Taeg-Keun Whangbo

Noninvasive identification of viable cell populations in docetaxel-treated breast tumors using ferritin-based magnetic resonance imaging.

Background Cancer stem cells (CSCs) are highly tumorigenic and are responsible for tumor progression and chemoresistance. Noninvasive imaging methods for the visualization of CSC populations within tumors in vivo will have a considerable impact on the development of new CSC-targeting therapeutics. Methodology/Principal Findings In this study, human breast cancer stem cells (BCSCs) transduced with dual reporter genes (human ferritin heavy chain [FTH] and enhanced green fluorescence protein [EGFP]) were transplanted into NOD/SCID mice to allow noninvasive tracking of BCSC-derived populations. No changes in the properties of the BCSCs were observed due to ferritin overexpression. Magnetic resonance imaging (MRI) revealed significantly different signal intensities (R2* values) between BCSCs and FTH-BCSCs in vitro and in vivo. In addition, distinct populations of pixels with high R2* values were detected in docetaxel-treated FTH-BCSC tumors compared with control tumors, even before the tumor sizes changed. Histological analysis revealed that areas showing high R2* values in docetaxel-treated FTH-BCSC tumors by MRI contained EGFP+/FTH+ viable cell populations with high percentages of CD44+/CD24− cells. Conclusions/Significance These findings suggest that ferritin-based MRI, which provides high spatial resolution and tissue contrast, can be used as a reliable method to identify viable cell populations derived from BCSCs after chemotherapy and may serve as a new tool to monitor the efficacy of CSC-targeting therapies in vivo.

Evidence Is Enough?: A Systematic Review and Network Meta-Analysis of the Efficacy of Tamsulosin 0.2 mg and Tamsulosin 0.4 mg as an Initial Therapeutic Dose in Asian Benign Prostatic Hyperplasia Patients

Purpose We compared the efficacy of tamsulosin between 0.2 mg and 0.4 mg in Asian prostatic hyperplasia (BPH) patients using network meta-analysis due to lack of studies with direct comparison. Methods The literature search was conducted using the MEDLINE, Embase, and Cochrane Library. Keywords used were “BPH,” “tamsulosin,” “placebo.” Experimental groups were defined as tamsulosin 0.2 mg (Tam 0.2) and 0.4 mg (Tam 0.4) and common control group was defined as placebo for indirect treatment comparison. Mixed treatment comparison was performed including one direct comparison study. Results Seven studies met the eligible criteria. Indirect treatment comparison revealed that total International Prostate Symptoms Score (IPSS) and quality of life score of IPSS were not significantly different in Tam 0.2 and Tam 0.4 (P>0.05). There was no significant difference of maximal flow rate and postvoid residual urine volume in Tam 0.2 and Tam 0.4 (P>0.05). Mixed treatment comparison including one direct comparison study showed inconsistency (P<0.001). Therefore, analysis using direct treatment comparison effect sizes of Tam 0.2 vs. placebo and Tam 0.4 vs. placebo was done and there was no significant difference. Conclusions Network meta-analysis showed no difference of efficacy between tamsulosin 0.2 mg and 0.4 mg and the evidence of tamsulosin 0.4 mg as initial dose for Asian BPH patient seems to be insufficient. Therefore, initial dose of tamsulosin for Asian BPH patient should be 0.2 mg.

Personalized Urination Activity Recognition Based on a Recurrent Neural Network Using Smart Band

Purpose Though it is very important obtaining exact data about patients’ voiding patterns for managing voiding dysfunction, actual practice is very difficult and cumbersome. In this study, data about urination time and interval measured by smart band device on patients’ wrist were collected and analyzed to resolve the clinical arguments about the efficacy of voiding diary. By developing a smart band based algorithm for recognition of complex and serial pattern of motion, this study aimed to explore the feasibility of measurement the urination time and intervals for voiding dysfunction management. Methods We designed a device capable of recognizing urination time and intervals based on specific postures of the patient and consistent changes in posture. These motion data were obtained by a smart band worn on the wrist. An algorithm that recognizes the repetitive and common 3-step behavior for urination (forward movement, urination, backward movement) was devised based on the movement and tilt angle data collected from a 3-axis accelerometer. The sequence of body movements during voiding has consistent temporal characteristics, so we used a recurrent neural network and long short-term memory based framework to analyze the sequential data and to recognize urination time. Real-time data were acquired from the smart band, and for data corresponding to a certain duration, the value of the signals was calculated and then compared with the set analysis model to calculate the time of urination. A comparative study was conducted between real voiding and device-detected voiding to assess the performance of the proposed recognition technology. Results The accuracy of the algorithm was calculated based on clinical guidelines established by urologists. The accuracy of this detecting device was high (up to 94.2%), proving the robustness of the proposed algorithm. Conclusions This urination behavior recognition technology showed high accuracy and could be applied in clinical settings to characterize patients’ voiding patterns. As wearable devices are developed and generalized, algorithms detecting consistent sequential body movement patterns reflecting specific physiologic behavior might be a new methodology for studying human physiologic behavior.

Development of Personalized Urination Recognition Technology Using Smart Bands

Purpose This study collected and analyzed activity data sensed through smart bands worn by patients in order to resolve the clinical issues posed by using voiding charts. By developing a smart band-based algorithm for recognizing urination activity in patients, this study aimed to explore the feasibility of urination monitoring systems. Methods This study aimed to develop an algorithm that recognizes urination based on a patient’s posture and changes in posture. Motion data was obtained from a smart band on the arm. An algorithm that recognizes the 3 stages of urination (forward movement, urination, backward movement) was developed based on data collected from a 3-axis accelerometer and from tilt angle data. Real-time data were acquired from the smart band, and for data corresponding to a certain duration, the absolute value of the signals was calculated and then compared with the set threshold value to determine the occurrence of vibration signals. In feature extraction, the most essential information describing each pattern was identified after analyzing the characteristics of the data. The results of the feature extraction process were sorted using a classifier to detect urination. Results An experiment was carried out to assess the performance of the recognition technology proposed in this study. The final accuracy of the algorithm was calculated based on clinical guidelines for urologists. The experiment showed a high average accuracy of 90.4%, proving the robustness of the proposed algorithm. Conclusions The proposed urination recognition technology draws on acceleration data and tilt angle data collected via a smart band; these data were then analyzed using a classifier after comparative analyses with standardized feature patterns.

A method for manipulating moving objects in panoramic image stitching

Detection of moving object and creating well-aligned image mosaics are a challenging task. In this paper, we propose a new approach for creating visually appealing and correctly stitched image panoramas in the presence of moving objects in the scene. Our main contributions are followings. The first component of our method is to find precise overlap region between images by using correctly matched feature points information after discarding outliers. Second main component is to check the existence of moving objects in the found overlap region. In order to detect moving objects in the overlapping regions, firstly, we find the absolute difference of overlap regions by subtracting overlaps, afterwards, the squares of the subtracted image pixels are calculated, thresholding operation is applied to the obtained image. Hence, median filtering is employed to remove noise from the derived binary image. Following that, we check the existence of moving object; afterwards, binary image masking is applied to the overlap region for extracting the moving object pixels or background scene pixels. Subsequently, obtained region pixels are placed on the other image using inverse hommography to make the overlap region same in both stitched images. As a result, the overlap region of images becomes same with the same object or background region for the movement of the object. Finally, we apply the blending operation to achieve seamless result. Experiments demonstrate that our proposed method is able to produce correct panorama images for moving objects and it is very effective on mobile phones in terms of speed and quality as compared to other existing methods.

Effective Fat Quantification Using Improved Least-Square Fit at High-Field MRI

In high-field magnetic resonance imaging (MRI), water-fat separation in the presence of B0 field inhomogeneity is important research. Various field map estimation techniques that use three-point multi-echo acquisitions have been developed for reliable water fat separation. Among the numerous techniques, iterative decomposition of water and fat with echo asymmetry and least squares estimation (IDEAL) has gained considerable popularity as an iterative method for acquiring high-quality water and fat images. However, due to the worsened B0 inhomogeneity at high-field, IDEAL cannot adjust for meaningful field map estimation, particularly for a large field of view. Previously, to improve the robustness of this estimation, a region-growing (RG) technique was developed to take advantage of the 2D linear extrapolation procedure through the seed point set by the median value in the target object. There are some limitations with this approach, such as the dependence on the initial seed point, such as a number, intensity, and position of the seed point. In this work, we introduce a effective method called the improved least square fit method that does not need to consider parameters related with accuracy. As a result of the proposed method, we obtained a effective fat quantification result that can be applied in high-fields, with an average water residual rate of 7.2% higher than the existing method.

How to Detect Obstacles within the Lane through Smartphone-Based Lane Recognition

This paper proposes a method of detecting obstacles existing on the lane making use of lane recognition based on smart-phone. In the proposed method, after detecting lanes by means of inverse perspective transformation, the 1st step is to detect a obstacle candidate region in terms of dispersion map using dispersion values at the interested regions set up in the detected lane, and when multiple candidate regions are detected, the 2nd step is to detect characteristic points at the interested region with a FAST corner detector and select the region that has the overlapping obstacle candidate position with the 1st step result as the obstacle. The proposed method showed good obstacle-detection performance of 80~90ms for processing 1 frame image by reducing processing region and simplifying processing process.

Improved Depth Map Generation Using Motion Vector and the Vanishing Point from a Moving Camera Monocular Image

As a clue often used to acquire 3D information from a monocular image, there is a vanishing point. This article compares the inclines among valid straight lines at searching for the vanishing point of a monocular image and calculates the proximity degree and eliminates unnecessary information in generating a node so as to elevate accuracy to estimate the vanishing point of Hough Transform and draw an initial depth-map through the vanishing point calculated. Moreover, to obtain a more accurate depth of the initial depth-map, the study uses the difference of vector values according to the distance between the camera and the object based on the motion vector information within the monocular image with a camera moving and suggests a method to calculate a more accurate depth of the monocular image.

Effective R2 Map-Based Liver Segmentation Method in an MR Image

Object recognition is usually processed based on region segmentation algorithm. Region segmentation in the IT field is carried out by computerized processing of various input information such as brightness, shape, and pattern analysis. If the information mentioned does not make sense, however, many limitations could occur with region segmentation during computer processing. Therefore, this paper suggests effective region segmentation method based on R2 information within the magnetic resonance (MR) theory. In this study, the experiment had been conducted using images including the liver region and by setting up feature points of R2 map as seed points for region growing to enable region segmentation even when the border line was not clear. As a result, an average area difference of 8.5%, which was higher than the accuracy of conventional region segmentation algorithm, was obtained.