. The texture features of average, contrast, correlation, and standard deviation with no filter, and fine to medium filter values as well as the presence of nodal metastasis were significantly different between the recurred (n = 70, 79.5%) and non-recurred group (n = 18, 20.5%). In the multivariate Cox regression analysis, lower standard deviation and contrast and higher correlation with lower average value representing homogenous texture were significantly associated with poorer DFS (disease free survival), along with the presence of lymph node metastasis. Texture parameters from routinely performed pre-operative CT images could be used as an independent imaging tool for predicting the prognosis in pancreas head cancer patients who underwent curative resection.

Predictors of Invasive Breast Cancer in Patients With Ductal Carcinoma In Situ in Ultrasound-Guided Core Needle Biopsy: Invasive Cancer Predictors in Core Needle Breast Biopsy

To determine predictors of invasiveness of lesions with US‐guided biopsy‐confirmed ductal carcinoma in situ (DCIS), focusing on US features, including shear wave elastography (SWE).

Interactive visual analysis of miRNA target prediction results

Integrated analysis of mRNA and miRNAs is essential to comprehend regulation of gene expression. In this paper, we present a case study with miRTarVis, a recently introduced visual analytics tool that supports effective visualizations of the miRNA target prediction results. In the case study, we evaluate the feasibility of miRTarVis by applying it to analyzing miRNA-mRNA expression profiles from TCGA (The Cancer Genome Atlas) breast cancer dataset. Our study results show that miRTarVis can be effective in confirming the previously reported miRNA-mRNA interactions, and it has potential to help researchers generate new hypotheses when it is applied to new dataset.

SwiftTuna: Responsive and incremental visual exploration of large-scale multidimensional data

For interactive exploration of large-scale data, a preprocessing scheme (e.g., data cubes) has often been used to summarize the data and provide low-latency responses. However, such a scheme suffers from a prohibitively large amount of memory footprint as more dimensions are involved in querying, and a strong prerequisite that specific data structures have to be built from the data before querying. In this paper, we present SwiftTuna, a holistic system that streamlines the visual information seeking process on large-scale multidimensional data. SwiftTuna exploits an in-memory computing engine, Apache Spark, to achieve both scalability and performance without building precomputed data structures. We also present a novel interactive visualization technique, tailed charts, to facilitate large-scale multidimensional data exploration. To support responsive querying on large-scale data, SwiftTuna leverages an incremental processing approach, providing immediate low-fidelity responses (i.e., prompt responses) as well as delayed high-fidelity responses (i.e., incremental responses). Our performance evaluation demonstrates that SwiftTuna allows data exploration of a real-world dataset with four billion records while preserving the latency between incremental responses within a few seconds.

Adaptive Zoom-based Gaze Tracking for Enhanced Accuracy and Precision

The accuracy and precision of video-based remote gaze trackers is affected by numerous factors (e.g. the head movement of the participant). However, it is challenging to control all factors that have an influence, and doing so (e.g., using a chin-rest to control geometry) could lead to losing the benefit of using gaze trackers, i.e., the ecological validity of their unobtrusive nature. We propose an adaptive zoom-based gaze tracking technique, ZoomTrack that addresses this problem by improving the resolution of the gaze tracking results. Our approach magnifies a region-of-interest (ROI) and retrieves gaze points at a higher resolution under two different zooming modes: only when the gaze reaches the ROI (temporary) or whenever a participant stares at the stimuli (omnipresent). We compared these against the base case without magnification in a user study. The results are then used to summarize the advantages and limitations of our technique.