Image Quality in Radiology: Resolution, Contrast, Noise

What is image quality?

DoseSave is not only about dose; the other half of the work is image quality. Because dose and quality are two sides of the same coin — you cannot discuss one without the other.

So what is a "quality image"? In medicine, this is not an aesthetic judgment. The value of a radiological image is measured by how well it serves the diagnosis — far more a technical evaluation than an artistic one.¹

130 years ago, the same mindset

When a reporter asked Wilhelm Röntgen, who discovered X-rays, "And what did you think?", his answer was: "I did not think; I investigated."² Image quality is exactly that kind of subject — assessed by measuring, not by feeling.

We can split image quality into three core components: spatial resolution, contrast and noise. Understanding these is also the key to understanding the dose–quality balance.

Spatial resolution

Spatial resolution is how fine a detail we can resolve — can we see two tiny structures very close together as separate, or do they blur into a single smudge? Think of it like a camera's "megapixels" — but in medical imaging, resolution is set not only by pixel count but also by detector design, focal-spot size, motion, reconstruction and imaging geometry.¹

Every modality has a limiting resolution. The representative values below make this concrete:¹

Modality	Limiting spatial resolution
Mammography (film)	0.03 mm — highest in radiology
Digital mammography	0.05–0.10 mm
Radiography (film)	0.08 mm
Fluoroscopy	0.125 mm
Digital radiography	0.17 mm
Computed tomography	~0.3 mm
Nuclear medicine (planar)	~2.5 mm
SPECT	~7 mm

This is why mammography needs the highest resolution: it must catch tiny structures such as microcalcifications.

Contrast

If resolution is "how small," contrast is "how different." Contrast is how much a structure stands out, in gray level, from neighboring tissue. Contrast resolution is the ability to detect very subtle differences in gray scale — i.e. to distinguish a low-contrast lesion from its surroundings.¹

In radiography, contrast arises because tissues attenuate the X-ray beam differently: as the beam crosses the patient, bone absorbs much and soft tissue little; these differences in beam intensity reaching the detector form the contrast.¹ CT's soft-tissue contrast surpasses plain radiography because of its tomographic nature: out-of-slice structures do not clutter the image.¹

Noise

Noise is the unwanted "snow" or speckle in an image — like the graininess of an old TV. In medical imaging its main source is quantum noise (quantum mottle): the number of X-ray photons forming the image is limited, and as that number drops, speckle rises.¹

This is exactly where dose and quality intersect. X-ray imaging is a "quantum-limited" process: how many photons form the image determines the quality. Collecting more photons (more dose) reduces noise; fewer photons (less dose) increases it.¹

The square-root rule

Photon counts are statistical (Poisson). The signal-to-noise ratio rises with the square root of the number of photons collected. So to halve the noise you need about four times as many photons — and therefore roughly four times the dose. The cost of buying quality grows quickly.¹

SNR and CNR

Two metrics condense these ideas into a single number:

SNR (signal-to-noise ratio): the signal relative to the noise. High SNR means a clean, readable image.¹

CNR (contrast-to-noise ratio): a structure's contrast against its background, relative to the noise — the quantitative version of "can this lesion be picked out from the speckle?" CNR is especially useful for optimizing acquisition parameters.¹

In practice it is often CNR that decides the diagnosis: a finding, however small, cannot be seen if it is lost in the noise.

The dose–quality balance

Now it all comes together. Higher dose → more photons → less noise → higher SNR/CNR → better low-contrast detectability. So why don't we keep increasing the dose without limit?

Because once diagnostic quality reaches a threshold, extra dose adds nothing meaningful to the image; it only raises the patient's risk. The job of radiology physics is to find that balance: deliver quality sufficient for the diagnosis at the lowest possible dose.³

This is a compromise inherent to medical imaging: lower dose means more noise, higher dose means more risk; what we seek is the balance between patient safety and image quality.¹ That is precisely why DoseSave's two pillars — dose and quality — cannot be separated.

Image quality is task-dependent

A lung nodule, a kidney stone, a liver lesion, a microcalcification and a bone fracture do not share the same image-quality requirement. The goal of optimization is not to produce the "prettiest" image but to deliver sufficient image quality to answer the clinical question, at the lowest reasonable dose.

For why dose is tracked: Why Does Dose Matter?. For how parameters affect quality: Exposure Parameters. For noise and reconstruction: FBP, Iterative and Deep Learning. Key terms: MTF · NPS · DQE · Scatter · Grid.

References

Bushberg JT, Seibert JA, Leidholdt EM, Boone JM. The Essential Physics of Medical Imaging, 3rd ed. Lippincott Williams & Wilkins, 2011. Bölüm 1 (s.3, Tablo 1-1 s.5) ve Bölüm 4 (Image Quality; çözünürlük s.60, kontrast s.76, gürültü s.77–79, SNR/CNR s.91). Atıflardaki sayfa numaraları bu baskıya aittir.
Röntgen WC, H. J. W. Dam ile söyleşi: “The New Marvel in Photography.” McClure's Magazine, Cilt 6, Sayı 5, Nisan 1896, s.416. (Röntgen'in bilinen tek basın söyleşisi.)
ICRP Publication 103. The 2007 Recommendations of the ICRP. Ann. ICRP 37(2–4) — gerekçelendirme ve optimizasyon ilkeleri. icrp.org

Note: This content is for education; for clinical decisions or regulatory compliance, consult a qualified medical physicist and current regulations.

← All articles