Introduction — the indispensable micron-level scan

The CT temporal bone protocol occupies a unique and exacting niche within the computed tomography landscape. No other routine CT examination demands the same combination of sub-millimetre spatial resolution, precise patient positioning, and specialised bone-kernel reconstruction as the temporal bone study — a scan whose entire diagnostic value can be undermined in seconds by incorrect isocenter placement or an inappropriate reconstruction algorithm. For radiographers, radiologists, and the ENT surgeons and neurologists who act on its findings, mastery of the CT temporal bone protocol is not optional; it is a clinical imperative.

The temporal bone encases the most anatomically complex region in the human skull. Within its petrous apex and squamous portions lie the malleus, incus, and stapes — ossicles weighing less than 30 mg that transmit acoustic energy from the tympanic membrane to the oval window — as well as the bony labyrinth, the facial nerve’s *canalis facialis*, the *cochlea*, and the semicircular canals. At the heart of this protocol is the internal auditory canal (IAC), a short bony corridor approximately 8–10 mm in length and 4–5 mm in diameter that houses the vestibulocochlear nerve (CN VIII) and facial nerve (CN VII) as they enter the posterior cranial fossa. Detecting cholesteatoma extension, diagnosing otosclerosis, classifying temporal bone fractures, and identifying superior semicircular canal dehiscence (SSCD) all hinge on the clarity of these millimetre-scale bony structures.

Clinical demand for temporal bone CT has grown considerably over the past decade. Population aging, rising rates of chronic otitis media, expanding cochlear implant candidacy criteria, and increasing ENT surgical volumes have all contributed to higher imaging referral rates. Simultaneously, CT scanner technology has evolved — from 16-slice scanners requiring dedicated high-resolution temporal bone protocols to modern 320-slice, dual-energy, and photon-counting CT (PCCT) systems capable of sub-0.25 mm isotropic resolution with dramatically reduced radiation burden.^[2] Understanding the entire spectrum of scanner capability, from departmental workhorses to frontier technology, equips the modern radiographer and radiologist to deliver consistently excellent diagnostic output regardless of institutional resources.

This article provides the complete evidence-based reference framework for the CT temporal bone and IAC protocol. It covers gross anatomy, Hounsfield Unit (HU) reference ranges for all key structures, a seven-step scanning technique validated across scanner generations, dual-energy and photon-counting protocol variants, contrast media rationale, radiation dose benchmarks aligned to international guidelines, the ten most important detectable pathologies, and an exhaustive three-tier pitfall matrix for radiographers, radiologists, and non-radiology clinicians.

Anatomy and HU values of the temporal bone

Gross anatomical narrative

The temporal bone is a paired, compound cranial bone formed from four embryological components: the squamous, tympanic, mastoid, and petrous parts. The petrous part — so named for its stone-like density — is the most clinically critical region for CT imaging. It projects anteromedially toward the clivus and forms the posterior wall of the middle cranial fossa. The petrous apex, its most medial tip, borders the cavernous sinus and is a frequent site of apicitis, cholesterol granuloma, and rare cholesteatoma extension.

The middle ear cleft (tympanic cavity) is an air-filled space approximately 15 mm tall, 15 mm deep, and only 6 mm wide at its centre. It contains the three ossicles and their ligamentous suspensions, two intratympanic muscles (tensor tympani and stapedius), the chorda tympani branch of CN VII, and the tympanic plexus. The Eustachian tube connects the middle ear to the nasopharynx at its anteromedial aspect. Superiorly, the *tegmen tympani* — a thin plate of bone separating the middle ear from the middle cranial fossa — is frequently the site of meningocele, encephalocele, or CSF otorrhoea when dehiscent.^[3]

The mastoid process projects inferiorly from the posterior temporal bone and contains the mastoid antrum — the largest air cell, communicating directly with the middle ear via the aditus ad antrum. In a well-pneumatised mastoid, hundreds of interconnected air cells fill the process. Mastoiditis and cholesteatoma exploit these air-cell tracts as pathways for spread. The inner ear (labyrinth) occupies the otic capsule, the densest bone in the human body. The cochlea completes approximately 2¾ turns around the central bony modiolus. The vestibule connects to three semicircular canals: superior (anterior), posterior, and lateral (horizontal). The superior semicircular canal’s thin bony roof overlying the middle fossa dura is the focus of the increasingly recognised SSCD syndrome.

The internal auditory canal (IAC) is a fundus-to-porus corridor approximately 8–12 mm in length. At its fundus, the falciform crest (transverse crest) divides the nerve bundle: the facial nerve occupies the anterosuperior quadrant, the superior vestibular nerve the posterosuperior quadrant, the cochlear nerve the anteroinferior quadrant, and the inferior vestibular nerve the posteroinferior quadrant. CT provides exceptional detail of the IAC bony walls and can detect widening (suggesting a large vestibular schwannoma) or narrowing (congenital stenosis), though MRI remains superior for soft tissue nerve evaluation.^[4]

Hounsfield Unit reference table

The otic capsule — why it matters diagnostically

The otic capsule’s extraordinary density — the highest of any bone structure routinely encountered in CT imaging — is both an asset and a challenge. It provides outstanding inherent contrast between the bony labyrinth and the air- or fluid-filled spaces within it, enabling detection of subtle fenestral otosclerosis lesions as small as 0.4 mm. However, this same density creates beam hardening artefacts when the X-ray beam traverses the petrosal pyramids bilaterally, generating dark pseudo-hypodense bands that can simulate labyrinthine pathology or obscure small IAC lesions.^[5] Recognition and mitigation of beam hardening (discussed in the pitfalls section) is an essential competency for both the acquiring radiographer and the interpreting radiologist.

Ossicular chain anatomy and surgical relevance

The three ossicles form the smallest articulating joint system in the human body. The malleus attaches via its *manubrium* to the tympanic membrane and articulates with the incus at the *incudomalleolar joint*, a saddle-type synovial joint. The incus’s long process descends to the *lenticular process*, which articulates with the stapes head at the *incudostapedial joint*. The stapes sits in the oval window, transmitting vibration to the perilymph via its footplate.

CT assessment of the ossicular chain is the definitive pre-operative evaluation for suspected ossicular discontinuity (post-traumatic or cholesteatoma-related), stapedial fixation (otosclerosis), and congenital ossicular dysplasia. The incudostapedial joint is the most radiographically delicate and the most commonly disrupted in temporal bone trauma. On axial CT, normal ossicular chain morphology is sometimes described as an ice cream cone sign (malleus head and incus body in the epitympanum), while the incus long process and stapes appear in coronal and parasagittal reformats.^[6]

The facial nerve canal

The facial nerve (CN VII) traverses the temporal bone through four distinct segments: the internal auditory canal segment, the labyrinthine segment (shortest, 3–5 mm), the tympanic (horizontal) segment along the medial wall of the middle ear, and the mastoid (vertical) segment descending to the stylomastoid foramen. CT reliably depicts the bony facial canal and identifies dehiscences — congenitally thin or absent bone — most common in the tympanic segment. Unrecognised facial canal dehiscence carries significant surgical risk during stapedectomy or mastoidectomy, making its CT identification clinically imperative.^[7]

Scanning technique — 7 steps and scanner comparison

Seven numbered steps for CT temporal bone acquisition

1. Patient positioning and isocenter alignment. Position the patient supine with the head in the neutral position, chin slightly tucked to bring the orbitomeatal line perpendicular to the gantry bore. This is the single most critical step of the entire examination. Use laser alignment guides and confirm that the skull is at the true gantry isocenter in both the cranio-caudal and lateral axes. Even 1–2 cm of off-isocenter positioning measurably degrades spatial resolution at the periphery of the reconstruction field, directly compromising ossicular chain assessment. Use a dedicated head holder or cradle to eliminate rotational drift.^[8]

2. Scout acquisition and anatomy verification. Acquire a biplane scout (lateral and PA) to confirm positioning symmetry. Verify that both petrous pyramids project symmetrically and that there is no head tilt. Adjust before committing to helical acquisition; the consequence of sub-optimal positioning in temporal bone CT is amplified compared to brain CT because the target structures are so small. For bilateral examinations, confirm full coverage from the external auditory canals to the posterior cranial fossa including the entire IAC.

3. Protocol parameters. Apply 120 kVp with a tube current range of 250–300 mA (or automatic tube current modulation targeting a reference mAs of 275). Set the pitch to 0.6 — this slow pitch is essential for maximising z-axis sampling density given the sub-millimetre slice targets. Set the rotation time to 1.0 second, which balances adequate tube output per rotation with manageable heat unit loading. The field of view (FOV) must be tightly targeted to the temporal bones bilaterally — typically 120–160 mm — to maximise the displayed pixel density from the available matrix.

4. Reconstruction kernel and slice thickness. Reconstruct exclusively using a bone-plus or ultra-sharp kernel (e.g., Siemens B80f, GE Bone Plus, Philips YD or YE kernel equivalent). This is non-negotiable; a soft tissue kernel will obscure the fine cortical margins of the ossicles, facial canal, and otic capsule. Generate a primary series at ≤0.5 mm slice thickness with ≤0.5 mm reconstruction increment (i.e., overlapping thin slices) for multiplanar reformatting. A secondary series at 0.8–1.0 mm may be generated for overview axial images but should not replace the sub-mm primary dataset.^[9]

5. Multiplanar reformatting (MPR). Generate three standard reformatted series: axial (parallel to the lateral semicircular canal plane), coronal (perpendicular to the petrous pyramid), and — crucially — Pöschl (along the long axis of the superior semicircular canal) and Stenvers projections (perpendicular to the long axis of the petrous pyramid). The Pöschl projection is the definitive view for diagnosing or excluding superior semicircular canal dehiscence (SSCD), which can appear falsely open on standard axial images due to partial volume averaging of the thin overlying bone.

6. Window settings. Apply bone window settings with window width (WW) of approximately 4,000 HU and window level (WL) of approximately 700 HU. This extremely wide window is mandatory to display the full dynamic range from dense otic capsule bone to air-filled tympanic cavity without clamping either extreme. Some institutions supplement this with a narrow soft-tissue window to assess middle ear soft tissue opacification character, though this is secondary. Customise brightness and contrast on a case-by-case basis for suspected otosclerosis (narrow the window to accentuate subtle otic capsule density changes).

7. Deep learning reconstruction (DLR) and dose optimisation. Where available, apply vendor-specific deep learning reconstruction (DLR) algorithms (e.g., Canon AiCE, Siemens Deep Resolve, GE TrueFidelity) at medium or strong settings. Clinical validation studies have confirmed that DLR at 50–80% reduced mAs can maintain or improve signal-to-noise ratio and modulation transfer function (MTF) at high spatial frequencies — directly relevant to ossicular chain and SSCD assessment.^[10] Always verify vendor validation data for the specific kernel and anatomical region before reducing dose.

Scanner comparison table: 16-slice to 320-slice

Dual-energy and photon-counting CT protocol variants

Contrast media protocol — rationale for non-contrast acquisition

The CT temporal bone protocol is performed as a non-contrast CT (NCCT) as the universal standard acquisition. This is not a default omission but a deliberate, evidence-based protocol decision grounded in the fundamental physics of what CT temporal bone is required to diagnose.

Why contrast is not indicated

The diagnostic target in CT temporal bone imaging is bony microarchitecture, not vascular or soft tissue enhancement patterns. The differential diagnoses most commonly evaluated — cholesteatoma versus granulation tissue, fracture line characteristics, ossicular erosion, SSCD, otosclerosis — are all fundamentally defined by their relationship to adjacent bony structures and do not require intravenous contrast to characterise. The high inherent density difference between otic capsule bone (1,600–3,000 HU), cortical petrous bone (400–900 HU), air-filled mastoid cells (−1,000 HU), and soft tissue opacification (20–60 HU) provides more than adequate natural contrast for diagnosis without the use of iodinated agents.

Safety check — the NCCT pathway

In routine temporal bone CT, the absence of contrast administration simplifies the pre-scan safety pathway. There is no requirement for renal function screening (eGFR), no metformin management protocol, no allergy pre-medication pathway, and no post-procedure hydration monitoring. This significantly streamlines patient throughput in ENT CT lists, which frequently include elderly patients with age-related renal impairment or multiple medications.^[11]

For sites employing dual-head power injectors as part of their standard CT workflow, the non-contrast nature of temporal bone CT means these devices remain unused for this specific protocol — though the injector line and patient IV access may still be placed in mixed-scan lists in anticipation of subsequent contrast-requiring studies. SATMED Health’s SATLine patient line systems and SATSyringe precision syringes support seamless crossover between non-contrast and contrast-requiring protocols within the same workflow session.

Radiation dose and optimisation

Despite targeting a small anatomical volume, the CT temporal bone protocol demands relatively high tube output to achieve the sub-millimetre spatial resolution required. This places the radiation dose per examination at the upper range of head CT studies, and it justifies a rigorous evidence-based approach to dose optimisation — particularly in paediatric patients and those requiring multiple serial follow-up studies (e.g., post-cholesteatoma surgery surveillance).

Dose reference levels table

Five dose reduction strategies

1. Deep learning reconstruction (DLR). Applying DLR at 40–60% mAs reduction has been demonstrated in peer-reviewed studies to preserve or improve high-spatial-frequency resolution in temporal bone CT without introducing unacceptable noise amplification.^[10] This is the highest-impact single dose reduction measure currently available on compatible scanner platforms and should be the first optimisation strategy implemented by any institution with DLR capability.

2. Tight field-of-view targeting. Restricting the reconstruction FOV to the temporal bones bilaterally (120–160 mm FOV versus a routine 200–250 mm head FOV) maximises the pixel density available from the fixed detector matrix. This improved displayed spatial resolution reduces the clinical temptation to over-irradiate to compensate for suboptimal image quality — making tight FOV a dose-reducing as well as quality-improving measure.

3. Paediatric dose scaling. Children undergoing temporal bone CT for congenital hearing loss, chronic otitis media, or suspected cholesteatoma require age- and weight-based dose scaling strictly aligned to ALARA principles. The paediatric temporal bone CT effective dose should be targeted to ≤0.3 mSv using age-appropriate reference charts published by the AAPM and endorsed by the ICRP.^[13] DLR or iterative reconstruction at high strength settings should be mandatorily applied in paediatric protocols.

4. Reduced kVp for small patients. Reducing the tube voltage from 120 kVp to 100 kVp in patients with low body weight (typically those with head circumference ≤55 cm or body weight ≤70 kg) increases inherent image contrast through photoelectric effect enhancement and reduces radiation dose by approximately 30–40% relative to 120 kVp for equivalent image quality. This is particularly applicable in paediatric and small adult temporal bone examinations.^[14]

5. Omission of unnecessary series. In straightforward clinical scenarios (e.g., unilateral conductive hearing loss for pre-operative planning), targeted unilateral temporal bone CT may be clinically acceptable rather than mandatory bilateral acquisition. Dose-audit data suggest that bilateral temporal bone CT contributes negligibly to cancer risk in the individual patient but represents a meaningful population-level dose reduction when applied protocol-wide across high-volume ENT CT lists. Always consult the referring clinician regarding unilateral versus bilateral acquisition requirements before defaulting to full bilateral coverage.

Top 10 pathologies detectable on CT temporal bone

The following pathology card grid covers the ten clinically most significant conditions diagnosed or evaluated on CT temporal bone and IAC imaging, with Hounsfield Unit ranges, key CT signs, and direct protocol impact on detection.

Pitfalls for radiographers — scanning

Pitfalls for radiologists — interpretation

Pitfalls for non-radiology physicians — clinical

Pitfall comparison summary

The three-column summary below synthesises the most critical pitfalls across the multidisciplinary temporal bone CT workflow, providing a rapid-reference framework for departmental governance, multi-disciplinary team meetings, and individual CPD.

AI and automation in temporal bone CT

Artificial intelligence applications in temporal bone CT represent one of the most technically demanding frontiers in medical image analysis. The sub-millimetre structures involved — ossicles weighing less than 30 mg, canal walls thinner than 0.3 mm — require AI models trained on extremely high-quality, radiologist-annotated datasets and validated specifically on the bone-kernel reconstructions used in clinical practice. Despite these challenges, the field has advanced significantly and several well-evidenced applications are now approaching clinical readiness.^[19]

Automated ossicular chain segmentation

Deep learning segmentation of the malleus, incus, and stapes has been achieved with clinically acceptable accuracy in several academic centres using multi-atlas and convolutional neural network (CNN) approaches. These tools support cochlear implant electrode planning by generating precise 3D models of the cochlea and ossicular chain from HRCT datasets, replacing laborious manual segmentation that previously required 30–90 minutes of specialist time per case.^[20] FDA-cleared cochlear implant surgical planning software incorporating automated inner ear segmentation (e.g., Otoplan/MED-EL, Simplant ENT) is available in multiple clinical centres and represents a validated AI application with direct surgical workflow benefit.

AI-assisted cholesteatoma and SSCD detection

Several academic groups have published CNN-based detection algorithms for cholesteatoma on non-contrast temporal bone CT, with sensitivity and specificity figures exceeding 85% in validation cohorts.^[21] These tools aim to flag erosive middle ear findings for radiologist attention in high-volume reporting environments. Similarly, automated detection of SSCD using multi-planar reconstruction AI analysis on Pöschl-equivalent reformats has been reported, with the specific goal of avoiding false-positive diagnosis from partial volume averaging — the primary interpretation pitfall of this condition. None of these specific tools are currently FDA-cleared as standalone diagnostic software, but several are under regulatory evaluation.

Automated fracture classification

Post-traumatic temporal bone CT presents a volume challenge in major trauma centres where pan-CT datasets are acquired on every major trauma patient and temporal bone assessment may be deprioritised in favour of life-threatening intracranial, thoracic, and abdominal findings. AI-assisted flagging of otic-capsule-violating temporal bone fractures — the most prognostically important subtype — from whole-head CT datasets represents a clinical triage application with genuine patient safety implications. Pilot studies have demonstrated automated identification of otic capsule disruption with AUC values exceeding 0.91 on external validation datasets.^[22]

Photon-counting CT and AI integration

Photon-counting CT systems offer sub-0.2 mm isotropic resolution that has been described as generating CT images approaching the clarity of anatomical specimens. Emerging research explores combining PCCT’s exceptional spatial resolution with AI segmentation models trained specifically on PCCT datasets — potentially enabling automated stapes footplate thickness measurement, automatic assessment of oval window narrowing in otosclerosis, and quantitative cochlear duct length measurement for cochlear implant electrode size selection. These applications remain research-stage in 2026 but represent the next-generation convergence of hardware and software innovation in temporal bone imaging.^[23]

Conclusion

The CT temporal bone protocol is a technically precise, anatomically demanding examination that sits at the diagnostic crossroads of otolaryngology, neurotology, and neurosurgery. Its diagnostic value is entirely contingent on the consistent application of exacting technique: positioning at the gantry isocenter, bone-kernel reconstruction at ≤0.5 mm slice thickness, targeted field-of-view selection, and systematic multiplanar reformatting in anatomically guided planes.

From the Hounsfield Unit reference framework — spanning from the densest otic capsule bone at 1,600–3,000 HU to air-filled mastoid cells at −1,000 HU — to the ten clinically critical pathologies from cholesteatoma and otosclerosis to superior semicircular canal dehiscence and labyrinthitis ossificans, the effective radiographer and radiologist must command an encyclopaedic knowledge of temporal bone anatomy and its pathological variants. The primary scanning pitfall — off-isocenter patient placement causing spatial resolution degradation — and the primary interpretation pitfall — false-positive SSCD diagnosis due to partial volume averaging — represent the two most impactful quality failure points in the entire temporal bone CT workflow and must be actively mitigated through protocol governance, reconstruction standards, and systematic reporting discipline.

For non-radiology physicians, the fundamental clinical lesson from this protocol framework is that the CT temporal bone examination is an exquisitely specialised study that cannot be substituted by a routine brain CT. Ordering the correct protocol — and referring to a gadolinium-enhanced MRI IAC when vestibular schwannoma or intralabyrinthine pathology is suspected — is the first and most consequential clinical decision in the temporal bone diagnostic pathway.

As photon-counting CT continues to enter clinical practice, and as deep learning reconstruction algorithms become standard on major scanner platforms, the sub-millimetre resolution benchmarks that define CT temporal bone excellence will become more achievable at lower radiation doses — improving access to high-quality temporal bone imaging across a wider range of institutional settings. The protocol principles described in this article remain the foundational framework on which these technological advances build.