Search

Karcinom dojke je najčešća zloćudna bolest u žena. Napretkom terapije karcinoma u zadnjih pola stoljeća naglo je narasla potreba za dijagnostičkim metodama sposobnim neinvazivno, precizno i pravovremeno pronaći i klasificirati lezije dojke. Automatski ultrazvuk dojke se razvio iz potrebe da se nadomjeste nedostatci konvencionalnog ručnog ultrazvuka i mamografije pri probiru i klasifikaciji lezija, posebno u mladih žena. Moderni uređaji za automatski ultrazvuk stvaraju trodimenzionalnu snimku dojke, koju radiolog naknadno interpretira. To omogućuje rekonstrukciju i analizu u više ravnina što nosi različite prednosti u odnosu na dvodimenzionalne slikovne metode. Reproducibilnost i neovisnost o operateru omogućuju korištenje podataka automatskog ultrazvuka dojke u treniranju modernih umjetnih neuronskih mreža i postavljaju podlogu za daljnji razvoj ovog obećavajućeg dijagnostičkog modaliteta. U ovom radu će biti opisane najčešće bolesti dojke i njihova klasifikacija, najčešće dijagnostičke metode za karakterizaciju lezija dojke, princip rada automatskog ultrazvuka dojke i tehnika snimanja, prednosti i nedostatci automatskog ultrazvuka u karakterizaciji i klasifikaciji lezija dojke te njegova usporedba s drugim dijagnostičkim modalitetima., Breast cancer is the most common malignant disease in women. With advancements in pharmacotherapy made in the last fifty years there has been an increasing demand for non-invasive, precise, and timely diagnostic methods to detect and classify breast lesions. Automated breast ultrasound was developed to make up for some shortcomings of handheld ultrasound and mammography. Modern automated ultrasound devices scan a three-dimensional image of the breast, that gets interpreted by the radiologist later. This enables multiplanar viewing and analysis that bring multiple advantages over conventional two-dimensional imaging modalities. Reproducibility and operator independence enable automated ultrasound data to be used in modern artificial neural networks and are laying the foundations for future development of this promising diagnostic method. In this paper I will describe the most common breast diseases and their classification, the most common diagnostic methods for breast lesion characterisation, the principles of automated ultrasound technique and image acquisition, advantages and disadvantages of automatic ultrasound in lesion characterisation and classification and its comparison with other diagnostic modalities.

Diese Arbeit beschäftigt sich mit der computerassistierten Diagnose (CAD) zur Detektion von Lungenknoten. Der Lungenkrebs ist einer der häufigsten Todesursachen auf der Welt. Der Hauptauslöser für Bronchialkarzinome ist der vermehrte Tabakkonsum. Die schädlichen Stoffe können Zellen schädigen und somit dem Atmungsorgan schaden. Um die Sterberate zu senken, ist eine Frühdiagnose des Bronchialkarzinoms notwendig, da diese Lungenerkrankung meistens symptomfrei entsteht. Bei späteren Stadien ist eine vollständige Heilung des Öfteren nicht möglich. Daher ist der Einsatz von CAD Systemen von großer Bedeutung. CAD Systeme sind in der Lage, anhand von „lernenden“ Algorithmen, Lungenknoten zu detektieren. Für das Trainieren der Systeme wird eine große Datenbank benötigt. Umso mehr Lernmaterial zur Verfügung steht, desto besser können CAD Systeme trainiert werden. Die künstliche Intelligenz spielt dabei eine wichtige Rolle, da während des Lernprozesses keine Beispiele auswendig gelernt, sondern bestimmte Muster eingeprägt werden. Diese erworbenen Kenntnisse werden dann an unbekannte Daten angewendet. Laut einigen Studien konnte festgestellt werden, dass anhand CAD Systeme die Detektionsrate und Sensitivität erhöht werden konnten. Weitere Studien haben auch gezeigt, dass CAD Systeme mehr an Bedeutung in der Medizin gewonnen haben. Die Befundungszeit konnte auch wesentlich gesenkt werden. Zusätzlich können CAD Systeme als Lerntools für Radiolog*innen und Radiologietechnolog*innen dienen. Da viele Berufsgruppen in der Radiologie häufig zusammenarbeiten ist es notwendig, sich gegenseitig zu unterstützen und bei Verdacht auf eine gefährliche Krankheit, Ärzt*innen auf die Röntgenaufnahme aufmerksam zu machen. Zu beachten ist auch, dass auf dem Gebiet der Radiologie, die kontinuierliche Entwicklung und Aktualisierung von Verfahren beziehungsweise Modalitäten mehr zunimmt. Als Radiologtechnolog*in ist es daher wichtig, stets auf dem neuesten Stand der Technik zu sein. This thesis deals with computer-assisted diagnosis (CAD) for the detection of lung nodules. Lung cancer is one of the leading causes of death in the world. The main trigger for bronchial carcinoma is increased tobacco consumption. The harmful substances can damage cells and thus damage the respiratory system. In order to reduce the mortality rate, the early diagnosis of bronchial carcinoma is necessary, as this lung disease usually develops without symptoms. Complete healing is often not possible in later stages. Therefore, the use of CAD systems is of great importance. CAD systems can detect lung nodules using machine learning algorithms. A large database is required to train the systems. The more learning material is available, the better CAD systems can be trained. Artificial intelligence plays an important role here, since no examples are learned by heart during the learning process, but rather certain patterns are imprinted. This acquired knowledge is then applied to unknown data. According to some studies, it could be determined that the detection rate and sensitivity could be increased using CAD systems. Further studies have also shown that CAD systems have gained more importance in medicine. The time of diagnosis could also be reduced significantly. In addition, CAD systems can serve as learning tools for radiologists. Since many professional groups in radiology often work together, it is often necessary to support each other and to draw the attention of doctors to the X-ray if a dangerous disease is suspected. It should also be noted that in the field of radiology the continuous development and updating of procedures and modalities is increasing. As a radiologist technologist, it is therefore important to always be up to date with the latest technology.

We have developed a refined computer-based method to detect joint space narrowing (JSN) progression with the joint space narrowing progression index (JSNPI) by superimposing sequential hand radiographs. The purpose of this study is to assess the validity of a computer-based method using images obtained from multiple institutions in rheumatoid arthritis (RA) patients. Sequential hand radiographs of 42 patients (37 females and 5 males) with RA from two institutions were analyzed by a computer-based method and visual scoring systems as a standard of reference. The JSNPI above the smallest detectable difference (SDD) defined JSN progression on the joint level. The sensitivity and specificity of the computer-based method for JSN progression was calculated using the SDD and a receiver operating characteristic (ROC) curve. Out of 314 metacarpophalangeal joints, 34 joints progressed based on the SDD, while 11 joints widened. Twenty-one joints progressed in the computer-based method, 11 joints in the scoring systems, and 13 joints in both methods. Based on the SDD, we found lower sensitivity and higher specificity with 54.2 and 92.8%, respectively. At the most discriminant cutoff point according to the ROC curve, the sensitivity and specificity was 70.8 and 81.7%, respectively. The proposed computer-based method provides quantitative measurement of JSN progression using sequential hand radiographs and may be a useful tool in follow-up assessment of joint damage in RA patients.

Rupture of cerebral aneurysms is the main cause of subarachnoid hemorrhage, which can have devastating effects on quality of life. The identification and assessment of unruptured aneurysms from medical images is therefore of significant clinical relevance. In recent years, the availability of clinical imaging data has rapidly increased, which calls for computer assisted detection (CAD) systems. Previous studies have shown that CAD systems based on convolutional neural networks (CNN) can help to detect cerebral aneurysms from magnetic resonance angiographies (MRAs). However, these CAD systems require large datasets of annotated medical images. Thus, more efficient tools for processing and categorizing medical imaging data are required. Previous studies of CNN-based classification for medical images used various patch configurations of input data. These studies showed that classification accuracy was affected by the patch size or image representation. Thus, we hypothesize that the accuracy of CADs to detect cerebral aneurysms can be improved by adjusting the configuration of the input patches. In the present study, we performed CNN-based medical imaging classification for varying input data configurations to examine the relationship between classification accuracy and data configuration.

Son yıllarda tıp alanında, görüntüleme tekniklerinin sıklıkla kullanılması sayesinde Bilgisayar Destekli Tespit (BDT) sistemleri de kendisine geniş bir yer bulmuştur. BDT sistemlerinde en önemli işlem bölütleme aşamasının doğru bir şekilde yapılmasıdır. Bu tez çalışmasında, Bilgisayarlı Tomografi (BT) görüntüleri üzerinde karaciğerin bölütlenmesi için bilgisayar destekli otomatik bir yöntem ve yazılım geliştirilmiştir. Bölütleme işlemleri için Bölge Büyütme (RG) ve Bulanık C-Ortalama (FCM) algoritmalarından yararlanılmıştır. Bu algoritmalar ile yapılan bölütlemelerin başarımını tespit etmek için alanında uzman bir hekimden destek alınmıştır. Bu amaçla uzman tarafından yapılan seçim ölçüt olarak kullanılarak RG ve FCM algoritmaları ile bilgisayar destekli olarak elde edilen bölütleme sonuçları karşılaştırılmıştır. Karşılaştırma ölçütü olarak Jaccard benzerlik ölçütü kullanılmıştır. 88 BT görüntüsü üzerinde yapılan deneysel çalışmalarda Jaccard benzerlik ölçütüne göre RG algoritmasında %91.15 ve FCM algoritmasında %75.16 bölütleme için ortalama başarım değerleri bulunmuştur. Sonuç olarak Bölge Büyütme algoritması ile yapılan bölütleme işlemlerinin daha başarılı olduğu görülmüştür. Ayrıca, benzerlik ölçümleri sonucunda bulunan nicel değerlerin istatiksel olarak anlamlılık testlerinin değerlendirilmeleri de gerçekleştirilmiş ve RG algoritması ile elde edilen daha başarılı bölütleme sonuçlarının, anlamlı bir fark ortaya koyduğu sonucuna ulaşılmıştır. Bunlara ek olarak, benzerlik ve anlamlılık testleri ile beraber işlem zamanlarının karşılaştırmaları da yapılmış ve burada da RG ile yapılan bölütleme işleminin daha hızlı olduğu sonucuna varılmıştır. Tüm bu elde edilen bulgular ile bu tez çalışmasında karaciğerin bölütlenmesi için önerilen bilgisayar destekli yöntemin ve geliştirilen yazılımın, hekimlerin karar verme aşamalarında ikincil bir yardımcı araç olarak kullanılabileceği görülmüştür. In recent years, computer assisted detection(CAD) systems have also found a large place in the medical field due to frequent use of imaging techniques. The most important process in CADsystems is the correct segmentation phase. In this thesis, computer assisted automated methods and software were developed to divide the liver over computerized tomography(CT) images. For segmentation operations, the area magnification (RG) and fuzzy c-average (FCM) algorithms were used. In order to determine the success of these algorithms, support has been obtained from a specialist physician in the field. For this purpose, computer-aided segmentation results were compared with RG and FCM algorithms using expert selection criteria. The comparison criteria were used. In the experimental studies performed on 88 BT image, the average performance values were 91.15% in RG algorithm and 75.16% in FCM algorithm according to jaccard similarity criteria. As a result, segmentation operations with the zone magnification algorithm were more successful. In addition, statistical significance tests of the quantitative values as a result of similarity measurements have been performed and more successful segmentation results obtained by the RG algorithm revealed a significant difference was found. In addition, comparisons were made with similar and semantic tests, and it was concluded that the splitting process with RG was faster. In this study, it was observed that the computer-aided method and the developed software for liver division could be used as a secondary tool in the decision-making process of physicians. 69

Asthma, Chronic obstructive pulmonary disease, influenza, pneumonia, tuberculosis, lung cancer and many other breathing problems are the leading causes of death and disability all over the world. These diseases affect the lung. Radiology is a primary assessing method with low specificity of the prediction of the presence of these diseases. Computer Assisted Detection (CAD) will help the specialists in detecting one of these diseases in an early stage. A method has been proposed by Ulas Bagci to detect lung abnormalities using Fuzzy connected object estimation, Ball scale encoding and comparing various features extracted from local patches of the lung images (CT scan). In this paper, the Tree-in-Bud patterns are selected after segmentation by using ball scale encoding algorithm.

本稿では,4時相の腹部3次元CT像からの肝血管腫検出手法を提案する.本手法では,まず門脈相CT像から肝血管腫の候補領域を検出する.事前に生成した濃度値分布クラスを用いて各画素の事後確率を計算し,各画素の事後確率が最も高くなる濃度値分布クラスに画素を分類することで,肝血管腫候補領域を得る.得られた肝血管腫候補領域から過検出の原因となる肝臓血管を抽出し,肝血管腫候補領域から削除する.次に,肝血管腫候補領域に対して塊状構造強調フィルタやガウシアンフィルタ,メディアンフィルタなどの特徴量計測用フィルタを適用し,フィルタの出力値から各画素ごとの特徴量ベクトルを生成する.各画素の特徴量ベクトルから多クラス識別器により各画素のカテゴリを決定し,肝血管腫であるとカテゴリ分けされた領域を最終的な肝血管腫領域として検出する.腹部3次元CT像15症例に対して実験を行ったところ,検出結果の領域単位における再現率は83.3%,拾い過ぎ数は15.4個/画像であった., In this paper, we propose a method for detecting hemangioma from four-phases contrasted 3D abdominal CT images. In our method, candidate regions of the liver hemangioma are detected from a portal phase CT image. The candidate regions of the liver hemangioma are obtained by classifying voxels based on the posterior probability of each voxel in a liver region. Since false positives of detection results usually include blood vessel regions, we remove the blood vessel regions in the liver from the candidate regions. Then, we calculate a feature value vector at each voxel in the candidate regions. Feature value vectors are calculated by using various filters such as median, gaussian, and blob enhancement filters. A category of each feature value vector is determined by a multi-class classifier. Finally, we obtain regions whose category is classified as the liver hemangioma. We apply this method to 15 cases of abdominal CT images. Recall rates of the segmentation accuracy and the number of False Positives were 83.3% and 15.4 per case, respectively., IEICE Technical Report;MI2011-146

The implementation of a database of digitised mammograms is discussed. The digitised images were collected beginning in 1999 by a community of physicists in collaboration with radiologists in several Italian hospitals as a first step in developing and implementing a computer-aided detection (CAD) system. All 3,369 mammograms were collected from 967 patients and classified according to lesion type and morphology, breast tissue and pathology type. A dedicated graphical user interface was developed to visualise and process mammograms to support the medical diagnosis directly on a high-resolution screen. The database has been the starting point for developing other medical imaging applications, such as a breast CAD, currently being upgraded and optimised for use in a distributed environment with grid services, in the framework of the Instituto Nazionale di Fisicia Nucleare (INFN)-funded Medical Applications on a Grid Infrastructure Connection (MAGIC)-5 project.

3-D object segmentation is an important and challenging topic in computer vision that could be tackled with artificial life models. A Channeler Ant Model (CAM), based on the natural ant capabilities of dealing with 3-D environments through self-organization and emergent behaviours, is proposed. Ant colonies, defined in terms of moving, pheromone laying, reproduction, death and deviating behaviours rules, is able to segment artificially generated objects of different shape, intensity, background. The model depends on few parameters and provides an elegant solution for the segmentation of 3-D structures in noisy environments with unknown range of image intensities: even when there is a partial overlap between the intensity and noise range, it provides a complete segmentation with negligible contamination (i.e., fraction of segmented voxels that do not belong to the object). The CAM is already in use for the automated detection of nodules in lung Computed Tomographies.

Computer-Assisted Detection in FLAIR and DT neuroimages: automatic segmentation and volume assessment of cerebral gliomas. Giorgio De Nunzio1,6, Marina Donativi1,6, Antonella Castellano2,6, Gabriella Pastore3,6, Matteo Rucco4,6, Antonella Iadanza2, Marco Riva5, Lorenzo Bello5, Andrea Falini2 (1) Univ. of Salento, Dept. of Mathematics and Physics, and INFN, Lecce (2) U.O. Neuroradiologia, Ospedale San Raffaele e Univ. Vita-Salute, Milano (3) Istituto di Ricerche Cliniche Ecomedica, Centro di Radioterapia e IGRT, Empoli (4) Univ. of Camerino, School of Science and Technology, Computer Science Division, Camerino (5) U.O. Neurochirurgia, Ist. Clinico Humanitas, Univ. di Milano, Milano (6) ADAM srl, Advanced Data Analysis in Medicine, http://adamgroup.it Purpose: tumor cells in cerebral gliomas invade surrounding tissues preferentially along WM tracts, spreading beyond the abnormal area depicted on conventional MR images. Diffusion Tensor Imaging can reveal larger peritumoral abnormalities in gliomas that are not apparent on conventional MRI. We aimed at characterizing pathological vs healthy tissue in FLAIR and DTI datasets by 3D statistical Texture Analysis, developing an automatic segmentation technique for cerebral glioma, hereafter called GlioCAD, especially useful in patient follow-up during chemotherapy, and for preoperative assessment of tumor extension. Methods and materials: thirty-four patients with gliomas were selected. 3T axial 3D-FLAIR, axial 3D-T1w, and DTI (single-shot EPI sequence, b=1000 s/mm2, 32 gradient directions) were acquired. Isotropic and anisotropic maps (FA, MD, p and q) were calculated, and pathological ROIs were manually drawn. 3D texture features were calculated with a sliding window approach in the segmented ROIs and in the contralateral healthy tissue, for CAD-system training. The feature-space dimensionality was reduced by Linear Discriminant Analysis, which allowed tissue classification by simple thresholding. Results: For each map, tumor-classification sensitivity, specificity and ROC curves (0.90≤AUC≤0.97) were calculated, and manual and automatic segmentations were compared by the Jaccard Coefficient, showing good concordance. The CAD system automatically calculated lesion volumes and histograms. With the purpose of allowing remote fruition of GlioCAD, a Graphical User Interface was designed as a plugin for OsiriX, a well-known radiological viewer. Conclusion: GlioCAD is proposed as a new tool, based on statistical textural analysis, for the automatic segmentation and volume assessment of brain gliomas, and for the quantitative analysis of the histograms in the regions of interest.