Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userAuto Measurement and Segmentation of Head Region in Fetal Ultrasound Images
تجزئة و قياس منطقة الرأس في الصور فوق الصوتية للجنين بشكل آلي
940
0 84
0
(
0
)
Created by
Shamra Editor
Ask ChatGPT about the research
يقدم البحث طريقة مبتكرة في تجزئة رأس الجنين آلياً في الصور فوق الصوتية Ultrasound Images قليلة التباين. حيث تعاني تلك الصور من كمية ضجيج مرتفعة تؤثّر على الظهور البصري لمنطقة الرأس, كذلك ضعف الحواف و عدم إحاطتها بالمنطقة المرغوبة بشكل كامل يجعل من عملية التجزئة صعبة و مهمة في نفس الوقت, خصوصاً أن البحث اعتمد التجزئة الآلية Auto Segmentation دون الحاجة إلى تدخل المستخدم في أي مرحلة من المراحل. اعتمدنا على تقنية ضبط المستوى Level-Set لتجزئة منطقة الرأس, بعد تحديد الإطار الأولي Initial Contour بشكل آلي عن طريق تابع خصائص المنطقة Region Properties. الطريقة المقترحة أثبتت فعاليتها في اقتطاع منطقة الرأس دون التأثّر بالضجيج الموجود أو بالانقطاعات الحاصلة أحياناً للحواف, بالرغم من عدم وجود مرحلة معالجة مسبقة Pre-Processing ضمن سلسلة الخطوات المتتالية المطبّقة على عدة صور فوق صوتية بأحجام و مصادر مختلفة. ليتم في النهاية حساب القطر الثانوي للقطع الناقص (قطاع رأس الجنين Head) الناتج بالاعتماد على تابع خصائص المنطقة, القياس النهائي يمثّل المسافة بين الجداريين Bi Parietal Diameter BPD, و هو قياس مهم يمكّن الطبيب من تقدير عمر الحمل و تحديد تاريخ الولادة للجنين. تمت مصادقة نتيجة التجزئة بالاعتماد على معايير التشابه, أما دقة القياس النهائي فقد تمت مقارنته مع قياسات يدوية قام بها طبيب مختص. و قد أبدت نتائج المقارنة فعالية الخوارزمية المقترحة و نجاحها بنسبة تصل إلى 98%.
This Paper offers an innovative way for auto segmentation of the fetal head in ultrasound US images. There is high amount of noise in US images, which it affects the visual appearance of the area of head. The research depends on auto segmentation mechanism without the need for user intervention at any stage of proposed method, so this is what makes segmentation process is difficult and important at the same, because the weakness of the edges and not fully enclosed in the desired region. We relied on a Level Set method to segment the head area, after determining the initial contour automatically by the Region Properties Function. The proposed method proves effective in the head area segmentation without being influenced by noise or the existence of discontinuities in the edges of the head, despite the absence of a pre-processing stage in a series of steps applied to several ultrasound images in different sizes and sources. The last step is to calculate the secondary diameter of the output ellipse (the fetal head sector) depending on the properties of the region, and this final measurement represents the Bi Parietal Diameter BPD, an important measure enables the physician to assess gestational age and determine the birth of the fetus date. Segmentation result has been authenticated based on similarity criteria, and the final measurement accuracy has been compared with manual measurements carried out by a specialist. The comparison results showed the effectiveness of the proposed algorithm and its success by up to 98%.
References used
SHAN, J. A fully automatic segmentation method for breast ultrasound images, UTAH STATE UNIVERSITY, Logan, Utah, 2011, Pages 12-63
CHEN, Y.; Huang, F.; Tagare, H.; and Rao, M., A coupled minimization problem for medical image segmentation with priors, Int. J. Comput. Vis. 71(3), 2007, 259–272
KALE, A. and S, AKSOY. Segmentation of Cervical Cell Images. 20th International Conference on Pattern Recognition (ICPR), 2010
rate research
Read More
This Paper offers an effective method to measure the length of the
femur in Fetal Ultrasound Images, it applies a series of steps
starting with the reducing amount of noise in these images, and
then converted them to a binary form and uses morphol
ogical
operations to segment the femur and isolate it from the rest of the
image objects, then it applies an Edge Detector in order to find the
edges of the bone, then uses the Hough Transform to detect straight
lines in the image. we apply overlapping for resulted lines on the
original image, finally we choose the most significant and longest
straight line which is corresponding to the length of the femur. The
proposed method facilitates the measurement of the femur without
the help of a physician through a series of steps.
Fetus images produced by 2D ultrasound devices are ambiguous and lack precision. This led to the need
for offering a 3D visualization of the fetus, which allows visualizing width, height, and angle, in order to
get additional information about the
fetus, and detect fetus abnormalities.
We explain in this paper our method in producing 3D models of the fetus from 2D images using a
computer system without the need for changing the 2D imaging devices, and without using position
sensors.
Our method is based on passing the probe over the pregnant woman's abdomen and make a manual scan
for the entire body of the fetus from top of the head till the bottom of his feet, then it saves this scan as a
video clip then send, it to the computer which segments the video into multiple images which are saved and
later processed using digital processing principles of images. Then these processed images are
reconstructed to produce the volume matrix and then display it in a 3D form using 3D model construction
methods.
We applied our software on various fetuses of different ages and got volume images which are considered
good in comparison with the images offered by currently available systems and devices. The precision of
the images we got, differs according to the change in fetus pose, amniotic liquid, and fetus size, The
obstetrician or gynecologist can retrieve more precise details by changing the angle and displaying volume
images of certain part of the fetus body.
Breast cancer is the second leading cause of death of women in the world. The early detection gives a
better chance to cure it. Physicians diagnose breast tumors by analyzing the characteristics of the lesion in
ultrasound images. Shape data, provi
ded by a tumor contour, is important to physicians in making
diagnostic decisions. However, due to the increasing use of technology in medicine, a computer aided
detection systems (CAD) have been built to help the expert. This research focuses on using a level-set
method as an effective lesion segmentation method for breast ultrasound images. By applying non-local
means filter on image, the unwanted speckle noise will be removed and the image's important details will
be preserved. Then the initial contours are sketched using the GUI in order to apply level-set method
which delineates the contour of the lesion in breast ultrasound image. The proposed method was found to
determine the breast tumor contours that are very similar to manual-sketched contours (about 96%).
Total Polyphenol was studied in olive leaves collected from Lattakia area
(AlQurdaha zone), where Phenolic compounds were extracted from dry Olive leaves by
using two extraction methods: maceration and Ultrasonic Device. By the both two methods
a
study was carried out on the extraction solvent concentration effect where the extraction
was done by using the mixture of Ethanol-Water with different percentages (60,70,80
%). As well, The extraction temperature effect was studied at the both methods at (20,30,
40°C); in addition to the study of the extraction time effect.
At the method of Maceration Extraction, the total polyphenols amount was studied in
dry olive leaves after extraction during different times (24,48,72 h). While at the Ultrasonic
Extraction the times were (10,20,30 min).
The study showed that the highest amount of phenolic compounds were existed in
dry olive leaves which were extracted by Ultrasonic Waves with the concentration of the
solvent ethanol-water 80%, temperature of 40°C, and with an extraction time of 20min.
Neural Machine Translation (NMT) for Low Resource Languages (LRL) is often limited by the lack of available training data, making it necessary to explore additional techniques to improve translation quality. We propose the use of the Prefix-Root-Post
fix-Encoding (PRPE) subword segmentation algorithm to improve translation quality for LRLs, using two agglutinative languages as case studies: Quechua and Indonesian. During the course of our experiments, we reintroduce a parallel corpus for Quechua-Spanish translation that was previously unavailable for NMT. Our experiments show the importance of appropriate subword segmentation, which can go as far as improving translation quality over systems trained on much larger quantities of data. We show this by achieving state-of-the-art results for both languages, obtaining higher BLEU scores than large pre-trained models with much smaller amounts of data.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
