Medical image analysis

This book places an emphasis on the basic principles and recent advances in medical imaging technologies. After an introduction, the rest of the text can be categorized into three main parts. Part 1 provides a comprehensive tutorial on the fundamentals of medical imaging modalities and image processing methodologies. Part 2 highlights the cutting-edge imaging modalities for radiography and the recent developments of information technology in biomedicine. Part 3 provides some case studies and applications in radiation therapy and computer-aided diagnosis along with a perspective on future trends in medical and molecular imaging. The book is well organized and well presented. A concise summary before each chapter highlights the most important points of the text. This book can be used as a textbook for graduate courses and as an excellent reference for radiologists, neurologists, biomedical engineers, and researchers who are interested in imaging processing and modeling.

Drawing on existing theoretical work, as well as field research, this dissertation examines the impact of medical imaging technologies on communication between physicians and older patients when diagnostics often privilege disembodied data over the patient voice. Current diagnostic trends are contextualized within the history of medicine, from Ancient Greece to the present, including the development of imaging. Since the 1970s, advanced medical imaging technologies (e.g., ultrasound, computed tomography, magnetic resonance imaging) have become the diagnostic norm in Western medicine. The rapidity of this shift, which renders the human body as flattened data, can outstrip considerations of the implications of applying such technologies to living patients. Focusing on older patients, who may be less technologically savvy than younger patients or medical professionals, the field research begins with semi-structured interviews of patients over age sixty-five, exploring their encounters with medical imaging equipment and professionals. This data is interrogated qualitatively using Foucauldian discourse analysis drawing on Andrea Doucet’s model of slow scholarship, and informed by Arthur Frank’s notion of letting stories breathe; themes were allowed to surface from the patients’ narratives, rather than imposed by the researcher. Information emerging from the data considers patients’ emotions, unexpected physical sensations, communicative strategies and rationalizations, as well as Foucauldian allusions to power. Observational research was also conducted during encounters between physicians and simulated patients in the presence of medical images; these encounters were followed by reflective exit interviews. Research indicates that although physicians are increasingly trained in patient-centred communication, it is not always optimally practised. Physicians are sometimes more comfortable with the medical discourse of disease than with the emotional, metaphoric language of the patient’s illness experience. Since the development of modern Western medicine in Europe of the late 1700s, physicians have been trained to seek pathology, with the increasing aid of medical technologies, rather than listening to their patients. For older patients, who may experience multiple co-morbidities, the lack of communication around advanced medical technologies can increase their sense of vulnerability and anxiety. The dissertation concludes with recommendations for both patients and practitioners to improve communication in the medical context.

Automatic change detection methods for identifying the changes of serial MR images taken at different times are of great interest to radiologists. The majority of existing change detection methods in medical imaging, and those of brain images in particular, include many preprocessing steps and rely mostly on statistical analysis of magnetic resonance imaging (MRI) scans. Although most methods utilize registration software, tissue classification remains a difficult and overwhelming task. Recently, dictionary learning techniques are being used in many areas of image processing, such as image surveillance, face recognition, remote sensing, and medical imaging. We present an improved version of the EigenBlockCD algorithm, named the EigenBlockCD-2. The EigenBlockCD-2 algorithm performs an initial global registration and identifies the changes between serial MR images of the brain. Blocks of pixels from a baseline scan are used to train local dictionaries to detect changes in the follow-up scan. We use PCA to reduce the dimensionality of the local dictionaries and the redundancy of data. Choosing the appropriate distance measure significantly affects the performance of our algorithm. We examine the differences between [Formula: see text] and [Formula: see text] norms as two possible similarity measures in the improved EigenBlockCD-2 algorithm. We show the advantages of the [Formula: see text] norm over the [Formula: see text] norm both theoretically and numerically. We also demonstrate the performance of the new EigenBlockCD-2 algorithm for detecting changes of MR images and compare our results with those provided in the recent literature. Experimental results with both simulated and real MRI scans show that our improved EigenBlockCD-2 algorithm outperforms the previous methods. It detects clinical changes while ignoring the changes due to the patient's position and other acquisition artifacts.

Advances in medical imaging technology have led to the acquisition of large number of images in different modalities. On some of these images the boundaries of key organs need to be accurately identified for treatment planning and diagnosis. This is typically performed manually by a physician who uses prior knowledge of organ shapes and locations to demarcate the boundaries of organs. Such manual segmentation is subjective, time consuming and prone to inconsistency. Automating this task has been found to be very challenging due to poor tissue contrast and ill-defined organ/tissue boundaries. This dissertation presents a genetic algorithm for combining representations of learned information such as known shapes, regional properties and relative location of objects into a single framework in order to perform automated segmentation. The algorithm has been tested on two different datasets: for segmenting hands on thermographic images and for prostate segmentation on pelvic computed tomography (CT) and magnetic resonance (MR) images. In this dissertation we report the results of segmentation in two dimensions (2D) for thermographic images; and two as well as three dimensions (3D) for pelvic images. We show that combining multiple features for segmentation improves segmentation accuracy as compared with segmentation using single features such as texture or shape alone.

Thanks to advances in medical imaging technologies and numerical methods, patient-specific modelling is more and more used to improve diagnosis and to estimate the outcome of surgical interventions. It requires the extraction of the domain of interest from the medical scans of the patient, as well as the discretisation of this geometry. However, extracting smooth multi-material meshes that conform to the tissue boundaries described in the segmented image is still an active field of research. We propose to solve this issue by combining an implicit surface reconstruction method with a multi-region mesh extraction scheme. The surface reconstruction algorithm is based on multi-level partition of unity implicit surfaces, which we extended to the multi-material case. The mesh generation algorithm consists in a novel multi-domain version of the marching tetrahedra. It generates multi-region meshes as a set of triangular surface patches consistently joining each other at material junctions. This paper presents this original meshing strategy, starting from boundary points extraction from the segmented data to heterogeneous implicit surface definition, multi-region surface triangulation and mesh adaptation. Results indicate that the proposed approach produces smooth and high-quality triangular meshes with a reasonable geometric accuracy. Hence, the proposed method is well suited for subsequent volume mesh generation and finite element simulations.

Health depiction and image are indispensable for patient instruction and self-care. Images characterize an imperative feature of the entire medical knowledge. Presently there is hardly any attempt in automatically construction inclusive image bases for universal health information customers. We proposed an approach to automatically construct a consumer -leaning image library contain images of individual organs, diseases, drugs and other medical entities. Health check imaging has become an important tool not simply in document patient presentation and clinical findings, but also considerate and managing various diseases. Image data present tangible visual evidence of disease manifestation. The number of digital images that needs to be acquired, analyzed, classified, stored and retrieved in the medical centers is exponentially growing with the advances in medical imaging technology. The goal of this work we proposed medical image retrieval technique for pathology images that implements recent improvements in feature representation, efficient indexing, and similarity matching.

In the last few years, with the rapid development of artificial intelligence (AI) technology, its application in the medical field has become more and more in-depth and extensive, especially the research of computer vision technology in medical image analysis and recognition has achieved many results. It is recognized by experts engaged in imaging diagnosis in the medical field, which has provided a great boost to the development of precision medicine. However, the application and development of artificial intelligence in the field of medical imaging is still in its infancy, and faces many technical and ethical challenges, such as data quality issues, data security issues and clinical application issues. The application of AI technology in medical imaging needs to be further adjusted and improved in combination with social development and actual needs. In this paper, we discussed the usages of artificial intelligence in medical imaging and provided new sights for the sustainable development and improvement in this area, as well as the challenges faced by others.

The fields of medical imaging and medical imaging instrumentation are increasingly important. The state-of-the-art continues to advance at a very rapid pace. In fact, various medical imaging modalities are under development at the National Synchrotron Light Source (such as MECT and Transvenous Angiography.) It is important to understand how these techniques compare with today`s more conventional imaging modalities. The purpose of this report is to provide some basic information about the various medical imaging technologies currently in use and their potential developments as a basis for this comparison. This report is by no means an in-depth study of the physics and instrumentation of the various imaging modalities; instead, it is an attempt to provide an explanation of the physical bases of these techniques and their principal clinical and research capabilities.

Bladder cancer is reported to be the fifth leading cause of cancer deaths in the United States. Recent advances in medical imaging technologies, such as magnetic resonance (MR) imaging, make virtual cystoscopy a potential alternative with advantages as being a safe and non-invasive method for evaluation of the entire bladder and detection of abnormalities. To help reducing the interpretation time and reading fatigue of the readers or radiologists, we introduce a computer-aided detection scheme based on the thickness mapping of the bladder wall since locally-thickened bladder wall often appears around tumors. In the thickness mapping method, the path used to measure the thickness can be determined without any ambiguity by tracing the gradient direction of the potential field between the inner and outer borders of the bladder wall. The thickness mapping of the three-dimensional inner border surface of the bladder is then flattened to a twodimensional (2D) gray image with conformal mapping method. In the 2D flattened image, a blob detector is applied to detect the abnormalities, which are actually the thickened bladder wall indicating bladder lesions. Such scheme was tested on two MR datasets, one from a healthy volunteer and the other from a patient with a tumor. The result is preliminary, but very promising with 100% detection sensitivity at 7 FPs per case.

National cancer institute initiative for development of novel imaging technologies

Early classification of brain tumors is the key to effective treatment. With advances in medical imaging technology, automated classification algorithms face challenges due to tumor diversity. Although Swin Transformer is effective in handling high-resolution images, it encounters difficulties with small datasets and high computational complexity. This study introduces SparseSwinMDT, a novel model that combines sparse token representation with multipath decision trees. Experimental results show that SparseSwinMDT achieves an accuracy of 99.47% in brain tumor classification, significantly outperforming existing methods while reducing computation time, making it particularly suitable for resource-constrained medical environments.

The main objective of this PhD research has been to utilize machine learning techniques on near infrared spectroscopy (NIRS) signals, for the development of highly accurate and clinically practical biomarkers for the objective assessment of pain perception. While advances in medical imaging technology have significantly improved the scientific knowledge in regards to the brain's response to noxious stimuli, there remains an unmet clinical need for a practical, inexpensive tool for the reliable and objective assessment of pain perception. Even though functional imaging modalities such as fMRI and PET scans deliver superior spatial information, they are not readily accessible for routine clinical use. On the other hand NIRS is non-invasive, safe, portable and affordable with a short setup time. These features make NIRS ideal for clinical applications. In this thesis we used the cold pressor test to induce different levels of pain in healthy subjects while the NIRS signal was recorded from the frontal regions of the brain. We extracted 54 features from each dataset and used machine learning techniques, logistic regression and support vector machine, to classify the signals based on the self-reported pain scores. To select the model for machine learning, we developed our feature selection algorithm based on a RFE-SVM (recursive feature elimination - support vector machine) method to find subsets of feature space with the highest classification capability. Through this process we identified a subset of 10 features which could distinguish high-pain from low-pain stimuli with an accuracy of 85% (Leave-one-out cross validation). Moreover we applied functional data analysis on the collected NIRS data and converted discrete samples to continuous curves. This time we used the same RFE-SVM method on the coefficients of fDA bases (as opposed to extracted features) and we achieved 94% of accuracy to classify low-pain high-pain signals. Then using machine learning techniques (k-means and hierarchical clustering) we found clusters in the data which covered low pain and high pain groups with an accuracy of 91.2%. The center of these clusters can represent the prototype NIRS response of that pain level. Our approaches provided trial-by-trial predictions of pain level from NIRS measurement for each individual (as opposed to methods based on responses averaged across many trials and subjects), and thus, represent a step towards the goal of establishing an objective clinical bio marker of pain perception. Further refinement of proposed methods, including incorporating more datasets and employing other noxious stimuli, is required to make the NIRS technique a powerful clinical tool for pain assessment.

The coming years will be an ideal time for investigators in radiology to develop a comparative effectiveness research (CER) initiative and to establish a lasting precedent for CER in our field.

Purpose:Recent advances in medical imaging technologies provide opportunities to quantify the tumor phenotype throughout the course of treatment non‐invasively. The emerging field of Radiomics addresses this by converting medical images into minable data using a large number of quantitative imaging algorithms. The ability of radiomics to predict clinical outcomes such as survival or nodal involvement, can provide valuable complementary information. This study aims to evaluate the predictability of CT based radiomics features for node stages and survival time.in non‐small cell lung cancer (NSCLC) patients.Method:We included 118 patients with NSCLC in our analysis. Seventy‐four quantitative radiomic features describing the tumor phenotype based on intensity, shape, and texture, were extracted from planning CT scans. Association of the radiomics features with node involvement (Yes/No) was assessed using the area under the curve (AUC) of the receiver operating characteristics (ROC). Association with overall survival was assessed using the C‐Index (CI).Results:We found 30 radiomic features that were significantly (FDR corrected) associated with nodal involvement (range AUC=0.73‐0.60). Seventeen features were significantly associated survival (range CI=0.56‐0.62). Gray Level Non‐Uniformity, describing intra‐tumor heterogeneity, was the best predictor for both nodal stage (AUC=0.73) and survival time (CI=0.62). It is noteworthy that tumor volume is only the 8th best predictor for both nodal stage (AUC=0.71) and survival (CI =0.61).Conclusion:Radiomics analysis was able to predict node stage and survival time by looking at complex image features. Although tumor volume has been routinely used to predict outcome, this study shows that radiomic features capturing detailed information of the tumor phenotype can provide better prediction for clinical relevant factors such as nodal involvement and survival.

The CSIRO (Commonwealth Scientific and Industrial Research Organisation) and the Queensland Government have jointly established the Australian e-Health Research Centre (AEHRC) with the aim of developing innovative information and communication technologies (ICT) for a sustainable health care system. The AEHRC, as part of the CSIRO ICT Centre, has access to new technologies in information processing, wireless and networking technologies, and autonomous systems. The AEHRC's 50 researchers, software engineers and PhD students, in partnership with the CSIRO and clinicians, are developing and applying new technologies for improving patients' experience, building a more rewarding workplace for the health workforce, and improving the efficiency of delivering health care. The capabilities of the AEHRC fall into four broad areas: smart methods for using medical data; advanced medical imaging technologies; new models for clinical and health care interventions; and tools for medical skills development. Since its founding in 2004, new technology from the AEHRC has been adopted within Queensland (eg, a mobile phone-based cardiac rehabilitation program), around Australia (eg, medical imaging technologies) and internationally (eg, our clinical terminology tools).