Wyniki wyszukiwania - BazTech

1

Segmentation of the melanoma lesion and its border

Surówka Grzegorz, Ogorzałek Maciej

International Journal of Applied Mathematics and Computer Science

|

2022

|

Vol. 32, no. 4

683--699

EN

Segmentation of the border of the human pigmented lesions has a direct impact on the diagnosis of malignant melanoma. In this work, we examine performance of (i) morphological segmentation of a pigmented lesion by region growing with the adaptive threshold and density-based DBSCAN clustering algorithm, and (ii) morphological segmentation of the pigmented lesion border by region growing of the lesion and the background skin. Research tasks (i) and (ii) are evaluated by a human expert and tested on two data sets, A and B, of different origins, resolution, and image quality. The preprocessing step consists of removing the black frame around the lesion and reducing noise and artifacts. The halo is removed by cutting out the dark circular region and filling it with an average skin color. Noise is reduced by a family of Gaussian filters 3×3−7×7 to improve the contrast and smooth out possible distortions. Some other filters are also tested. Artifacts like dark thick hair or ruler/ink markers are removed from the images by using the DullRazor closing images for all RGB colors for a hair brightness threshold below a value of 25 or, alternatively, by the BTH transform. For the segmentation, JFIF luminance representation is used. In the analysis (i), out of each dermoscopy image, a lesion segmentation mask is produced. For the region growing we get a sensitivity of 0.92/0.85, a precision of 0.98/0.91, and a border error of 0.08/0.15 for data sets A/B, respectively. For the density-based DBSCAN algorithm, we get a sensitivity of 0.91/0.89, a precision of 0.95/0.93, and a border error of 0.09/0.12 for data sets A/B, respectively. In the analysis (ii), out of each dermoscopy image, a series of lesion, background, and border segmentation images are derived. We get a sensitivity of about 0.89, a specificity of 0.94 and an accuracy of 0.91 for data set A, and a sensitivity of about 0.85, specificity of 0.91 and an accuracy of 0.89 for data set B. Our analyses show that the improved methods of region growing and density-based clustering performed after proper preprocessing may be good tools for the computer-aided melanoma diagnosis.

2

An improved Otsu method for oil spill detection from SAR images

Yu F., Sun W., Li J., Zhao Y., Zhang Y., Chen G.

Oceanologia

|

2017

|

No. 59 (3)

311--317

EN

In recent years, oil spill accidents have become increasingly frequent due to the development of marine transportation and massive oil exploitation. At present, satellite remote sensing is the principal method used to monitor oil spills. Extracting the locations and extent of oil spill spots accurately in remote sensing images reaps significant benefits in terms of risk assessment and clean-up work. Nowadays the method of edge detection combined with threshold segmentation (EDCTS) to extract oil information is becoming increasingly popular. However, the current method has some limitations in terms of accurately extracting oil spills in synthetic aperture radar (SAR) images, where heterogeneous background noise exists. In this study, we propose an adaptive mechanism based on Otsu method, which applies region growing combined with both edge detection and threshold segmentation (RGEDOM) to extract oil spills. Remote sensing images from the Bohai Sea on June 11, 2011 and the Gulf of Dalian on July 17, 2010 are utilized to validate the accuracy of our algorithm and the reliability of extraction results. In addition, results according to EDCTS are used as a comparator to further explore validity. The comparison with results according to EDCTS using the same dataset demonstrates that the proposed self-adapting algorithm is more robust and boasts high-accuracy. The accuracy computing by the adaptive algorithm is significantly improved compared with EDCTS and threshold method.

3

Segmentation of pulmonary vascular tree from 3D CT thorax scans

Fabijańska A.

Biocybernetics and Biomedical Engineering

|

2015

|

Vol. 35, no. 2

106--119

EN

This paper considers the problem of pulmonary vessels identification in thoracic 3D CT scans. In particular, the method for pulmonary vascular tree segmentation is introduced. The main idea behind the introduced method is to extract both thoracic trees together (i.e. the vascular tree and the airway tree) and then remove airway walls. Therefore, firstly segmentation of vessels and airway walls is performed using 3D region growing where the growth of the region is guided and constrained by results of random walk segmentation applied to consecutive CT slices. In particular, results of random walk segmentation of one slice are used to determine seeds for random walk segmentation of the following slice. Next step is airway tree segmentation using 3D region growing algorithm guided and constrained by the morphological gradient. Finally, morphological processing is applied in order to extend airway lumen onto airway walls and remove the overlapping regions. The main steps of the proposed approach are described in detail. Results of pulmonary vascular tree segmentation from example thoracic volumetric CT datasets provided by the introduced approach are presented and discussed. Based on a manually selected and radiologist's verified ground truth pixels and the resulting quality measures it can be concluded, that the average accuracy of the introduced approach is about 90%.

4

Application of region growing method to brain tumor segmentation - preliminary results

Piekar E., Szwarc P., Sobotnicki A., Momot M.

Journal of Medical Informatics & Technologies

|

2013

|

Vol. 22

153--160

EN

In this article image have been subject to segmentation using Matlab software, i.e. T1 in normal conditions, perfusion images and images after administering a contrast agent. The tumor in images made in normal conditions was difficult to identify. The images obtained after administering the contrast agent confirmed that the homogeneity criterion has been appropriately selected. In perfusion images the pixels of the background were added to the tumor. When the parameters were changed i.e. pixel counter or neighborhood type the method became more efficient; the tumor boundaries were outlined more precisely. The region growing method enables precise tumor detection; however, the selection of an appropriate homogeneity criterion is a prerequisite for correct segmentation.

5

HIST - an application for segmentation of hepatic images

Reska D., Krętowski M.

Zeszyty Naukowe Politechniki Białostockiej. Informatyka

|

2011

|

Z. 7

71-93

EN

HIST (Hepatic Image Segmentation Tool) is a Java-based application for segmentation and visualisation of medical images, specialised for hepatic image analysis. This paper contains an overview of the application features, a description of adapted segmentation algorithms and their experimental validation. The application provides two main segmentation methods, based on region growing and active contour model methods, adapted for the case of liver segmentation. HIST also offers data visualisation tools, including multiplanar reconstruction, volume rendering and isosurface extraction.

PL

HIST (ang. Hepatic Image Segmentation Tool – narzędzie do segmentacji obrazów wątroby) jest napisaną w języku Java aplikacją do segmentacji i wizualizacji obrazów medycznych, wyspecjalizowaną segmentacji w obrazów wątroby. Artykuł ten zawiera przegląd możliwości aplikacji, opis zaadaptowanych algorytmów segmentacji i wizualizacji oraz ich eksperymentalną walidację. Aplikacja oferuje dwie główne metody segmentacji, oparte o algorytmy rozrostu regionów i aktywnego konturu, dostosowane do segmentacji wątroby. Narzędzia wizualizacyjne aplikacji wykorzystają rekonstrukcję multiplanarną, rendering wolumetryczny oraz ekstrakcję izopowierzchni.

6

The concept of image processing algorithms for assessment and diagnosis of hydrocephalus in children

Węgliński T., Fabijańska A.

Prace Instytutu Elektrotechniki

|

2011

|

Z. 251

155-177

EN

This paper presents a concept of image processing and analysis algorithms for an automatic assessment of hydrocephalus in children's brain. Presented research was inspired by the medical need for tools performing an automatic (or at least semi-automatic) detection and quantitative evaluation of this lesion. Algorithms for precise segmentation of hydrocephalus and determination of its volume from three dimensional CT brain scans were introduced. Specifically, for brain and hydrocephalus segmentation, region growing approach was proposed. Results of applying the developed method to real CT data sets were presented and discussed. The analysis of the results show that in future, the proposed algorithms can be helpful tool for diagnosis of hydrocephalus.

PL

W niniejszym artykule przedstawiono koncepcję wykorzystania algorytmów przetwarzania i analizy obrazów do automatycznej oceny i diagnostyki wodogłowia u dzieci. Inspiracją do badań była potrzeba stworzenia skutecznych narzędzi do automatycznej (lub co najmniej półautomatycznej) detekcji i ilościowej oceny tego schorzenia dla potrzeb współczesnej neurochirurgii. Artykuł zawiera opis opracowanych algorytmów segmentacji obszaru wodogłowia oraz całego mózgu metodą rozrostu obszaru, a także przedstawia algorytm obliczający stosunek objętości zmiany chorobowej do objętości całego mózgu. Niniejszy artykuł prezentuje rezultaty zastosowania proponowanych algorytmów do rzeczywistych danych obrazowych pochodzących z tomografu komputerowego. Analiza otrzymanych rezultatów pokazuje, że proponowane algorytmy mogą stanowić użyteczne narzędzie diagnostyczne do detekcji i oceny wodogłowia u dzieci.

7

Segmentacja drzewa oskrzelowego z wykorzystaniem algorytmu zamykania otworów

Postolski M., Janaszewski M., Fabijańska A., Babout L., Jędrzejczyk M., Stefańczyk L.

Automatyka / Akademia Górniczo-Hutnicza im. Stanisława Staszica w Krakowie

|

2009

|

T. 13, z. 3/1

949-957

PL

Niezawodna segmentacja ludzkich drzew oskrzelowych ze zbiorów wolumetrycznych pochodzących z tomografii komputerowej (CT) jest ważnym elementem analizy danych w zastosowaniach klinicznych. W tym artykule został zaprezentowany nowatorski algorytm segmentacji drzewa oskrzelowego bazujący na geometrii i topologii dyskretnej. Proponowana metoda jest w pełni automatyczna, i posiada zalety dobrze zdefiniowanych pojęć matematycznych. Otwory występują w ścianach oskrzeli z wielu powodów np. są wynikiem szumów. Otwory są częstym problemem w poprzednio zaprezentowanych metodach, wykorzystujących rozrost obszaru i mogą powodować wyciek algorytmów segmentacji do otaczających oskrzela części płuc. Nowoczesność prezentowanego algorytmu polega na zastosowaniu dedykowanego algorytmu zamykania otworów, który zamknie wszystkie występujące otwory w drzewie oskrzelowym. Wyniki eksperymentów wykazały, że algorytm jest niezawodny i generuje wyniki dokładne oraz dobrej jakości.

EN

Reliable segmentation of a human airway tree from volumetric computer tomography (CT) data sets is the most important step for further analysis in many clinical applications. In this paper the original airway segmentation algorithm based on discrete topology and geometry is presented. The proposed method is fully automated and takes advantage of well defined mathematical notions. Holes occur in bronchial walls due to many reasons, for example they are results of noise. Holes are common problem in previously proposed methods because in some areas they can cause the segmentation algorithms to leak into surrounding parenchyma parts of a lung. The novelty of the approach consist in the application of a dedicated hole closing algorithm which closes all disturbing holes in a bronchial tree. The experimental results showed that the method is reliable and generate good quality and accurate results.

8

Analiza porównawcza algorytmów segmentacji drzewa oskrzelowego

Fabijańska A., Postolski M. M., Janaszewski M., Babout L.

Automatyka / Akademia Górniczo-Hutnicza im. Stanisława Staszica w Krakowie

|

2009

|

T. 13, z. 3/1

821-830

PL

W artykule rozważono problem segmentacji drzewa oskrzelowego z trójwymiarowych tomogramów klatki piersiowej. W szczególności dokonano porównania dwóch autorskich algorytmów wykorzystujących segmentację przez rozrost obszaru. Pierwsza z rozważonych metod w celu uniknięcia wycieku do płuc wykorzystuje algorytm zamykania otworów, druga - rozrost obszaru ograniczony przez gradient morfologiczny. W artykule przedstawiono i przedyskutowano również porównanie wyników obu metod uzyskanych dla przykładowych danych.

EN

In this paper problem of airway tree segmentation from 3D CT chest scans was considered. Especially comparison of two authors' algorithms was provided. The algorithms are 3D region growing approaches. The first method uses hole closing algorithm in order to avoid leakages into the lungs. The second approach guides and constrains region growing by morphological gradient. Results of both considered methods for exemplary data are presented and discussed.

9

Segmentacja i dopasowywanie cyfrowych obrazów medycznych: przetwarzanie nagrań wideo-endoskopowych strun głosowych oraz danych tomograficznych zmian rakowych

Skalski A., Zieliński T.

Pomiary Automatyka Kontrola

|

2008

|

R. 54, nr 6

330-333

PL

Celem artykułu jest wprowadzenie do zagadnienia segmentacji i dopasowywania cyfrowych obrazów medycznych 2D i 3D, np. z endoskopii i tomografii komputerowej, oraz krótki przegląd stosowanych metod. Na tym tle zaprezentowano nowe, oryginalne wyniki prac własnych autorów, dotyczących analizy cyfrowych nagrań wideo strun głosowych. Badania te mają na celu estymację parametrów ruchu tych strun dla ludzi zdrowych oraz chorych, np. ze zmianami nowotworowymi. W tym ostatnim przypadku wskazana jest analiza danych wideo przed i po terapii laserowej. W artykule porównano poprzednie wyniki autorów uzyskane dla metody segmentacji metodą poziomic (level sets) z metodą rozrostu obszarów (region growing). W końcowej części pracy zaprezentowano przykład zastosowania dopasowywania danych tomograficznych pacjenta podczas radioterapii zmian nowotworowych.

EN

In the paper introduction to segmentation and registration of medical 2D/3D data, coming from medical endoscopy and computed tomography, is done and a brief description of the most popular methods is presented. On this background, as an example, new original results of vocal folds video analysis are given. In this case evaluation of vocal folds motion parameters for people in good health and sick persons with cancer changes is addressed, especially before and after the laser treatment. In the paper previous segmentation results obtained for level sets methods are compared with application of a region growing approach. Finally, application of registration to computed tomography data before cancer radiotherapy is shown in the paper.

10

Wydzielanie kompleksów krajobrazowo-roślinnych na zdjęciach Landsat ETM+ z zastosowaniem procedury region growing

Kosiński K., Hoffmann-Niedek A., Zawiła A.

Archiwum Fotogrametrii, Kartografii i Teledetekcji

|

2006

|

Vol. 16

341--350

PL

Celem pracy jest delimitacja kompleksów krajobrazowo-roślinnych na wieloczasowych zdjęciach Landsat ETM+. Kompleks krajobrazowo-roślinny to niewielka jednostka geobotaniczna nawiązująca do nanochory w hierarchicznym systemie jednostek fizycznogeograficznych. Kompleksy wydziela się jako względnie jednorodne segmenty obrazu z zastosowaniem procedury Region Growing. Analizowany jest dobór opcji procedury i granicznej odległości euklidesowej (SED). Zalecana jest opcja aktualizacji średniej. Nie stwierdzono możliwości ustalenia konkretnej wartości SED dla poszczególnych typów kompleksów. Stąd konieczność ręcznego doboru SED dla każdego z segmentów.

EN

The aim of the article was the delimitation of landscape-vegetation complexes on multitemporal Landsat ETM+ images. A landscape-vegetation complex is a small geobotanic unit corresponding to a nanochore level of physico-geographical units (Matuszkiewicz, 1990, 1992; Richling, Solon, 2002). A landscape-vegetation complex consists of one or both of the two main components: plant communities (phytocoenoses) and anthropogenic forms of relief and land cover (excavations, water reservoirs, ditches, dams, roads, buildings etc.). A landscape-vegetation complex is useful for mapping terrains under spatially-differentiated anthropogenic pressure. Landscape-vegetation complex may be visually interpreted on diachronic composition of panchromatic data. The composition used in this paper consist of the three bands: - red component: ETM(8) acquired in September, - green component: ETM(8) acquired in May, - blue component: ( ETM(1) + ETM(2) + ETM(3) ) / ETM(8), September data. Nine types of complexes may be distinguished on the composition (Kosiński, 2005): - waters, - build up and railway areas, - pine woods, - broadleaf forests and thickets, - arable lands and four categories of grasslands. In this project, landscape-vegetation complexes have been interactively delimited as semi-homogeneous image segments using a Region Growing procedure. 6 interpreters participated in this work. Procedure options and Spectral Euclidean distance (SED) were interactively adjusted for each segment and different options were tested. The update region mean option was assumed to be recommended. In forest areas, the required spectral Euclidean distance was low and in grasslands it is higher. However, there is no possibility to use one SED value for any type of complex. It may be manually adjusted for each segment. Complexes were classified on the basis of colour component values. Aerial photos and topographic maps were used as additional data.

11

Self-learning model-based segmentation of medical images

Pohle R., Toennies K.

Image Processing & Communications

|

2001

|

Vol. 7, no 3-4

97-113

EN

Interaction increases flexibility of segmentation but it leads to undesirable behaviour of an algorithm if knowledge being requested is inappropriate. In region growing, this is the case for defining the homogeneity criterion, as its specification depends also on image formation properties that are not known to the user. We developed a region growing algorithm that learns its homogeneity criterion automatically from characteristics of the region to be segmented. The method is based on a model that describes homogeneity and simple shape properties of the region. Parameters of the homogeneity criterion are estimated from sample locations in the region. These locations are selected sequentially in a random walk starting at the seed point, and the homogeneity criterion is updated continuously. In contrast to other adaptive region growing methods our approach produces results that are far less sensitive to the seed point location, and it allows a segmentation of individual structures. The model-based adaptive region growing approach was extended to a fully automatic and complete segmentation method by using the pixels with the smallest gradient length in the not yet segmented image region as a seed point. Both methods were tested for segmentation on test images and of structures in CT images. The performance of the semi-automatic method is compared with the adaptive moving mean value region growing method, and the automatic method is compared with the watershed segmentation. We found our method to work reliable if the model assumption on homogeneity and region characteristics were true. Furthermore, the model is simple but robust thus allowing for a certain amount of deviation from model constraints and still delivering the expected segmentation result.