Suchergebnisse

Background Medical Image segmentation is an important image processing step. Comparing images to evaluate the quality of segmentation is an essential part of measuring progress in this research area. Some of the challenges in evaluating medical segmentation are: metric selection, the use in the literature of multiple definitions for certain metrics, inefficiency of the metric calculation implementations leading to difficulties with large volumes, and lack of support for fuzzy segmentation by existing metrics. Result First we present an overview of 20 evaluation metrics selected based on a comprehensive literature review. For fuzzy segmentation, which shows the level of membership of each voxel to multiple classes, fuzzy definitions of all metrics are provided. We present a discussion about metric properties to provide a guide for selecting evaluation metrics. Finally, we propose an efficient evaluation tool implementing the 20 selected metrics. The tool is optimized to perform efficiently in terms of speed and required memory, also if the image size is extremely large as in the case of whole body MRI or CT volume segmentation. An implementation of this tool is available as an open source project. Conclusion We propose an efficient evaluation tool for 3D medical image segmentation using 20 evaluation metrics and provide guidelines for selecting a subset of these metrics that is suitable for the data and the segmentation task. ; European Union Seventh Framework Programme (FP7/2007-2013) ; 1

? Segmentation of anatomical structures in medical image data is an essential task in clinical practice. Dagmar Kainmueller introduces methods for accurate fully automatic segmentation of anatomical structures in 3D medical image data. The author's core methodological contribution is a novel deformation model that overcomes limitations of state-of-the-art Deformable Surface approaches, hence allowing for accurate segmentation of tip- and ridge-shaped features of anatomical structures. As for practical contributions, she proposes application-specific segmentation pipelines for a range of anatom.

Medical Ultrasound (US) has many features that make it widely used in the world. These features are safety, availability and low cost. However, despite these features, the ultrasound suffers from problems. These problems are speckle noise and artifacts. In this paper, a new method is proposed to improve US images by removing speckle noise and reducing artifacts to enhance the contrast of the image. The proposed method involves algorithms for image preprocessing and segmentation. A median filter is used to smooth the image in the pre-processing. Additionally, to obtain best results, applying median filter with different kernel values. We take the better output of the median filter and feed it into the Gaussian filter, which then feeds the output of the Gaussian filter into histogram equalization to improve image visualization. The segmentation is done by thresholding and region growing segmentation. The value of threshold 128 was found to be better after we tested many values of thresholding. This value of thresholding combined with region growing gave accurate result segmentation of images. This paper demonstrates how image noise, artifacts and techniques were used effectively to improve image quality, and the analysis of performance of various techniques.

BACKGROUND. As medical marijuana legislation becomes more common, concerns arise about the overconcentration of dispensaries, raising questions about the number of medicinal marijuana dispensaries (MMD) needed to serve medicinal users. OBJECTIVES. This paper applies niche-marketing theory—which suggests dispensaries market to specific types of people—to examine if MMDs might be targeting recreational users. Observed differences between dispensary populations and between dispensary clients and local residents may indicate that dispensaries are drawing in patients based on factors other than medical need. METHODS. Data were collected via exit surveys with patients at four dispensaries in Long Beach, CA. A total of 132 patients were surveyed regarding demographic data, purchase information, medical condition, and nearest cross street for their home address. Census tract information was collected for every dispensary. RESULTS. Chi-squared tests show significant associations between dispensary visited and race (χ(2) = 31.219, p < .001) and significant associations between medical condition and dispensary visited (χ(2) = 22.123, p < .05). Lastly we found that all four of the dispensaries had patients who were different from community residents in some characteristics. CONCLUSIONS. There were significant differences relating to race, medical condition, and distance traveled across dispensaries. Results suggest dispensary users do not necessarily reside in the same area in which dispensaries are located and do not necessarily reflect the local population. Taken together these results provide some support for market segmentation.

Image segmentation is a way which applied to split an image for many parts. It will generate image smooth and easy to evaluate. A useful image segmentation technique is help the area authority in medical images. Also, a good segmented images are important in many medical fields like radiologist, pathologist of quick and successful analysis. It has used effectively in diagnosis for many disease. Nature inspired algorithms imitate the mathematical and innovative techniques for non-linear and actuality problems and can be achieved to segment or analyses the images. In this paper, a watershed as segmentation technique (WAWOA) has applied to segment the medical dataset image then proposed a whale based segmentation technique which can be used on medical bones image anyways the nature or style of the image. For fair analysis of these techniques, peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM) and Universal Image Quality Index (UIQI) of segmented images are analyzed by using MATLAB. The success of the proposed technique has confirmed with evaluated via comparing the obtained with outed results beside a standard watershed technique. The results showed this technique develops the segmentation of medical images and can help with better diagnosis. The evaluation results of our proposed technique show the success and efficiency of segmented images in high rates.

Automatic detection and segmentation of anatomical structures in medical images are prerequisites to subsequent image measurements and disease quantification, and therefore have multiple clinical applications. This book presents an efficient object detection and segmentation framework, called Marginal Space Learning, which runs at a sub-second speed on a current desktop computer, faster than the state-of-the-art. Trained with a sufficient number of data sets, Marginal Space Learning is also robust under imaging artifacts, noise and anatomical variations. The book showcases 35 clinical applications of Marginal Space Learning and its extensions to detecting and segmenting various anatomical structures, such as the heart, liver, lymph nodes and prostate in major medical imaging modalities (CT, MRI, X-Ray and Ultrasound), demonstrating its efficiency and robustness.