Imaging Evidence Path

Who this book is for

This book is for technically competent readers entering medical imaging: software and data engineers, data scientists, quantitative researchers, healthcare-IT professionals, technically minded clinicians, and people building medical-imaging AI. It assumes programming and basic machine-learning and statistical literacy. It does not assume prior mastery of DICOM, image geometry, imaging physics, or radiomics. Those are part of what the book teaches.

The common starting point is not a particular job title. It is the need to reason across boundaries: between an image and the object that stores it, between an array and the physical space it represents, between a measurement and the procedure that produced it, and between a model result and the clinical question it is meant to inform.

What this book follows

The book follows the path from scanner to clinical claim. It asks what was measured, how that measurement was represented in DICOM, where it sits in space, which preprocessing decisions change it, how quantitative features are computed, how models are trained and validated, and how a result enters an imaging workflow. The chapters make the links between these layers explicit so that a reader can see where a reasonable technical choice may also be a choice about meaning.

The scope is broader than radiomics alone. CT and DICOM, geometry, segmentation, resampling, quantitative imaging, statistical learning, informatics, medical imaging AI, research methodology, MRI, multimodality, deployment, and lifecycle questions belong to the same evidence path. The book’s promise is not mastery of every specialty; it is a durable way to inspect the transitions on which an imaging result depends.

How to read the book

Readers who want a foundation can follow the accepted eleven-chapter route in Start here: CT, MRI, and DICOM, followed by geometry, quantitative measurement, radiomics, statistical learning and validation, medical imaging AI, evidence and claims, imaging informatics, and clinical deployment. This is a coherent route, not a required schedule.

Readers arriving with those foundations can enter through the chapter that matches their question. Imaging informatics and medical imaging AI follow how images, representations, and results move through systems. Research methodology provides a direct route into study design and critical reading. MRI and multimodality foundations extends the measurement discussion to other acquisition settings, while clinical deployment and the AI lifecycle follows a result toward use and monitoring.

The chapters are written as connected arguments, not as a required schedule or completion sequence. A chapter overview states the assumptions that matter for an independent entry. Read the prose first; use worked examples and practical tasks to test an idea against a real imaging or evidence question, and return to the reference sections when a definition, standard, or source needs checking. Prerequisites are guidance rather than gates.

One manuscript, several reading forms

The English manuscript in learning/ is the canonical teaching source. A structurally parallel Polish manuscript is translated from that source, and the Web, EPUB, and print forms render the same manuscript rather than maintaining separate editorial versions. The stable content is therefore the explanation in the manuscript, not a particular page layout or package.

Some linked material changes more quickly than the book: tool documentation, current standards editions, active literature, and operational project notes. Those links support verification and further work, but the stable chapters are written so that their central reasoning does not depend on a temporary version or an internal project artifact.