• Journal of Biomedical Engineering Research
• /
• v.44 no.2
• /
• pp.139-146
• /
• 2023

Osteoporosis is a disease in which the risk of bone fractures increases due to a decrease in bone density caused by aging. Osteoporosis is diagnosed by measuring bone density in the total hip, femoral neck, and lumbar spine. To accurately measure bone density in the lumbar spine, the vertebral region must be segmented from the lumbar X-ray image. Deep learning-based automatic spinal segmentation methods can provide fast and precise information about the vertebral region. In this study, we used 695 lumbar spine images as training and test datasets for a deep learning segmentation model. We proposed a lumbar automatic segmentation model, CM-Net, which combines the center point of the spine and the modified U-Net network. As a result, the average Dice Similarity Coefficient(DSC) was 0.974, precision was 0.916, recall was 0.906, accuracy was 0.998, and Area under the Precision-Recall Curve (AUPRC) was 0.912. This study demonstrates a high-performance automatic segmentation model for lumbar X-ray images, which overcomes noise such as spinal fractures and implants. Furthermore, we can perform accurate measurement of bone density on lumbar X-ray images using an automatic segmentation methodology for the spine, which can prevent the risk of compression fractures at an early stage and improve the accuracy and efficiency of osteoporosis diagnosis.

• Journal of Korea Multimedia Society
• /
• v.7 no.4
• /
• pp.579-591
• /
• 2004

This paper presents a new arc segmentation method which extracts curves from simple drawing consisted of straight lines and curves and segments them into circular arcs. First, it finds center points and finds line segments and curve segments by tracing connected center points. Next, it expands the segment by searching neighbor segment at the two endpoints. Next, it removes straight lines and segments the extracted curves into circular arcs by using the recursive subdivision method. The proposed method has been compared with previous vectorization software and vector based arc segmentation method. Experimental results show that the proposed method produces more correct results for the curves which contain intersection with other lines or curves.

• Journal of Biomedical Engineering Research
• /
• v.44 no.5
• /
• pp.354-361
• /
• 2023

Most of spinal diseases are diagnosed based on the subjective judgment of a specialist, so numerous studies have been conducted to find objectivity by automating the diagnosis process using deep learning. In this paper, we propose a method that combines segmentation and feature extraction, which are frequently used techniques for diagnosing spinal diseases. Four models, U-Net, U-Net++, DeepLabv3+, and M-Net were trained and compared using 1000 X-ray images, and key-points were derived using Douglas-Peucker algorithms. For evaluation, Dice Similarity Coefficient(DSC), Intersection over Union(IoU), precision, recall, and area under precision-recall curve evaluation metrics were used and U-Net++ showed the best performance in all metrics with an average DSC of 0.9724. For the average Euclidean distance between estimated key-points and ground truth, U-Net was the best, followed by U-Net++. However the difference in average distance was about 0.1 pixels, which is not significant. The results suggest that it is possible to extract key-points based on segmentation and that it can be used to accurately diagnose various spinal diseases, including spondylolisthesis, with consistent criteria.

• Investigative Magnetic Resonance Imaging
• /
• v.21 no.1
• /
• pp.9-19
• /
• 2017

Background: Normalized cerebral blood volume (nCBV) can be measured using manual or semiautomatic segmentation method. However, the difference in diagnostic performance on brain tumor differentiation between differently measured nCBV has not been evaluated. Purpose: To compare the diagnostic performance of manually obtained nCBV to that of semiautomatically obtained nCBV on glioblastoma (GBM) and primary central nervous system lymphoma (PCNSL) differentiation. Materials and Methods: Histopathologically confirmed forty GBM and eleven PCNSL patients underwent 3T MR imaging with dynamic susceptibility contrast-enhanced perfusion MR imaging before any treatment or biopsy. Based on the contrast-enhanced T1-weighted imaging, the mean nCBV (mCBV) was measured using the manual method (manual mCBV), random regions of interest (ROIs) placement by the observer, or the semiautomatic segmentation method (semiautomatic mCBV). The volume of enhancing portion of the tumor was also measured during semiautomatic segmentation process. T-test, ROC curve analysis, Fisher's exact test and multivariate regression analysis were performed to compare the value and evaluate the diagnostic performance of each parameter. Results: GBM showed a higher enhancing volume (P = 0.0307), a higher manual mCBV (P = 0.018) and a higher semiautomatic mCBV (P = 0.0111) than that of the PCNSL. Semiautomatic mCBV had the highest value (0.815) for the area under the curve (AUC), however, the AUCs of the three parameters were not significantly different from each other. The semiautomatic mCBV was the best independent predictor for the GBM and PCNSL differential diagnosis according to the stepwise multiple regression analysis. Conclusion: We found that the semiautomatic mCBV could be a better predictor than the manual mCBV for the GBM and PCNSL differentiation. We believe that the semiautomatic segmentation method can contribute to the advancement of perfusion based brain tumor evaluation.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.2
• /
• pp.542-551
• /
• 2000

In this paper, a novel technique is presented for automatic brain region segmentation in single channel MR image data sets for 3D visualization and analysis. The method detects brain contours in 2D and 3D processing of four steps. The first and the second make a head mask and an initial brain mask by automatic thresholding using a curve fitting technique. The stage 3 reconstructs 3D volume of the initial brain mask by cubic interpolation and generates an intermediate brain mask using morphological operation and labeling of connected components. In the final step, the brain mask is refined by automatic thresholding using curve fitting. This algorithm is useful for fully automatic brain region segmentation of T1-weighted, T2-weighted, PD-weighted, SPGR MRI data sets without considering slice direction and covering a whole volume of a brain. In the experiments, the algorithm was applied to 20 sets of MR images and showed over 0.97 in comparison with manual drawing in similarity index.

• 제어로봇시스템학회:학술대회논문집
• /
• 1996.10a
• /
• pp.430-433
• /
• 1996

The shape of an object plays a very important role in pattern analysis and classification. Roughly, the researches on this topic can be classified into three fields, i.e. (i) edge detection, (ii) dominant points extraction, and (iii) shape recognition and classification. Many works have been done in these three fields. However, it is very seldom to see the research that discusses the connection relationship of objects. This problem is very important in robot assembly systems. Therefore, here we focus on this problem and discuss how to recover the connection relationship of planar objects. Our method is based on the partial curve identification algorithm. The experiment results show the efficiency and validity of this method.

• Journal of IKEEE
• /
• v.11 no.4
• /
• pp.307-314
• /
• 2007

In this paper, we propose an image contrast enhancement algorithm for an LCD TV. The proposed algorithm consists of two processes: the image segmentation process and the tone curve control process. The first process uses an automatic threshold technique to decompose an input image into two regions and then utilizes a hierarchical structure for real-time processing. The second process generates a gray level tone curve for contrast enhancement using a weighted sum of average tone curves for two segmented regions. Experimental result shows that the proposed algorithm outperforms the conventional contrast enhancement methods for an LCD TV.

• Journal of the Institute of Convergence Signal Processing
• /
• v.1 no.1
• /
• pp.23-31
• /
• 2000

Iterative curve evolution techniques are powerful methods for image segmentation. Classical methods proposed curve evolutions which guarantee close contours at convergence and, combined with the level set method, they easily handled curve topology changes. In this paper, we present a new geometric active contour model based on level set methods introduced by Osher & Sethian for detection of object boundaries or shape and we adopt anisotropic diffusion filtering method for removing noise from original image. Classical methods allow only one-way curve evolutions : shrinking or expanding of the curve. Thus, the initial curve must encircle all the objects to be segmented or several curves must be used, each one totally inside one object. But our method allows a two-way curve evolution : parts of the curve evolve in the outward direction while others evolve in the inward direction. It offers much more freedom in the initial curve position than with a classical geodesic search method. Our algorithm performs accurate and precise segmentations from noisy images with complex objects(jncluding sharp angles, deep concavities or holes), Besides it easily handled curve topology changes. In order to minimize the processing time, we use the narrow band method which allows us to perform calculations in the neighborhood of the contour and not in the whole image.

• Journal of the Korea Society of Computer and Information
• /
• v.23 no.4
• /
• pp.57-64
• /
• 2018

In this paper, we propose the automatic detection method for individual region separation using panorama image. Finding areas that contain individual teeth is one of the most important tasks in automating 3D models through individual tooth separation. In the conventional method, the maxillary and mandibular teeth regions are separated using a straight line or a specific CT slide, and the tooth regions are separated using a straight line in the vertical direction. In the conventional method, since the teeth are arranged in a curved shape, there is a problem that each tooth region is incorrectly detected in order to generate an accurate tooth region. This is a major obstacle to automating the creation of individual tooth models. In this study, we propose a method to find the correct tooth curve by using the jawbone curve which is very similar to the tooth curve in order to overcome the problem of finding the area containing the existing tooth. We have proposed a new method to accurately set individual tooth regions using the feature that individual teeth are arranged in a direction similar to the normal direction of the tooth alignment curve. In the proposed method, the maxillary and mandibular teeth can be more precisely separated than the conventional method, and the area including the individual teeth can be accurately set. Experiments using real dental CT images demonstrate the superiority of the proposed method.

• Journal of Digital Contents Society
• /
• v.18 no.1
• /
• pp.159-165
• /
• 2017

In order to solve problems in automatic quality inspection of tablet capsules, computation-efficient image processing technique, appropriate threshold setting, edge detection and segmentation methods are required. And since existing automatic system for quality inspection of tablet capsules is of very high cost, it needs to be reduced through the realization of low-price hardware system. This study suggests a technique that uses low-cost camera module to obtain image and inspects dents on tablet capsules and sorting them by applying TLS curve fitting technique and edge-based image segmentation. In order to assess the performance, the major classifications algorithm of PCA, ICA and SVM are used to evaluate training time, test time and accuracy for capsule image area and curve fitting edge data sets.