It is not. The image raised a probability. It did not settle one.
A 64-year-old woman has a 2-centimeter mass in the tail of her pancreas, found on a CT ordered for back pain. The radiologist's report reads: suspicious for adenocarcinoma. Her family hears the word and assumes the diagnosis is made. It is not. The scan was never the diagnosis. It was a probability. That distinction is what this chapter is about.
Imaging raises or lowers probability. Pathology is ground truth — but only when the tissue is the right tissue.
The diagnostic chain in oncology has five links: clinical suspicion, imaging, biopsy, pathology, molecular workup. Each link can break in its own way. Imaging is probabilistic. Biopsy samples a tiny fraction of a mass. Pathology is the ground truth — but only if the sample came from the right place. Understanding where each link fails is what lets you read a report correctly.
Neither modality eliminates the false-positive rate. Neither eliminates the false-negative rate.
CT reconstructs anatomy from rotating X-rays. It is the workhorse for detecting masses and for measuring treatment response under RECIST — the standardized rules that allow tumor size to be compared across scans. Its soft-tissue contrast is limited, and it exposes patients to roughly 8 to 15 millisieverts. MRI uses magnetic fields and has superior soft-tissue contrast for the brain, pelvis, and breast — at the cost of lower specificity and more false positives. Neither modality eliminates the false-positive or the false-negative rate.
Ultrasound uses sound waves — real-time, radiation-free, excellent for guiding needles to superficial lesions. Its weakness is operator-dependence: the same lesion looks different in different hands, and gas and bone block transmission entirely. PET-FDG images metabolic function. Cancer cells with high glucose uptake trap the tracer. But so does inflammation, healing surgical sites, and brown fat. And some tumors — mucinous cancers, some lymphomas — take up FDG poorly and appear cold. A hot spot raises probability. A cold spot lowers it. Neither is a diagnosis.
Malignancy is defined by what cells look like under a microscope. The pathology report — not the imaging report — is the foundation document for treatment.
An image shows a mass. It does not show the cells inside it. Malignancy is defined by what cells look like, how they are organized, whether they have invaded through basement membranes. These features are visible only under a microscope. The standard preparation is FFPE — formalin-fixed, paraffin-embedded — tissue sectioned to a few micrometers and stained with hematoxylin and eosin. The pathology report is the foundation document for treatment. Everything else establishes what to do about it.
Morphology tells you what kind of thing it is. IHC tells you which one.
Immunohistochemistry extends the morphological reading. Antibodies against specific proteins stain where the protein is present. TTF-1 identifies lung or thyroid origin among adenocarcinomas. Estrogen and progesterone receptor status determines whether hormonal therapy applies. HER2 overexpression determines trastuzumab eligibility. PD-L1 expression informs checkpoint inhibitor selection. Mismatch-repair deficiency, detectable by IHC loss of MLH1 or MSH2, identifies tumors likely to respond to pembrolizumab regardless of histological type. Morphology tells you what kind of thing it is. IHC tells you which one.
A biopsy samples a tiny fraction of a heterogeneous mass through a needle a millimeter or two in diameter. Three distinct failures are possible. The needle misses the tumor entirely — in pancreatic adenocarcinoma, the tumor often constitutes a minority of the mass; the bulk is dense desmoplastic stroma. The needle can hit the mass but sample a necrotic core or a low-grade zone that does not carry the mutations driving lethal behavior. Or the needle enters the right region but retrieves too little tissue for the molecular profiling that modern treatment requires.
Still open: how heterogeneous a tumor can be before a single core is fundamentally inadequate — and when imaging certainty alone justifies treating without tissue.
The whole diagnostic chain is only as strong as its weakest link. And the weakest link is almost always the biopsy — not the imaging, not the pathology reading, but the question of whether the needle was in the right place. Refusing to accept negative biopsies drives repeat procedures, each with its own complication risk. The skill is calibrated interpretation: how much does this specific negative result, from this specific technique, against this specific pretest probability, move the posterior? That requires knowing the test sensitivity and the pretest probability. Neither is printed on the pathology report.
Cancer Research · Chapter 2 · Cancer Diagnosis — Imaging and the Tissue Sample
That is the frame for everything that follows. The scan raises a probability. The tissue settles it — but only when the tissue is the right tissue. Learn to read the chain: where the imaging can mislead, why the biopsy can fail silently, and how to apply the pretest probability to interpret what the needle brought back. The chain is only as strong as its weakest link. Almost always, the weakest link is the biopsy.