Front Matter

The editor acknowledges the enormous administrative effort from Dr. Kara Rader in making this book a reality.

Quantitative imaging (QI) is becoming useful in a wide range of medical research and clinical areas. This book and Quantitative Imaging in Medicine: Applications and Clinical Translation (2021) make the case that QI is a process demanding attention to the imaging hardware, the acquisition methods, image processing, and final display. The applications of QI are diverse. QI can be applied to the images formed in the laboratory environment, such as images created by a confocal laser scanning microscope attempting to measure fluorescence intensities to ascertain relative fluorophore concentrations for levels of protein expression in cells; or in a clinical setting, such as images created by a magnetic resonance scanner looking for malignant brain tumor expression. In each case, the QI process requires attention to detail. This attention is manifested through rigorous application of standards and standard operating procedures.

QI is the extraction of measurable features from medical images for the assessment of normal, or the severity, degree of change, or status of a disease, injury, or chronic condition relative to normal. This is a comprehensive definition provided by the Quantitative Imaging Biomarker Alliance (QIBA) of the Radiological Society of North America. Several important factors are implied here. First, there is the obvious assumption that the features within medical images can be isolated, so various aspects of these features can be measured quantitatively. Second, there is the implication that the measure of such features can be used to assess the condition of the tissue under investigation, be it normal or otherwise. That is, there is a relationship between the measured value of the chosen feature and the physical condition of the tissue. When the objective measurement of the selected feature is accurate and reproducible, and the relationship between its value and status of the tissue is strong, the measurement is a biomarker. A robust biomarker measurement enables the radiologist to distinguish between normal and diseased regions in the image, and that metric can be used across patients, imaging platforms, and time.

Biomarkers in medicine are classified as molecular biomarkers, cellular biomarkers, and imaging biomarkers. All three are important as clinical aids to patient diagnosis, prediction of therapy response, and prognosis. Imaging biomarkers and the QI methods presented in this book are closely related. The methods used to isolate relevant features in an image and assign measured values to them are commonly referred to as QI tools, and the resulting feature measurement is called a quantitative imaging biomarker (QIB).

Without generalizability of quantitative methods, the clinical value of QI is eliminated. Generalizability implies that a quantitative measurement made on a patient will provide accurate information on the health or condition of the patient, no matter when or where the measurement was made. It therefore requires the application of standards and quality control processes to ensure generalizability. Of course, standards in clinical medicine are not new with QI. What is new is the extent to which standards must be applied if quantitative feature measurements from the medical images are to have clinical meaning. Chapter 2, on standards, phantoms, and site qualification, presents strong arguments for the need and degree of imaging standards and quality assurance in biomedical imaging research and clinical applications. Standards reduce variability between images, and although such variability often has little or no consequence when providing some useful clinical information, in clinical trials, image variability can compromise the ability of the trial to achieve its objectives.

If imaging standards are the foundation of QI, then radiomics and deep learning (DL) are becoming the heart of the subject. There is no area of clinical imaging that has not at least tried image analysis using radiomics or DL. This is evident in the various chapters of this book. While radiomics warrants its own chapter to introduce and discuss the topic, it is also repeatedly brought up in other chapters as an important tool for extracting quantitative information from the specific images being discussed.

To validate that a measured value of a feature in an image or series of images corresponds to the disease status of a particular tissue, extensive data must be collected across a large patient population using different imaging devices. Of course, a gold standard, such as biopsy, must be used to determine the true status of the tissue, which is then catalogued with the quantitative measure of the feature or features. Patient outcome can also be correlated with the biomarker information to determine if there is prognostic value to the biomarker. The study of QI, then, has been the correlation of measured features in images with the diagnosis or prognosis of disease. If this process has any hope of succeeding, standards for data collection, image processing, and device operation and maintenance must be applied.

Imaging biomarkers are discussed in several chapters. For example, the visual presence of a lesion in an image of the lung serves as a visual biomarker of disease, and the measure of the size of that lesion during the course of therapy can serve as QIB for prognosis or therapy response. A reduction in the size or number of lesions can be seen as a biomarker for favorable outcome. Response evaluation criteria in solid tumors (RECIST), discussed in Chap. 4, presents the change of tumor size as a commonly used biomarker to assess treatment response. Many other QIBs are presented and discussed in this book.

Chapter 3, on QIBA protocols and profiles, presents a detailed discussion of the QIBA organization and its mission to bring QI to clinical utility. The case is made that if precision is to be a part of the measurement process, it must also be a part of the process to define operations used in QI. That is, if a QIB is to be useful in clinical trials, it must be properly defined and the procedures for its measurement must be clearly stated.

The extreme breadth of the field of QI is demonstrated, not only by the content of the chapters in this book, but also by the number of distinguished authors enlisted to tell the story of QI. QI is now being applied, in one degree or another, to all areas of clinical imaging. Its primary function is as a measure of response to therapy in clinical trials. The tools of QI are the means by which a region of interest is identified, and features within it are recognized and measured. Some of these features have direct anatomic connection; disease size, volume, location, and lesion number are examples. Other features have a connection with disease physiology; perfusion, diffusion, and glucose uptake are examples. Interestingly, with the growth of the areas of radiomics, DL, and artificial intelligence (AI), the measured features in QI can have little apparent connection with traditional visual or physiological features in the images. For example, image transform coefficients, such as wavelet and Fourier, or other methods, have been used to extract information from images to determine early detection or eventual outcome of disease.

All these topics are presented in this book but, as with any very broad area of discussion, a complete, in-depth presentation of any one topic is not possible. Therefore, there are ample references in each chapter to guide the reader to a more thorough study of QI in the specific chapter area. The book has been organized in two volumes. The first volume presents the various concepts and tools used in QI. Concepts include the requirement for standards in data collection and processing to ensure that the quantitative results obtained are representative of the nature of the tissue being imaged. Quantitative tools include the algorithms and models that are used to convert spatial image information into quantitative results.

Volume 2 focuses on the applications of the concepts and tools in QI. Each major clinical imaging modality is presented, showing how the tools presented in Volume 1 are implemented. Examples of the application of QI at various organ sites are also presented. Because certain imaging modalities tend to be preferred for specific organ sites (e.g. MRI for brain imaging, or CT for lung imaging), there is unavoidable repetition, which reinforces the significance of QI from the perspectives of image modality and organ site.

Finally, if quantitative methods are to become adapted in clinical trials and clinical standard of care, these tools and methods must be thoroughly validated in the clinical environment. This requires not only that the methods be accurate and robust as a biomarker, but the quantitative process must also fit into the overall clinical and administrative workflow. Ultimate commercialization of QI tools will require discussions with the Food and Drug Administration.

The study of QI in medicine is only just beginning. In the past ten to twelve years, the number of publications with the phrase “quantitative imaging” in the title has increased by an order of magnitude, and it is highly likely this exponential increase will continue for the next decade. Fueled by successes in AI and DL, the qualitative aspects of image interpretation will be augmented by quantitative inputs from computers. Quantitative methods in medicine will continue on a path of steady growth, but it is unlikely that computer diagnosis of disease will completely replace the need for human intervention.

Hugo J. W. L. Aerts

Director, Artificial Intelligence in Medicine (AIM) Program, Brigham and Women’s Hospital, and Associate Professor, Harvard Medical School, Boston, MA 02115

Firas S. Ahmed

Columbia University Medical Center, 622 W 168th Street, New York, NY 10032

Thomas L. Chenevert

University of Michigan, 1500 Medical Center Drive, Ann Arbor, MI 48109

Janet F. Eary

Associate Director DCTD/Cancer Imaging Program, National Cancer Institute, 9609 Medical Center Drive, Rockville, MD, 20850

Maryellen L. Giger

University of Chicago, Department of Radiology, MC 2026, 5841 South Maryland Avenue, Chicago, IL 60637

Annie (Ping) Gu

Emory University, 100 Woodruff Circle, Atlanta, GA 30332

Alexander R. Guimaraes

Chair, Quantitative Imaging Biomarkers Alliance, Vice Chair Research, Department of Diagnostic Radiology, Oregon Health & Sciences University, Portland, OR 97239

Jayashree Kalpathy-Cramer

Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02115

Paul E. Kinahan

University of Washington, 1959 NE Pacific St., Seattle, WA 98195

Dariya Malyarenko

University of Michigan, 1500 Medical Center Drive, Ann Arbor, MI 48109

Sandy Napel

Stanford University Department of Radiology, 318 Campus Drive, Stanford, CA 94305

Robert J. Nordstrom

Director, Quantitative Imaging Network, National Cancer Institute, 9609 Medical Center Drive, Rockville, MD, 20850

Lawrence H. Schwartz

James Picker Professor and Chairman, Radiology, Columbia University Irving Medical Center, 630 W 168th Street, New York, NY 10032

Ashish Sharma

Emory University, 100 Woodruff Circle, Atlanta, GA 30332

Jakob Weiss

Senior Resident, Radiology, Artificial Intelligence in Medicine Program, Harvard Medical School, Boston, MA 02115

Heather M. Whitney

University of Chicago, Department of Radiology, MC 2026, 5841 South Maryland Avenue, Chicago, IL 60637

Wheaton College, Department of Physics, 501 College Avenue, Wheaton, IL 60187