Practical and Clinical Medical Imaging Informatics and Artificial Intelligence PDF

IMAGING SCIENCE AND INFORMATICS CHAPTER V ========================================= Practical imaging informatics is the study of information technology (IT) and clinical image management in the field of medical imaging. It\'s a branch of biomedical informatics that focuses on the development, app...

IMAGING SCIENCE AND INFORMATICS CHAPTER V ========================================= Practical imaging informatics is the study of information technology (IT) and clinical image management in the field of medical imaging. It\'s a branch of biomedical informatics that focuses on the development, application, and assessment of IT for clinical medical imaging. Here are some things to know about practical imaging informatics: What it covers -------------- Practical imaging informatics covers a wide range of topics, including image creation, processing, storage, distribution, and security. It also includes workflow, standards, interoperability, quality assurance, customer relations, user training, compliance, and billing. ENTERPRISE DIAGNOSTIC MEDICAL IMAGING MODALITIES ================================================ Enterprise imaging is a set of strategies, initiatives, and workflows implemented across a healthcare enterprise to consistently and optimally capture, index, manage, store, distribute, view, exchange, and analyze all clinical imaging and multimedia content to enhance the electronic health record. For some audiences, "enterprise imaging" refers to consolidating various hospitals' or imaging centers' radiology departments into a unified imaging system, aiding in image transmission for interpretation. Others see it as integrating medical images into a centralized archive often linked with the electronic medical record (EMR). - X-ray scanners - Computed tomography (CT) scanners - Magnetic resonance imaging (MRI) - Nuclear medicine - Ultrasound  Users within the medical enterprise have protected access through a LAN. Also depicted are an "offsite" backup archive for disaster recovery and real-time customer care monitoring to provide around the clock support. A mirror archive provides on-site backup within the enterprise firewall with immediate availability in case of failure of the primary archive OPEN-SOURCE SOFTWARE VS PROPRIETARY SOFTWARE ============================================ The main difference between open-source software and proprietary software is that open-source software is freely available without restrictions, while proprietary software requires payment and has restrictions on use. Software for RIS ---------------- Radiology Information System (RIS) software is a tool that manages and stores medical imaging data, as well as patient data, scheduling, and reporting. RIS software can help with billing, automate tasks, and improve the imaging process. Some popular RIS software options include: Open Source RIS softwares: -------------------------- - **KloudRIS** - A web-based RIS that\'s multilingual, multi-tenant, and open source. It\'s designed to manage radiology workflows from any device, location, or browser. - **LibreHealth RIS -** An open-source RIS that was developed in collaboration with OpenMRS. It\'s used by hospitals and clinics in multiple countries. - **ThaiRIS -** An open-source radiology information system. Proprietary RIS softwares: -------------------------- - **MedInformatix RIS -** Includes scheduling, registration, workflow, and tracking functionality, as well as revenue cycle management. It can also integrate with third-party solutions for digital dictation, voice recognition, and transcription. - **Optomed RIS** - Integrates with imaging equipment, image archives, patient information systems, and other hospital information systems. It can also transfer statistical data to other reporting systems. - **RISynergy -** Offers optional modules like biometrics-based login, appointment reminder, and DICOM worklist manager. It\'s designed for practices with as few as three doctors on staff. - **Centricity RIS -** Includes tools like rules based booking templates and color to help streamline scheduling and prioritize medical appointments and procedures. - **MedicsRIS -** Includes electronic prior authorizations, insurance discovery options, out-of-network alerts, eligibility verifications, and proactive denial alerts. When choosing a RIS vendor, you can consider things like: Integration with other healthcare systems, Customizability and scalability, Vendor support and customer service, and Costs and return on investment. PACS VS DICOM ============= DICOM is a standard for medical image file formats and communication, while PACS is a system for storing, retrieving, and sharing medical images: DICOM ===== - An international standard for medical imaging data exchange that defines the file format for medical images. DICOM files have the file extension \".dcm\" and can also accept other file formats like JPEG, TIFF, GIF, and PNG. DICOM compression can be lossy or lossless, with lossy compression irreversibly losing imaging data. PACS ==== - A system for managing and storing medical images in a centralized database. PACS can store images produced by various medical hardware modalities, such as X-ray machines, CT scans, MRI scans, and ultrasound machines. PACS storage can be online (cloud storage) or offline (on-premises). RIS VS HIS ========== A radiology information system (RIS) and a hospital information system (HIS) are both important parts of a healthcare facility\'s information technology infrastructure, but they have different functions: RIS === - A radiology-specific electronic health record (EHR) that allows radiologists to store, edit, and report on patient imaging. RIS can also be used for patient scheduling, billing, and generating reports. HIS === - A comprehensive system that manages a hospital\'s administrative, financial, legal, and medical aspects. HIS maintains a patient\'s electronic medical record (EMR), which includes vitals, treatment, and other information. HIS also provides access to patient information, billing information, and reports from other services. ARTIFICIAL INTELLIGENCE (AI) ============================ Artificial intelligence (AI) has been defined by some as the \"branch of computer science dealing with the simulation of intelligent behavior in computers\", however, the precise definition is a matter of debate among experts. An alternative definition is the branch of computer science dedicated to creating algorithms that can solve problems without being explicitly programmed for all the specificities of the problems. AI algorithms and in particular deep learning (part of [machine learning](https://radiopaedia.org/articles/machine-learning-1?lang=us)) aim to either assist humans with solving a problem or solve the problem without human input. The exponential increase in computational processing and memory capability has opened up the potential for AI to handle much larger datasets, including those required in radiology. History and etymology --------------------- The term artificial intelligence is credited to John McCarthy, a mathematician (and the creator of the LISP programming language) who proposed and organized a summer research conference that happened in 1956 at Dartmouth on artificial Intelligence, who used the term. The conference is considered by many to be the moment that AI was founded as an area of academic research, however, it could be argued that the creation of the field began earlier with Alan Turing, who developed the Turing test, or even before. Turing Test The Turing test is a thought experiment that assesses a machine\'s ability to mimic human intelligence. The test was proposed in 1950 by Alan Turing, an English mathematician, computer scientist, and cryptanalyst. The test is still used today to study artificial intelligence (AI) and to better understand how AI interacts with humans. The Turing Test involves three participants: - The interrogator: Asks questions to the other two participants - The human: One of the participants who answers the interrogator\'s questions - The computer: The other participant who answers the interrogator\'s questions The interrogator\'s goal is to determine which participant is the human and which is the computer. The computer\'s goal is to trick the interrogator into thinking it is the human. The test is repeated multiple times. If the interrogator is unable to determine which participant is human, the computer is considered the winner. The Turing Test is a useful tool for studying how machines interact with humans and for defining intelligence and thinking. It\'s also important for understanding the capabilities of artificial intelligence (AI) as it becomes more integrated into our lives. The term AI encompasses numerous specific areas and approaches, including: COMPUTER AIDED DIAGNOSIS (CAD) ============================== Computer aided diagnosis (CAD) is the use of a computer- generated output as an assisting tool for a clinician to make a diagnosis. It is different from automated computer diagnosis, in which the end diagnosis is based on a computer algorithm only. As an early form of artificial intelligence, computer aided diagnosis systems have been used extensively within radiology for many years. The most common applications are for detection of [breast cancer](https://radiopaedia.org/articles/breast-neoplasms?lang=us) on [mammography](https://radiopaedia.org/articles/mammography?lang=us) and of [pulmonary nodules](https://radiopaedia.org/articles/pulmonary-nodule-1?lang=us) on [chest CT](https://radiopaedia.org/articles/computed-tomography-of-the-chest?lang=us). These systems traditionally relied on manual feature engineering based on domain knowledge, but newer approaches are employing [machine learning](https://radiopaedia.org/articles/machine-learning-1?lang=us) to discover latent features within imaging data. The term is often used broadly for both computer-aided detection and computer-aided diagnosis: - computer-aided detection (CADe): marks specific areas of images that may seem abnormal, designed to reduce the risk of missing pathologies of interest - computer-aided diagnosis (CADx): helps a practitioner assess and classify pathology in medical images History ------- Research into CAD began in the mid-1980s at the University of Chicago. How it works --------------------------------- CAD analyzes medical images, such as X-rays, MRIs, and CT scans, to identify abnormalities or signs of disease that might be missed by the human eye. CAD systems use complex algorithms to process the images and highlight areas of interest. How it\'s used -------------- CAD is a \"second opinion\" that helps radiologists and other medical professionals make more accurate diagnoses and provide better patient care. CAD systems are regularly updated to keep up with new medical developments. MACHINE LEARNING ================ History and etymology --------------------- Arthur Samuel first defined the term \"machine learning\" in 1959, a component in artificial intelligence where a computer learns from a set of data to improve its future performances Machine learning is a specific practical application of computer science and mathematics that allows computers to extrapolate information based on observed patterns without explicit programming. A defining characteristic of machine learning programs is the improved performance at tasks such as classification when more data, known as training data, is processed.  Machine learning is becoming an increasingly important tool in the medical profession for primary computer-aided diagnosis algorithms and decision support systems. Interest in the practical applications of machine learning, including applications for imaging, is high. The availability of large scale data sets, substantial advances in computing power, and [deep-learning](https://radiopaedia.org/articles/deep-learning?lang=us) algorithms have driven this area forward substantially since it began in the 1950s. NATURAL LANGUAGE PROCESSING =========================== Natural language processing (NLP) is an area of active research in artificial intelligence concerned with human languages. Natural language processing programs use human written text or human speech as data for analysis. The goals of natural language processing programs can vary from generating insights from texts or recorded speech to generating text or speech. The first area of natural language processing to gain wide usage in radiology was speech recognition. In earlier literature, speech recognition was often referred to as voice recognition, but the trend in nomenclature is towards differentiating voice recognition and speech recognition, with only the latter implying the use of dictated recordings to create reports. In many radiology practices, radiologists use speech recognition programs to create reports routinely. Areas of active research for the application of natural language processing in radiology include areas of natural language understanding (NLU) such as topic modelling, other forms of information extraction and keyword searching. Natural language processing also includes natural language generation (NLG). Large language models --------------------- Traditionally natural language processing utilized recurrent neural networks, until around 2017 when researchers from Google Bain published a paper exploring the use of transformers. Transformers improve the efficiency of the algorithm via its ability to focus on different parts of the input sequence while encoding or decoding it. Generative pretrained transformer models such as ChatGPT have made a significant impact on natural language processing due to its efficiency and ability to replicate human writing RULE-BASED EXPERT SYSTEMS ========================= A rule-based expert system is the simplest form of artificial intelligence and uses prescribed knowledge- based rules to solve a problem. The aim of the expert system is to take knowledge from a human expert and convert this into a number of hardcoded rules to apply to the input data. In their most basic form, the rules are commonly conditional statements (if a, then do x, else if b, then do y). These systems should be applied to smaller problems, as the more complex a system is, the more rules that are required to describe it, and thus increased difficulty to model for all possible outcomes. Note: with problems related to radiological images, often preprocessing of the images is required prior to the expert system being applied.  Example ------- A very basic example of rule-based expert system would be a programme to direct the management of abdominal aneurysms. The system would input the diameter of an aneurysm. Using conditional arguments, the input diameter would be stratified to recommend whether immediate intervention was required, and if not what appropriate follow up is recommended. RADIOMICS =============================== Radiomics (as applied to radiology) is a field of medical study that aims to extract a large number of quantitative features from medical images using data characterization algorithms. The data is assessed for improved decision support. It has the potential to uncover disease characteristics that are difficult to identify by human vision alone. Process ------- The process of creating a database of correlative quantitative features, which can be used to analyze subsequent (unknown) cases includes the following steps: Initial image acquisition and processing ---------------------------------------- Using a variety of reconstruction algorithms such as contrast, edge enhancement, etc. influences the quality and usability of the images, which in turn determines how easily and accurately an abnormal characteristic could be detected and characterized. Region of interest segmentation ------------------------------- Identify/create areas (2D images) or volumes of interest (3D images). Can be done either manually, semi- automated, or fully automated using artificial intelligence. NOTE: many tools allow manual checking and adjustment of automated outputs, which is recommended especially with tools that use atlas based segmentation. Features extraction and qualification ------------------------------------- Features include volume, shape, surface, density, and intensity, [texture](https://radiopaedia.org/articles/texture-analysis?lang=us), location, and relations with the surrounding tissues. **Semantic features** are those that are commonly used in the radiology lexicon to describe regions of interest. **Agnostic features** are those that attempt to capture lesion heterogeneity through quantitative mathematical descriptors. Examples of semantic features: ------------------------------ - shape - location - vascularity - spiculation - necrosis - attachments Equivalent examples of agnostic features: ----------------------------------------- - kurtosis or skewness (of the image histogram) - Haralick textures - Laws textures - wavelets - Laplacian transforms - Minkowski functions - fractal dimensions Uses ---- Radiomics can be applied to most imaging modalities including radiographs, ultrasound, CT, MRI and PET studies. It can be used to increase the precision in the diagnosis, assessment of prognosis, and prediction of therapy response, particularly in combination with clinical, biochemical, and genetic data. The technique has been used in oncological studies, but potentially can be applied to any disease. A typical example of radiomics is using [texture analysis](https://radiopaedia.org/articles/texture-analysis?lang=us) to correlate molecular and histological features of diffuse high-grade gliomas. REDUCTION OF NOISE [(NOISE REDUCTION)](https://radiopaedia.org/articles/noise-reduction-1?lang=us) AND OPTIMIZATION OF IMAGE ACQUISITION ======================================================================================================================================== Noise reduction --------------- Noise reduction, also known as noise suppression or denoising, commonly refers to the various algorithmic techniques to reduce noise in [digital](https://radiopaedia.org/articles/digital-image?lang=us) [images](https://radiopaedia.org/articles/digital-image?lang=us) once they are created although a few sources use the term more broadly to imply anything that reduces [noise.](https://radiopaedia.org/articles/noise?lang=us) In digital image processing various techniques, most of which are filtering techniques are applied to images at various stages after acquisition. These methods can involve both spatial filters ([convolutions](https://radiopaedia.org/articles/kernel-image-reconstruction-for-ct-1?lang=us)), frequency filters ([discrete](https://radiopaedia.org/articles/fourier-transform?lang=us) [Fourier transform](https://radiopaedia.org/articles/fourier-transform?lang=us)), morphological filters or even statistical filters. Practical points ---------------- Many radiologists are familiar with [CT reconstruction kernels](https://radiopaedia.org/articles/kernel-image-reconstruction-for-ct-1?lang=us) and using a smooth kernel would be an example (in most cases) of noise reduction. The use of CT reconstruction kernels is often a choice, but some noise reduction techniques are automated and performed on raw imaging data without the radiologist necessarily being aware of it. Advanced noise reduction techniques are considered part of [AI](https://radiopaedia.org/articles/artificial-intelligence?lang=us). AUTOMATIC EXPOSURE CONTROL (AEC) ================================ Automatic exposure control (AEC) is a device in radiology that automatically terminates exposure to X- rays or other radiation when a predetermined amount of radiation has been reached. The goal of AEC is to consistently produce high-quality images with minimal technical factors set by the radiographer. AEC systems can also adjust other exposure factors, such as milliamperage, kilovoltage, or exposure time. They can be used in radiography, fluoroscopy, and CT scans. AEC systems work by sensing the amount of radiation in front of the image receptor. They can help reduce \"dose creep\", which is when a technologist accidentally overexposes a patient to radiation. However, AEC systems can also result in high patient doses, especially with digital image receptors. Radiographers should still use their judgment to select the appropriate kVp, mA, image receptor, and grid when using AEC systems. They should also understand how AEC systems work, how to compensate for variations, and how to follow quality control procedures. -End