Do digital radiographs detect caries better than film?

Digital radiographs and film detect caries with broadly similar accuracy [17]. A 2021 Cochrane review of early, enamel-only caries put pooled sensitivity at 0.44 (95% CI 0.38 to 0.50) for film and 0.49 (95% CI 0.42 to 0.56) for digital, and specificity at 0.90 (95% CI 0.86 to 0.93) and 0.87 (95% CI 0.82 to 0.91), and rated that evidence low certainty [1].

That runs against a natural assumption: that a newer receptor must see more. UK guidance now regards digital equipment as standard for contemporary practice [17], so the question isn't which one to buy. It's whether switching changes what you find on a bitewing, and what actually moves the answer when studies disagree.

What the question actually is

"Better" needs a yardstick. Diagnostic-accuracy studies radiograph teeth, have observers call each surface sound or carious, then check those calls against a reference standard. In the Cochrane review, that reference was histology of the sectioned tooth in 64 studies (83%) [1].

Two numbers come out of that comparison. Sensitivity is the share of carious surfaces the readers found. Specificity is the share of sound surfaces they correctly left alone. A receptor can gain on one and lose on the other, so "better" can point in two directions at once:

• In the Cochrane pooling, digital radiographs had the higher sensitivity and film the higher specificity [1].
• One study of 320 extracted teeth found the reverse for a CCD sensor against E-speed film: sensitivity consistently lower, specificity consistently higher, and overall accuracy not significantly different [4].

Some studies ask a different question again: not "is there a lesion?" but "does this surface need restoring?" With the treatment threshold set at macroscopic cavitation, one study of 84 surfaces found no significant difference in ROC area between film, a PSP system and a CCD system [6].

The receptors in play are film, photostimulable phosphor (PSP) plates, and solid-state sensors, the CCD and CMOS systems [13]. PSP plates and sensors are both digital receptors. Where a study below says "digital", it means one or both.

What the evidence shows

Pooled across studies

The Cochrane review pooled 55 film datasets and 42 digital datasets for non-cavitated, enamel-only caries [1]:

The review rated this evidence low certainty [1]. These figures describe enamel-only lesions, not caries into dentine, and come mostly from in vitro studies on extracted teeth [1].

The stronger comparison tests film and digital on the same teeth. The review found 32 such datasets from 15 studies, covering 2152 tooth sites or surfaces. There, film had sensitivity 0.41 (95% CI 0.30 to 0.53) and specificity 0.91 (95% CI 0.86 to 0.94); digital had sensitivity 0.47 (95% CI 0.36 to 0.59) and specificity 0.88 (95% CI 0.83 to 0.92) [1]. The difference in sensitivity reached statistical significance (P = 0.005); the difference in specificity didn't (P = 0.13) [1]. The confidence intervals overlap widely, so treat that sensitivity gap as an early signal rather than a settled advantage.

Head to head, study by study

Individual comparisons, all on extracted teeth, mostly land on "no significant difference". A few don't, and they don't agree with each other about which way:

UK guidance reaches the same overall reading: the diagnostic yield of digital radiography compared with film "has been shown to be broadly equivalent" [17].

Does image enhancement change caries detection?

Digital receptors let you adjust the image after exposure. UK guidance lists contrast enhancement among digital's benefits [17]. The evidence that it improves caries detection is thin and mixed, and no systematic review of it turned up. Every study below is laboratory work with a small reader panel:

• Observer-applied enhancement lowered accuracy. With a CCD system, Az was 0.7650 unenhanced, 0.7499 enhanced, and 0.8008 for Ektaspeed Plus film. The enhanced images were significantly less accurate than film [9].
• A contrast-enhancement routine didn't change ROC area for restorative decisions, although it gave the highest positive likelihood ratio: 11.37, against 9.67 for the same CCD system unenhanced, 8.14 for PSP and 5.29 for film [6].
• Brightness and contrast came out best of the post-processing settings in a 2024 PSP study with two radiologists: accuracy 82.5% and 83.8 for brightness and 82.5% for contrast, against 78.8% and 77.5% for noise adjustment [10].
• Brightness and contrast made no overall difference in a 2020 study of two digital systems, though it improved accuracy for one system when a metal-ceramic crown was in the field [11].
• Grey-scale reversal didn't help. Inverting the image did not optimise proximal caries detection and made dentine lesions harder to see [12].
• Task-specific filters traded specificity for sensitivity. Filtered images from one sensor had the highest sensitivity of seven solid-state receptors, but that came mainly from one observer, specificity was the lowest, and overall accuracy didn't differ between receptors [13].

Read that as a caution rather than a boost: in these laboratory studies, image enhancement did not reliably improve caries detection, and some kinds made it worse.

Where it's unsettled

The spread between studies is wide, and studies disagree for reasons beyond the receptor.

Most of the evidence comes from extracted teeth. In the Cochrane review, 60 of 77 studies (78%) were in vitro. Sensitivities ranged from 0 to 0.96 and specificities from 0 to 1.00, and none of the review's pre-specified explanations accounted for that spread [1]. The review downgraded its certainty for risk of bias, for indirectness arising from the in vitro studies, and for inconsistency [1].

Lesion depth changes the answer. In one comparison, lesion depth significantly affected how well observers performed, and depth tended to be underestimated [3]. Across a broader 2015 review, radiographic detection was highly accurate for cavitated proximal lesions and seemed also suitable for dentine lesions, but had low sensitivity for initial lesions [2]. A study weighted towards shallow lesions will report a different receptor comparison from one weighted towards deeper lesions. The early enamel lesion is where every receptor struggles.

Readers vary more than receptors, in at least one direct test. Syriopoulos and colleagues compared two films and four digital systems with eight observers. Radiologists performed significantly better than general practitioners whatever the receptor, and the authors concluded that the dentist's ability to recognise caries, not the imaging modality, was the main source of variation [3]. Reader variation shows up inside individual studies too. A filter's sensitivity gain came mainly from one observer [13]. Hooding a laptop in a bright clinic helped faculty readers, while students' performance stayed highly variable [15]. Some readers were more accurate on a second viewing of unfamiliar digital images [7].

The studies that do find a difference point in opposite directions. Film beat a sensor in the mixed dentition [7]; a PSP system beat film and a sensor in primary molars, significantly only at 50 kV [8]; pooled data favour digital on sensitivity and film on specificity [1]. Most head-to-head studies here date from 1997 to 2010.

What this means at the chairside

The limit is the radiograph's, whichever receptor made it. Low-certainty evidence suggests that, for early caries, imaging may have poor sensitivity but acceptable specificity [1]. The Cochrane review's worked example: in 1000 tooth surfaces with an enamel caries prevalence of 63%, imaging would call 337 carious surfaces sound and 43 sound surfaces carious [1]. A clean proximal surface on a bitewing doesn't rule out enamel caries, on film or on a sensor.

The radiograph follows the clinical examination. The 2026 ADA and AAOMR recommendations put patient history and clinical findings before any radiograph, and ask you to review previously obtained images [18]. For children, UK guidance calls posterior bitewings "an essential adjunct to clinical examination" [17].

How you view the image matters too. UK guidance asks for:

• suitable monitors, viewed under optimum conditions and after adjusting brightness and contrast
• regular quality assurance of plates, sensors and plate readers
• for film, an appropriate light box and magnifier, with extraneous light masked out [17]

The evidence on room light is reassuring but conditional. With a monitor calibrated for its ambient light, light level made no difference to caries detection [14]. On bitewings from patients, monitor type and room light didn't change accuracy, though room light did change how confident readers felt [16]. In a bright clinic, hooding a laptop raised experienced readers' Az from 0.79 to 0.87 [15].

Enhancement has thin evidence. UK guidance asks for digital radiographs to be viewed after adjustment of brightness and contrast [17]. In single laboratory studies, grey-scale reversal worsened detection of dentine lesions and task-specific filters gave the lowest specificity [12, 13].

The evidence gives you no strong reason to read a digital radiograph with more confidence than film, or less.

Film or sensor, a radiograph only shows so much. Practise reading what it does show on real bitewings: start here.