Dissertations / Theses on the topic 'Chest X-ray image'

The thesis develops a model (that includes a conceptual framework and an implementation) for analysing and classifying traditional X-ray images (MACXI) according to the severity of diseases as a Computer-Aided-Diagnosis tool with three initial objectives. � The first objective was to interpret X-ray images by transferring expert knowledge into a knowledge base (CXKB): to help medical staff to concentrate only on the interest areas of the images. � The second objective was to analyse and classify X-ray images according to the severity of diseases through the knowledge base equipped with an image processor (CXIP). � The third objective was to demonstrate the effectiveness and limitations of several image-processing techniques for analysing traditional chest X-ray images. A database was formed based on collection of expert diagnosis details for lung images. Five important features from lung images, as well as diagnosis rules were identified and simplified. The expert knowledge was transformed into a Knowledge base (KB) for analysing and classifying traditional X-ray images according to the severity of diseases (CXKB). Finally, an image processor named CXIP was developed to extract the features of lung images features and image classification. CXKB contains 63 distinct lung diseases with detailed descriptions. Some 80-chest X-ray images with diagnosis details were collected for the database from different sources, including online medical resources. A total of 61 images were used to determine the important features; 19 chest X-ray images were not used because of low visibility or the difficulty of diagnosis. Finally, only 12 images were selected after examining the diagnosis details, image clarity, image completeness, and image orientation. The most important features of lung diseases are a pattern of lesions with different levels of intensity or brightness. The other major anatomical structures of the chest are the hilum area, the rib area, the trachea area, and the heart area. Seven different severity levels of diseases were determined. Development and simplification of rules based on the image library were analysed, developed, and tested against the 12 images. A level of severity was labelled for each image based on a personal understanding of all the image and diagnosis details. Then, MACXI processed the selected 12 images to determine the level of severity. These 12 images were fed into the CXIP for recognition of the features and classification of the images to an accurate level of severity. Currently, the processor has the ability to identify diseased lung areas with approximately 80% success rate. A step by step demonstration of several image processing techniques that were used to build the processor is given to highlight the effectiveness and limitations of the techniques for analysing traditional chest X-ray images is also presented.

In recent years, image segmentation by using deep neural networks has made great progress. However, reaching a good result by training with a small amount of data remains to be a challenge. To find a good way to improve the accuracy of segmentation with limited datasets, we implemented a new automatic chest radiographs segmentation experiment based on preliminary works by Chunliang using deep learning neural network combined with shape context information. When the process was conducted, the datasets were put into origin U-net at first. After the preliminary process, the segmented images were then repaired through a new network with shape context information. In this experiment, we created a new network structure by rebuilding the U-net into a 2-input structure and refined the processing pipeline step. In this proposed pipeline, the datasets and shape context were trained together through the new network model by iteration. The proposed method was evaluated on 247 posterior-anterior chest radiographs of public datasets and n-folds cross-validation was also used. The outcome shows that compared to origin U-net, the proposed pipeline reaches higher accuracy when trained with limited datasets. Here the "limited" datasets refer to 1-20 images in the medical image field. A better outcome with higher accuracy can be reached if the second structure is further refined and shape context generator's parameter is fine-tuned in the future.
Under de senaste åren har bildsegmentering med hjälp av djupa neurala nätverk gjort stora framsteg. Att nå ett bra resultat med träning med en liten mängd data kvarstår emellertid som en utmaning. För att hitta ett bra sätt att förbättra noggrannheten i segmenteringen med begränsade datamängder så implementerade vi en ny segmentering för automatiska röntgenbilder av bröstkorgsdiagram baserat på tidigare forskning av Chunliang. Detta tillvägagångssätt använder djupt lärande neurala nätverk kombinerat med "shape context" information. I detta experiment skapade vi en ny nätverkstruktur genom omkonfiguration av U-nätverket till en 2-inputstruktur och förfinade pipeline processeringssteget där bilden och "shape contexten" var tränade tillsammans genom den nya nätverksmodellen genom iteration.Den föreslagna metoden utvärderades på dataset med 247 bröströntgenfotografier, och n-faldig korsvalidering användes för utvärdering. Resultatet visar att den föreslagna pipelinen jämfört med ursprungs U-nätverket når högre noggrannhet när de tränas med begränsade datamängder. De "begränsade" dataseten här hänvisar till 1-20 bilder inom det medicinska fältet. Ett bättre resultat med högre noggrannhet kan nås om den andra strukturen förfinas ytterligare och "shape context-generatorns" parameter finjusteras.

Thesis (PhD) -- University of Canberra, 2006.
Thesis submitted in fulfilment of the requirements for the degree of Master of Information Science in the School of Information Sciences and Engineering under the Division of Business, Law and Sciences at the University of Canberra, May 2006. Bibliography: leaves 120-132.

Дисертаційна робота присвячена розв’язанню актуальної науково-прикладної задачі автоматизації процесу діагностування вірусного пневмонічного запалення за медичними зображеннями легень через розроблення інформаційної технології раннього діагностування пневмонії за індивідуальним підбором параметрів моделі класифікації медичних зображень легень. Застосування розробленої інформаційної технології раннього діагностування пневмонії в клінічній практиці дає змогу підвищити точність та надійність ідентифікації пневмонії на ранніх стадіях за медичними зображеннями грудної клітини людини. Об’єктом дослідження є процес діагностування пневмонії за медичними зображеннями грудної клітини людини. Предметом дослідження є моделі, методи та засоби інформаційної технології для раннього діагностування пневмонії за медичними зображеннями грудної клітини людини. У дисертаційній роботі визначено актуальність застосування інформаційних технологій у галузі цифрового діагностування захворювань легень за медичними зображеннями грудної клітини. На основі проведено аналізу методів та підходів до виявлення пневмонії встановлено, що нейромережеві моделі є найкращим рішенням для розроблення інформаційної технології раннього діагностування. Досліджено методи для налаштування нейромережевої моделі та підходи до пояснення та інтерпретування результатів ідентифікації захворювання легень. За аналізом сучасних підходів, методів та інформаційних технологій для діагностування захворювання легень на ранніх стадіях за медичними зображеннями грудної клітини обґрунтовано потребу в створенні інформаційної технології раннього діагностування пневмонії.
The present thesis is devoted to solving the topical scientific and applied problem of automating the process of diagnosing viral pneumonia by medical images of the lungs through the development of information technology for early diagnosis of pneumonia by the individual selection of parameters of the classification model by medical images of the lungs. Applying the developed information technology for the early diagnosis of pneumonia in clinical practice by medical images of the human chest increases the accuracy and reliability of pneumonia identification in the early stages

This thesis addresses the extraction of medical knowledge from clinical text using deep learning techniques. In particular, the proposed methods focus on cancer clinical trial protocols and chest x-rays reports. The main results are a proof of concept of the capability of machine learning methods to discern which are regarded as inclusion or exclusion criteria in short free-text clinical notes, and a large scale chest x-ray image dataset labeled with radiological findings, diagnoses and anatomic locations. Clinical trials provide the evidence needed to determine the safety and effectiveness of new medical treatments. These trials are the basis employed for clinical practice guidelines and greatly assist clinicians in their daily practice when making decisions regarding treatment. However, the eligibility criteria used in oncology trials are too restrictive. Patients are often excluded on the basis of comorbidity, past or concomitant treatments and the fact they are over a certain age, and those patients that are selected do not, therefore, mimic clinical practice. This signifies that the results obtained in clinical trials cannot be extrapolated to patients if their clinical profiles were excluded from the clinical trial protocols. The efficacy and safety of new treatments for patients with these characteristics are not, therefore, defined. Given the clinical characteristics of particular patients, their type of cancer and the intended treatment, discovering whether or not they are represented in the corpus of available clinical trials requires the manual review of numerous eligibility criteria, which is impracticable for clinicians on a daily basis. In this thesis, a large medical corpora comprising all cancer clinical trials protocols in the last 18 years published by competent authorities was used to extract medical knowledge in order to help automatically learn patient’s eligibility in these trials. For this, a model is built to automatically predict whether short clinical statements were considered inclusion or exclusion criteria. A method based on deep neural networks is trained on a dataset of 6 million short free-texts to classify them between elegible or not elegible. For this, pretrained word embeddings were used as inputs in order to predict whether or not short free-text statements describing clinical information were considered eligible. The semantic reasoning of the word-embedding representations obtained was also analyzed, being able to identify equivalent treatments for a type of tumor in an analogy with the drugs used to treat other tumors. Results show that representation learning using deep neural networks can be successfully leveraged to extract the medical knowledge from clinical trial protocols and potentially assist practitioners when prescribing treatments. The second main task addressed in this thesis is related to knowledge extraction from medical reports associated with radiographs. Conventional radiology remains the most performed technique in radiodiagnosis services, with a percentage close to 75% (Radiología Médica, 2010). In particular, chest x-ray is the most common medical imaging exam with over 35 million taken every year in the US alone (Kamel et al., 2017). They allow for inexpensive screening of several pathologies including masses, pulmonary nodules, effusions, cardiac abnormalities and pneumothorax. For this task, all the chest-x rays that had been interpreted and reported by radiologists at the Hospital Universitario de San Juan (Alicante) from Jan 2009 to Dec 2017 were used to build a novel large-scale dataset in which each high-resolution radiograph is labeled with its corresponding metadata, radiological findings and pathologies. This dataset, named PadChest, includes more than 160,000 images obtained from 67,000 patients, covering six different position views and additional information on image acquisition and patient demography. The free text reports written in Spanish by radiologists were labeled with 174 different radiographic findings, 19 differential diagnoses and 104 anatomic locations organized as a hierarchical taxonomy and mapped onto standard Unified Medical Language System (UMLS) terminology. For this, a subset of the reports (a 27%) were manually annotated by trained physicians, whereas the remaining set was automatically labeled with deep supervised learning methods using attention mechanisms and fed with the text reports. The labels generated were then validated in an independent test set achieving a 0.93 Micro-F1 score. To the best of our knowledge, this is one of the largest public chest x-ray databases suitable for training supervised models concerning radiographs, and also the first to contain radiographic reports in Spanish. The PadChest dataset can be downloaded on request from http://bimcv.cipf.es/bimcv-projects/padchest/. PadChest is intended for training image classifiers based on deep learning techniques to extract medical knowledge from chest x-rays. It is essential that automatic radiology reporting methods could be integrated in a clinically validated manner in radiologists’ workflow in order to help specialists to improve their efficiency and enable safer and actionable reporting. Computer vision methods capable of identifying both the large spectrum of thoracic abnormalities (and also the normality) need to be trained on large-scale comprehensively labeled large-scale x-ray datasets such as PadChest. The development of these computer vision tools, once clinically validated, could serve to fulfill a broad range of unmet needs. Beyond implementing and obtaining results for both clinical trials and chest x-rays, this thesis studies the nature of the health data, the novelty of applying deep learning methods to obtain large-scale labeled medical datasets, and the relevance of its applications in medical research, which have contributed to its extramural diffusion and worldwide reach. This thesis describes this journey so that the reader is navigated across multiple disciplines, from engineering to medicine up to ethical considerations in artificial intelligence applied to medicine.

A grande utilização de exames de diagnóstico por imagem em crianças trouxe à tona a preocupação com a crescente dose de radiação absorvida na realização de um exame radiográfico. O objetivo desta pesquisa foi realizar avaliação das práticas de raios X na radiologia pediátrica, visando a otimização dos procedimentos radiológicos e a produção de imagens com qualidade para o diagnóstico com a menor dose ao paciente. A metodologia foi baseada no acompanhamento de exames pediátricos e medidas dosimétricas através do uso de dosímetros termoluminescentes TLDs e software específico (DoseCal) para a constatação da realidade dos serviços de radiologia pediátrica. Medidas de pacientes pediátricos em exames radiográficos de tórax foram realizadas em um hospital público de Curitiba e em uma clínica em Cascavel. Grupos com diferentes faixas etárias foram formados na avaliação de exames rotineiros de tórax nas projeções AP/PA e LAT, e ossos da face na projeção lateral, onde foram divididos em grupos de 0-1 ano, 1-5 anos, 5-10 anos e 10-15 anos. As doses obtidas através do software DoseCal foram comparadas entre si para determinar sua variabilidade. A DEP determinada pelos TLDs foi comparada com os valores de referência dados pela comunidade européia para verificar as doses utilizadas. Os valores de dose para crianças de até 1 ano apresentaram-se altos em comparação com os demais grupos avaliados, um fator justificado em partes pela limitação dos equipamentos utilizados. Na radiologia convencional os valores obtidos através dos TLDs foram satisfatórios, obedecendo a referência máxima descrita pela comissão européia. Na radiologia digital indireta obtivemos valores acima dos referenciados, fator este resultante da implantação e da adaptação das técnicas radiológicas a nova forma de captação de imagem. Concluí-se que o aprimoramento técnico das equipes em radiologia pediátrica é uma das melhores maneiras de se obter bons resultados na diminuição da dose.
The great use of examinations of diagnosis for image in children brought the concern with the increasing dose of radiation absorbed in the accomplishment of a radiographic examination. The objective of this research is to carry through evaluation of the practical ones of x-rays in pediatric radiology, aiming at to optimize the radiological rocedures and the production of images with quality for the diagnosis with the lesser dose to the patient. The methodology is based on the accompaniment of pediatric examinations and dosimetry measures through the use of dosemeters TLD and specific software (DoseCal) for the evidence of the reality in a radiology service. Measures of pediatric patients in radiographic examinations of thorax had been carried through in a public hospital in the Curitiba. Groups with different age groups had been formed in the evaluation of routine examinations of thorax in projections AP/PA and LAT, where they are divided in groups of 0-1 year, 1-5 years, 5-10 years and 10-15 years. Part of the carried through examinations had been evaluated with thermoluminescence dosemeters TLD-100 for the collection of the entrance surface dose (ESD). The measured doses are compared with the gotten ones with the DoseCal software, that makes the calculation of dose for each patient from the income of the device of rays X. The ESD is evaluated always that it has diagnostic quality in the radiographic image. The objective is to verify if the minimum requirements had been reached, for a good quality of image and bringing a small dose to the patient, as party to suit of to optimize procedures.

碩士
輔仁大學
統計資訊學系應用統計碩士在職專班
107
In recent years, the expeditious development of Artificial intelligence penetrates into all fields. We discover the application of artificial intelligence in every field, and in-depth learning is an emerging technology which has received much attention in recent years. It is a very important part in the field of medical image recognition. The convolutional neural network algorithm has achieved excellent results in the field of image recognition. However, the superparametric configuration rules of CNN, it is still time-consuming to rely on data scientists constantly adjusting the test during the modeling process. In this study, convolutional neural network and machine learning KNN algorithm are used to compare the chest X-ray images. The results show that the accuracy of CNN algorithm is better than that of KNN algorithm. In future, the original model needs to be retrained because of the increase of image samples. As a result, the original model needs to be retrained, so that analysts can speed up the retraining of the model and quickly find stable and efficient models. On the other hand, it is also hoped to find out the difficulties that may be encountered in the application of CNN algorithm in image recognition, for those who are interested in engaging in relevant research in the future.

碩士
國立臺灣大學
物理學研究所
104
Respiratory distress syndrome is one of the main cause of death for premature baby. Its symptoms include difficulty breathing, purple skin, expiratory moan and uncoordinated breathing. It’s mortality rate is highest in the first 48 hours. However, the symptoms are not shown immediately after birth. Some auxiliary examination is required for early detection of RDS. The auxiliary examination includes amniotic fluid examination, bubble test, PG examination, blood examination and chest x-ray(CXR) examination etc. All auxiliary examinations exclude CXR examination have specific numbers for classification, as the number of bubbles in bubble test, the molecule density in PG, amniotic and blood examination. CXR is a qualify but not quantify examination. The four grades classification standard used by NTUH are as follow: 1. If frosted glass pattern appear 2. If vessels are congested. 3. if the boundary between heart and lung, diaphragm and lung disappears. This research tries to numerically describe the RDS level of CXR, hoping to find a quantitative method for CXR classification. Medical image processing is not an easy work, especially for the case involving irregularity disease area. For RDS new born, there will be white out and boundary blurring effect on CXR. It makes most of the methods used for segmentation for adult CXR invalid. New approach must be developed. We use filter that are sensitive to ribs and use edge following technique to find the approximate region of thoracic cage. And by use of the property that spine nodes will be connected to ribs, to segment out the spine region by threshold. By mixing all approaches above, the lung area can be roughly segment out. After calculating the average and variation of selected regions, we see a correspondence (87% correctness) between the numerically grading and the qualitative grading which decided by doctors. It is concluded that it is possible to quantify RDS CXR by image processing.

碩士
國立交通大學
工業工程與管理系所
107
In recent years, rapid development of artificial intelligence in all fields, such as Drones, Smart home or autonomous cars, brought significant convenience to people's lives. However, advances in medical imaging techniques have completely changed the diagnosis method of medical images. Through the development of this technology, all traditional medical images gradually evolved from manual interpretation to digital assisting-interpretation. Therefore, the high development of artificial intelligence in medical field can not only assist physicians in disease diagnosis, but also boost the future development and innovation in medical field. Therefore, this study formed up a chest X-ray image classification model based on convolutional neural network. Actual chest X-ray images provided by Kaggle were used for modeling, testing and verifying the feasibility and effectiveness of the model constructed by this study. After verifying results in this study, the model in this study can provide an effective diagnostic technique for physicians in medical imaging diagnosis and expecting for more development of imaging technology in the medical field based on artificial intelligence.

no
Background Neonatal chest radiography is a frequently performed diagnostic examination, particularly in preterm infants where anatomical and/or biochemical immaturity impacts on respiratory function. However, the quality of neonatal radiographic images has been criticized internationally and a prevailing concern has been that radiographers (radiologic technologists) fail to appreciate the unique nature of neonatal and infant anatomical proportions. The aim of this study was to undertake a retrospective evaluation of neonatal chest radiography image acquisition techniques against key technical criteria. Methods Hundred neonatal chest radiographs, randomly selected from all those acquired in 2014, were retrospectively evaluated. Inclusion criteria for radiographs acquisition were as follows: anterior-posterior supine; within 30 days of birth; and with all preprocessed collimation boundaries visible. Image evaluation was systematically undertaken using an image assessment tool. To test for statistical significance, Student's t-test, χ2 test, and logistic regression were undertaken. Results Only 47% of the radiographs were considered straight in both upper and lower thoraces. The cranial collimation border extended beyond the upper border of the third cervical vertebra in 30% of cases, and the caudal border extended below the lower border of the first lumbar vertebra in 20% of cases, suggesting high possibility of neonatal overirradiation. Upper thorax rotation was significantly associated with head position (χ2 = 10.907; P < .001) as has been stated in many published textbooks internationally, but arm position had no apparent influence on rotation of the upper thorax (χ2 = 5.1260; P = .275). Birth weight was associated with accurate midline centering of central ray (logistic regression; OR = 1.0005; P = .009; CI, 1.00139–1.000957) with greater accuracy observed in images of neonates with higher birth weight. Conclusion This study has highlighted areas for neonatal chest radiography improvement. Importantly, the findings bring into question commonly advocated radiographic techniques relating to arm positioning and assessment of rotation while confirming the importance of other technical factors. These findings begin the work toward developing the evidence base to underpin neonatal chest radiograph acquisition, but further prospective work and multicenter/multinational data comparison are required to confirm the findings.

碩士
國立中正大學
資訊管理系醫療資訊管理研究所
106
We introduce an innovative application that combines 2 Faster Region with Convolution Neural Network (Faster R-CNN) It is applied to the automatic recognition of chest X-ray images. Because the shape and appearance of lung tumors vary greatly, radiologists have a lot of images to watch every day. Traditional methods cannot automatically and quickly identify tumor locations, but with deep learning, we only need to adapt to different tumor phases and cases through the training set. Compared to other traditional recognition methods, CNN is excellent in target recognition and it became the preferred algorithm in many target recognition challenge. Faster R-CNN uses the CNN to extract the image features, improves the region proposal method, shares the convolution feature with the Fast R-CNN, and makes the target detection and recognition almost instantaneous. This study details the Faster R-CNN and is used for chest X-ray images. We tested the clinical data provided by the doctors and the results confirmed that our approach achieved a certain level of accuracy in a fully automated challenge with very competitive execution time.

碩士
元智大學
資訊管理學系
99
In the past few years, malignant tumor has been a primary factory that which causes death. According to the statistics, lung cancer is the most frequently suffered disease among different cancers. In this paper, we develop a computer-aided detection (CAD) system for the detection of lung nodules in chest X-ray images. There are three main steps in this study. First, the lung areas are marked based on the histogram information. Second, we use the region growing method to find the contour of lung area. Finally, locations of nodule candidates are marked with circle detection method. Moreover, improper candidates are removed by the elimination criteria. Experimental results show that the average candidates are 47.7 per image with the accuracy of 82.47% to find lung nodules. In every subtle degree, the accuracy of the practicable degree and the hard degree were 86% and 76%, respectively. The proposed method showed the capability of enhancing the accuracy for the detection of subtle nodules.

碩士
國立臺南大學
資訊工程學系碩士班
106
Cancer is the top ten cause of death in Taiwan, and lung cancer is the highest among cancers. Early detection of cancer have a great impact on improving survival, but it is not easy to diagnose lung cancer early, often leading to delays in the golden age of treatment. In order to improve the detection rate of lung tumors in early stage, our study proposes a DANN(Domain-adversarial Neural Network) structure based on Faster RCNN(Regions with CNN features), and attempts to extract the characteristic information shared by obvious and nonobvious tumors in order to improve the detection of tumors. This experiment uses the JSRT database and the experimental results show that the proposed method is better than the faster RCNN.

碩士
國立交通大學
統計學研究所
107
Deep learning nowadays has attracted attention, especially in medical images classification because of its effectiveness and good performance that can compete with the medical images expert. Despite these successes there are the strong belief among experts that deep learning only efficient for the big datasets and for small datasets deep learning would produce a bad performance. For this study, it is aimed to build a deep learning model for image classification that can achieve high accuracy using chest x-ray images with relatively small dataset. We classify all normal chest x-ray images and all abnormalities in chest x-ray images into a binary classifier. We built and tested our model using the public dataset of Shenzen Hospital dataset. We use different type of input images based on different preprocessing and different type of learning technique so that the model can perform accurate classification for this particular dataset. Based on the result, pre-trained CheXNet with new trained fully connected network on cropped dataset can achieve the accuracy 0.8761, the sensitivity 0.8909, and the specificity 0.8621. The performance of the model also influenced by the certain area in the images, like other region outside the lung and black region outside the body.

碩士
國立臺中科技大學
資訊管理系碩士班
105
Digital image processing has been applied in medical domain widely, but the multitude still needs to manual processing. Automatic image segmentation and features analysis can assist doctor treatment and diagnose diseases more accurately, reduce the time of diagnosing and improve efficiency. Automatic medical image segmentation is difficult in that the image quality varied by equipment and dosage. In this thesis, the automatic method employed image multiscale intensity texture analysis and segmentation to surmount this problem. The proposed method automatically recognize and classify abnormal region without manual segmentation. Generally, automatic identification is based on the difference of the texture and organ shape, or any pathological changes of lung area. Therefore, the important features could be retained to identify abnormal areas. In this thesis, the chest x-ray images for finding whether lung region is healthy or not. The first proposed identifying common pneumothorax is based on SVM to classification method. Features are extracted from the lung image by the local binary pattern. Then, classification of pneumothorax lung is determined by support vector machines. The second proposed automatic pneumothorax detection is based on multiscale intensity texture segmentation. Remove the background and noises in the chest images for segmenting the lung of abnormal region. The segmenting the abnormal region. is used texture transforms from computing multiple overlapping blocks. Because the ribs boundaries are affected easily, the rib boundaries are identified by using Sobel edge detection. Finally, in order to obtain a complete disease region, the rib boundary is filled up in the rib boundary located between the abnormal regions.

碩士
元智大學
通訊工程學系
106
This study presents a deep learning method of segmenting lungs in chest X-ray image using encoder-decoder convolutional neural network. It also compares two modules as downsampling and upsampling algorithms, a bicubic interpolation over 4x4 pixel neighborhood and USEQ (Ultra-Fast Superpixel Extraction via Quantization). They are used as preprocessing modules to downsample the input image for segmentation and postprocessing module to upsample the network output to the original space for proper analysis. The experimental datasets consist of JSRT (Japanese Society of Radiological Technology), LIDC (Lung Image Database Consortium) and TMANH (Tainan Municipal An-Nan Hospital). Four measurement criteria were used in this research to determine the performance of the proposed method, Dice similarity coefficient (DSC), specificity, sensitivity, and Hausdorff distance. USEQ outperformed bicubic interpolation with an average score over individual lungs greater 95 for all the measurement criteria. An additional measurement criterion Jaccard index overlap (Ω) was used together with DSC to compare the proposed method to other segmentation algorithms that used the JSRT dataset. The proposed method is not only comparable to other methods with respect to mean scores of the two measurements but also achieved the most minimized bias-variance tradeoff. The result of the segmentation has proven efficient enough for the method to be applicable in real-world medical environments to bring ease in determining the area occupied by the lungs and some other medical diagnosis.

碩士
國立臺灣科技大學
醫學工程研究所
106
Image registration is important in medical applications accomplished with the advance of healthcare technology in recent years. Through the image registration task of finding the spatial relationship between input images, various studies have been proposed in the medical applications, including clinical track of events, updating the treatment plan for radiotherapy and surgery. This study presents a fully automatic registration system for chest X-ray images to generate fusion results for difference analysis. Through the accurate alignment of the proposed system, the fusion result indicates the difference of thoracic area during the treatment process. Registration of chest X-ray images is a challenging task due to variations on data appearance, imaging artifacts and complex data deformation problems, making existing registration approaches unstable and performs poor. The proposed method consists of a data normalization method, a hybrid L-SVM model to detect lungs, ribs and clavicles for object recognition, a landmark matching algorithm, two-stage transformation approaches, and a fusion method for difference analysis to highlight the difference of thoracic area. In evaluation, a preliminary test to compare three transformation models in the proposed system and a full evaluation process to compare the proposed method with two existing elastic registration methods for medical images have been conducted. The results show that the proposed method performs significantly better result than two benchmark methods (P value ≤ 0.001). The proposed system achieves the lowest mean registration error distance (MRED) (8.99 mm, 23.55 pixel) and the lowest mean registration error ratio (MRER) w.r.t. the length of image diagonal (1.61%) compared to the two benchmark approaches with MRED (15.64 mm, 40.97 pixel) and (180.5 mm, 472.69 pixel) and MRER (2.81%) and (32.51%), respectively.