Mach band effect
The Mach band effect is an optical illusion in which the eye exaggerates contrast where two densities meet, creating an apparent dark band that is not present in the image. On a bitewing radiograph it can appear in the dentine along the proximal dentinoenamel junction, where it mimics a carious lesion.
Synonyms: Mach bands; Mach effect; border contrast illusion
What is the Mach band effect?
It's the one caries mimic that isn't on the radiograph.
Everything else you learn to rule out — burnout, superimposition, a restoration edge — is a real pattern of density that a densitometer could measure. A Mach band is not. It's produced in your visual system as you look, and the radiograph itself is uniform across the spot where you see the band.
The Austrian physicist Ernst Mach described the effect in 1865. Put a dark region next to a light region and the retina enhances the edge between them: the dark side looks darker right at the border, and the light side looks lighter. That's all a Mach band is — a strip of exaggerated contrast, invented at the boundary.
You can prove it's an illusion. Photodensitometric tracing across the affected area reads a consistent density with no alteration where the band appears to be. The eye reports a step that the emulsion never recorded.
What it looks like
Berry's 1983 description is still the one to learn. On dental radiographs, Mach bands may appear in the dentine along the proximal dentinoenamel junction, or in the dentinal peaks bounded by occlusal and proximal enamel. A 2025 systematic review restates it for the modality you'll actually be holding: at the enamel–dentine junction on a bitewing, the effect can appear as an area of apparent radiolucency suggestive of caries.
So the classic false alarm is a thin dark line hugging the inside of the enamel cap, right where an early dentine lesion would sit.
Bands come in two directions. A negative Mach band is a narrow darker strip at the border between a radiolucent and a radiopaque area — darker than the surrounding tissue. A positive Mach band is a narrow lighter strip along the margin where a denser area meets a less dense one. The negative one is the diagnostic problem, because dark reads as caries.
The DEJ isn't the only interface that produces them. A 2022 survey of 77 dental publications that mention the effect found the two most frequently named locations were the interface between a very radiopaque restorative material and tooth structure, and the enamel–dentinal junction. The interface between a dental implant and the surrounding bone came up often as well. Anywhere a hard density step meets softer tissue on the image, the eye can add a line to it.
What it's confused with
| Looks like | What tells them apart |
|---|---|
| Proximal caries at the DEJ | It survives the masking test. Cover the enamel side of the interface and a true lesion is still there; a Mach band goes away |
| Cervical burnout | It's in the image, not in the eye. Overpenetration of the thinner cervical tooth structure produces radiolucency with rounded, diffuse inner borders and an intact tooth edge, sited between the CEJ and the alveolar crest rather than at the DEJ |
| Triangular-shaped radiolucency (TSR) | A superposition effect of maxillary molar crown anatomy, including the Carabelli cusp. It sits on mesial surfaces of upper molars — in the one frequency study retrieved here, on no lower molar at all |
Why it happens
The standard account in the radiology literature is lateral inhibition among the neural receptors of the retina. Neighbouring receptors suppress each other, so a receptor sitting just inside the dark side of a boundary is inhibited harder than one further in, and the border ends up over-represented as a band.
Vision science hasn't fully closed the question. A 2014 modelling paper explicitly criticises the validity of lateral inhibition as the principle behind the illusion and proposes a diffusion-based reconstruction model instead. That argument is about the mechanism, not about whether the illusion is real or whether it turns up on radiographs — both of which are settled.
What matters more chairside is that the effect isn't uniform across observers. Not everyone perceives Mach bands equally, and perception is modified by projection, contour, and film and object density. Two dentists at the same light box can genuinely see different things in the same radiograph, and neither is being careless.
What it means clinically
The consequence is over-call. A 2025 systematic review and meta-analysis in Oral Radiology found that radiographic optical effects — triangular-shaped radiolucencies or Mach bands, pooled together — were interpreted as caries or fracture in approximately 13% of observations (60 of 464). Read that number carefully: it covers two distinct phenomena together, not the Mach band effect alone, heterogeneity was high (I² > 90%), and the review graded the certainty of the evidence low to very low. It tells you the direction of the problem, not its size.
There is no published prevalence figure for the Mach band effect on its own on bitewings. If you see one quoted, check whether it's actually a TSR figure — the review's other headline number, a 26.44% prevalence on maxillary molars (270 of 1021), is TSR, not Mach band.
The effect may also cut the other way. The same review notes experimental work showing Mach bands can obscure the detection of small features nearby, which would make them a source of false negatives as well as false positives — a real lesion sitting close to a strong density boundary getting masked. That's an early line of evidence, not an established clinical finding.
The review's own recommendation is the conservative one: interpret radiographic findings with caution and always correlate them with a thorough clinical examination.
How to be sure
Mask the interface. This is Berry's technique and it's still the direct test. If a radiolucency sits in the dentine underneath the enamel–dentinal junction, cover the enamel with an opaque card or a piece of paper. That negates the lateral inhibitory effect and the Mach band disappears. Take the card away and it comes back. If the radiolucency is there whether the enamel is covered or not, it isn't a Mach band — on a bitewing that points to true demineralisation or a lesion in the outer dentine.
The critical detail is where you put the card: mask the interface between the radiopaque and the less radiopaque structure. Covering the wrong edge tests nothing.
Change the contrast. Manual contrast adjustment can help mask the effect. Digital post-processing has also been described — software that analyses grey levels inside a radiolucent area, treating varying grey levels as suggestive of caries and uniform levels as more likely the Mach band effect. That's a single reported approach rather than a validated standard.
Correlate clinically. The radiograph supports the clinical examination; it doesn't replace it.
One honest caveat on all of this. The masking test is a described technique, not a measured one. The 2022 literature survey found no study that adopted a masking technique while observers interpreted the images, despite earlier work suggesting it. Nobody has published how much it actually improves accuracy.
Common errors
Letting the referral write the report. In a study of 33 fourth-year dental students and 40 dentists shown the same periapical radiograph of a maxillary central incisor, students were more likely than dentists to misread a Mach band at the junction of alveolar crestal bone and root as a horizontal root fracture. When the same radiograph came with a history of recent trauma, both groups were more likely to call it a fracture. Expectation moves what you see, and experience only partly protects you.
Knowing about it and not testing for it. Awareness isn't the same as a method. Of 77 dental publications mentioning the Mach band effect, eight presented radiographic images illustrating it and three actually investigated it — and none used the masking test at the point of interpretation.
Quoting the pooled numbers as Mach band figures. Both of the widely cited percentages come from a review that combines Mach bands with triangular-shaped radiolucencies. Neither is a Mach band statistic.
Assuming it's a dental quirk. It isn't. A negative Mach band across the base of the dens has been mistaken for a cervical spine fracture, and a Mach band on a chest radiograph has mimicked pneumopericardium. The eye does this everywhere; dentistry just happens to put a bright enamel cap next to darker dentine on every posterior proximal surface you look at.
References
- Berry HM. Cervical burnout and Mach band: two shadows of doubt in radiologic interpretation of carious lesions. J Am Dent Assoc. 1983 May;106(5):622-5. PMID 6575081. DOI
- Dioguardi M, Guerra C, Sovereto D, et al. Radiographic artifacts in the diagnosis of dental caries: systematic review with meta-analysis. Oral Radiol. 2025;42(2):281-296. PMID 41331196. DOI
- Yeung AWK. The diagnostic relevance and interfaces covered by Mach band effect in dentistry: an analysis of the literature. Healthcare (Basel). 2022 Mar 28;10(4). PMID 35455810. DOI
- Lane EJ, Proto AV, Phillips TW. Mach bands and density perception. Radiology. 1976 Oct;121(1):9-17. PMID 959562. DOI
- Chasen MH. Practical applications of Mach band theory in thoracic analysis. Radiology. 2001 Jun;219(3):596-610. PMID 11376243. DOI
- Nielsen CJ. Effect of scenario and experience on interpretation of mach bands. J Endod. 2001 Nov;27(11):687-91. PMID 11716082. DOI
- Daffner RH. Pseudofracture of the dens: Mach bands. AJR Am J Roentgenol. 1977 Apr;128(4):607-12. PMID 403789. DOI
- Panikkath R, Panikkath D. Mach band sign: an optical illusion. Proc (Bayl Univ Med Cent). 2014 Oct;27(4):364-5. PMID 25484514. DOI
- Kühnisch J, Pasler FA, Bücher K, Hickel R, Heinrich-Weltzien R. Frequency of non-carious triangular-shaped radiolucencies on bitewing radiographs. Dentomaxillofac Radiol. 2008 Jan;37(1):23-7. PMID 18195251. DOI
- Geier J, Hudák M. Modelling the Mach bands illusion by means of a diffusion model. Perception. 2014;43(9):896-913. PMID 25420330. DOI
Practice this
Now you know your eye invents this one. Try three real bitewings and see whether you can tell.