Introduction

Low dose CT paranasal sinuses has become the unequivocal imaging benchmark for evaluating chronic rhinosinusitis, sinonasal polyposis, and presurgical anatomy before functional endoscopic sinus surgery (FESS). Unlike MRI, which excels in soft-tissue discrimination, CT uniquely maps the delicate bony architecture of the ostiomeatal unit (OMU), the ethmoid labyrinth, and the critical surgical danger zones adjacent to the orbit, skull base, and carotid canal — all structures that the operating ENT surgeon must navigate with millimetre precision.^[1]

The move toward low dose acquisition — using 100 kVp and a tight mA range of 80–120 mA — reflects a fundamental truth about sinonasal anatomy: the air-to-bone interface provides inherently high intrinsic contrast that renders elevated radiation exposure not only unnecessary but actively counterproductive in the context of ALARA (as low as reasonably achievable) compliance. Contemporary iterative reconstruction algorithms have made it feasible to reduce effective dose to as little as 0.1–0.3 mSv — roughly equivalent to 10–30 days of background radiation — while preserving diagnostic-quality bone-window resolution.^[4]

This article is structured as a complete clinical and technical reference for radiographers responsible for acquisition, radiologists responsible for reporting, and non-radiology physicians responsible for clinical integration of findings. All three professional groups encounter characteristic and well-documented error patterns; the pitfall matrices in sections 10–13 address each group with specificity, grounded in published evidence.

Beyond inflammation and infection, the paranasal sinuses harbour a clinically diverse range of neoplastic, traumatic, and vascular conditions — many of which share overlapping CT appearances and demand systematic, methodical review to avoid diagnostic error. This article covers all ten primary pathological categories, with HU values, protocol impact, and reporting guidance for each.

Anatomy and HU values

Gross anatomy of the paranasal sinuses

The paranasal sinuses are four paired, air-filled cavities embedded within the facial skeleton. They are grouped into anterior sinuses — the maxillary, frontal, and anterior ethmoid cells — and posterior sinuses, comprising the posterior ethmoid cells and the sphenoid sinuses. The drainage anatomy differs between these groups: the anterior sinuses share a common drainage corridor, the ostiomeatal unit (OMU), which empties into the middle meatus of the lateral nasal wall. The posterior sinuses drain separately into the sphenoethmoid recess.^[5]

The OMU is the single most clinically critical anatomical region visible on CT paranasal sinuses. It is bounded medially by the middle turbinate, laterally by the medial orbital wall (lamina papyracea), and posteriorly by the basal lamella of the middle turbinate. The uncinate process, a thin, curved blade of bone, forms the medial boundary of the infundibulum, through which maxillary sinus secretions must pass to reach the middle meatus. Any OMU compromise — whether from mucosal oedema, anatomical variants, or polyps — propagates obstructive disease to the entire anterior sinus group.

The ethmoid labyrinth and skull base relations

The ethmoid labyrinth is a honeycomb of 3 to 18 air cells per side, separated by paper-thin bony septa (the basal lamellae). The roof of the ethmoid, the fovea ethmoidalis, forms the lateral extension of the cribriform plate and is one of the most variable — and most dangerous — structures in sinonasal surgery. Its depth below the level of the cribriform plate is classified by the Keros classification into three types (Type I: 1–3 mm; Type II: 4–7 mm; Type III: 8–16 mm), with Keros Type III representing the highest risk for inadvertent intracranial entry during FESS.^[6] Radiologists must document the Keros type in every presurgical sinus CT report.

The lamina papyracea (paper plate), separating the ethmoid cells from the orbital contents, is typically less than 0.5 mm thick and can be naturally dehiscent in up to 14% of individuals, creating direct channels for orbital spread of sinusitis. Similarly, the internal carotid artery is directly adjacent to the lateral wall of the sphenoid sinus in approximately 22% of patients, and may even bulge into the sinus lumen without intervening bone — a critical presurgical finding that must be documented proactively.^[7]

Key anatomical variants relevant to reporting

Several developmental variants are encountered frequently on sinus CT and carry direct surgical implications:

• Concha bullosa — pneumatisation of the middle turbinate, present in up to 34% of patients, may narrow the infundibulum and predispose to OMU obstruction.
• Haller cells (infraorbital ethmoid cells) — ethmoid cells extending inferior to the orbital floor along the roof of the maxillary sinus; reduce the natural ostium and predispose to maxillary sinusitis.
• Agger nasi cells — the most anterosuperior ethmoid cells, immediately anterior to the attachment of the middle turbinate; obstruct frontal sinus drainage when enlarged.
• Onodi cells (sphenoethmoid cells) — posterior ethmoid cells that pneumatise superiorly and laterally to the sphenoid, placing the optic nerve at risk during posterior ethmoidectomy; present in up to 40% of individuals.
• Deviated nasal septum (DNS) — common, often associated with compensatory contralateral middle turbinate hypertrophy, may bias drainage patterns and complicate endoscopic access.

Full HU reference table — paranasal sinuses

Scanning technique

Seven-step acquisition protocol

1. Patient positioning and chin tilt: Position the patient supine with arms at sides. The single most important positioning step is adequate chin tilt — the head must be extended sufficiently to orient the hard palate perpendicular to the scan table. This places the axial plane true to the dental occlusal plane, preventing partial volume artefacts across the maxillary alveolus and ensuring the full frontal sinus is captured within the superior extent of the FOV. Use the gantry tilt laser to confirm alignment. Document chin elevation angle for reproducibility.
2. Scout radiograph review: Acquire a lateral scout (scanogram) at low dose before committing to helical acquisition. Review the scout to confirm: (a) both frontal sinus superior margins are visible, (b) the hard palate is visible and horizontally oriented at the lower extent, (c) the posterior sphenoid wall is included, and (d) no neck flexion is introducing obliquity. Adjust FOV and start/end coordinates accordingly before pressing scan.
3. Technical parameters: Acquire at 100 kVp, 80–120 mA (with dose modulation active), pitch 1.0, rotation time 0.5 s, and collimation of 0.625 mm × 64 (or equivalent). Use a tight helical pitch to ensure overlapping data for high-resolution multiplanar reformats. The narrow beam creates a raw dataset from which 1.0 mm coronal, 1.0 mm axial, and 1.0 mm sagittal reconstructions can all be generated without re-scanning.
4. FOV and matrix selection: Centre the FOV symmetrically on the nasal septum using a small (16–18 cm) targeted FOV to maximise spatial resolution. Use a high-resolution bone kernel (e.g. B70 sharp on Siemens, Bone+ on GE, FC81 on Toshiba). A 512 × 512 matrix is standard; select 1024 × 1024 when available and when evaluation of fine ossicular-like structures (e.g. superior turbinate lamellae, ethmoid septa) is paramount. Avoid extending the FOV to include the entire face — this dilutes spatial resolution without clinical benefit for a sinus-specific study.
5. Multiplanar reconstruction (MPR): Generate coronal reformats at 1.0 mm slice thickness, 0.5 mm increment, in the true coronal plane (perpendicular to the hard palate line on the lateral scout). Coronals are the primary diagnostic plane for OMU anatomy and anterior sinus drainage assessment. Generate axials at 2.0 mm for surgical overview and sagittals at 1.5 mm for frontal recess and skull base depth assessment. All planes must be window-levelled for bone (window: 2000–3000, level: 600–700) before being sent to PACS.
6. Patient instruction during scan: Instruct the patient to remain completely still, breathe quietly through the nose without swallowing, and avoid any motion during the 4–8 second helical acquisition. Swallowing motion during scan introduces starburst artefacts across the nasopharynx and posterior nasal choanae. If the patient is paediatric or uncooperative, consider scan time optimisation and liaise with the referrer regarding the use of sedation before the appointment, as restoring motion-corrupted sinus data is not possible post-acquisition.
7. Quality check before patient dismissal: Before releasing the patient, review the coronal bone-window reconstructions at the PACS or console for: complete coverage from frontal sinus to hard palate, absence of swallowing artefact, adequate delineation of the uncinate process and ethmoid bullae, visibility of the sphenoid sinus and clivus, and dental amalgam streak artefact assessment. If any of these checkpoints fails, the cause must be identified and — where correctable without significant added dose — addressed with a targeted repeat.

Scanner comparison table — 16-slice to 320-slice

Dual-energy and photon-counting CT protocol options

Deep learning reconstruction (DLR)

Deep learning reconstruction (DLR) algorithms — commercially available as Canon’s AiCE, GE’s TrueFidelity, Siemens’ ADMIRE at the AI tier, and Philips’ TrueFidelity equivalent — have demonstrated significant benefits in low dose paranasal sinus CT. In phantom and clinical studies, DLR reduces image noise by up to 50–60% compared to filtered back-projection (FBP) at equivalent dose levels, while preserving or enhancing edge sharpness at bony interfaces critical for sinus reporting.^[9]

For low dose sinus CT (effective dose target ≤ 0.15 mSv), DLR is now the reconstruction of choice on compatible platforms, as it can maintain diagnostic image quality at mA settings as low as 40–60 mA in adults of normal habitus. Radiographers should use DLR as the default output for sinus studies where available, archive the raw FBP series for technical QA purposes, and clearly annotate the DICOM header with the reconstruction type applied. Centres not yet equipped with DLR should default to iterative reconstruction at its highest strength tier.

Contrast media protocol

Rationale for non-contrast acquisition

The diagnostic objectives of sinus CT — characterising mucosal disease burden, mapping the OMU drainage pathways, identifying anatomical variants, detecting bony erosion, and localising high-attenuation foreign material such as fungal concretions or calcified osteomata — are all achievable exclusively through bone-windowed and soft-tissue-windowed NCCT reconstructions. Adding contrast would increase both cost and patient risk without meaningful diagnostic incremental value for these specific objectives.

The physiological basis further supports non-contrast acquisition: in chronic rhinosinusitis and inflammatory polyp disease, the pathological diagnosis rests on mucosal volume, OMU obstruction architecture, and bone remodelling patterns — none of which require vascular opacification. Even incidental mucous retention cysts, the single most common sinus CT finding (present in up to 33% of asymptomatic adults), are characterised entirely by their HU profile (typically +10 to +25 HU) and their smooth, dome-shaped morphology without contrast.

When to escalate to contrast-enhanced CT or MRI

Safety checklist for non-contrast sinus CT

Although no contrast is administered, the following pre-scan safety checks remain mandatory:

• Pregnancy status check in all women of childbearing age — document and follow institutional fetal dose policy
• Implants and metallic foreign body screening — dental implants generate streak artefact; document prior to scan and note in report
• Claustrophobia / cooperation status — sinus CT requires a minimum of 4–8 seconds of stillness; pre-empt patient coaching
• Radiation history for radiation dose accumulation tracking — particularly in paediatric patients or those with chronic sinusitis requiring repeat imaging
• Confirm referral indication is appropriate for NCCT (not a contrast-requiring indication per escalation criteria above)

Radiation dose

Diagnostic reference levels and typical values

The EC RP 185 document establishes European diagnostic reference levels (DRLs) for CT paranasal sinuses at a CTDIvol of 25 mGy for adult patients.^[10] It is important to understand that DRLs represent the 75th percentile of dose distributions across surveyed facilities — they are not targets but ceilings; the goal of any optimisation programme is to be substantially below the DRL while maintaining diagnostic quality. Contemporary low dose protocols using 100 kVp and 80–120 mA with iterative or DLR reconstruction routinely achieve CTDIvol values of 8–14 mGy — 40–65% below the DRL — without diagnostic compromise.

Five dose reduction strategies for paranasal sinus CT

1. Tube voltage reduction to 100 kVp: Reducing from the conventional 120 kVp to 100 kVp lowers dose by approximately 30–36% while maintaining diagnostic bone resolution, as the high intrinsic contrast of the air-bone interface compensates for the modest increase in image noise at lower kVp. In patients of small head size, 80 kVp can be considered when supported by vendor dose modulation algorithms, yielding up to 55% dose saving.^[11]
2. Aggressive mA modulation (AEC): Automatic exposure control (AEC) systems adapt tube current on a slice-by-slice or projection-by-projection basis. For paranasal sinuses — where alternating high-density bone and very low-density sinus air create extreme variation in x-ray path length — angular AEC can reduce dose by a further 15–25% compared to fixed mA. Calibrate AEC thresholds specifically for head/sinus protocols as body presets will significantly over-dose the head region.
3. Iterative and deep learning reconstruction: Transition from FBP to hybrid iterative reconstruction (iDose, SAFIRE, ASIR) and ultimately to DLR (AiCE, TrueFidelity) enables dose reduction of up to 50–60% at equivalent perceived image quality. For sinus CT, DLR is particularly effective because its noise suppression is tuned to preserve thin, high-frequency bony structures — precisely the structures of most clinical interest.^[9]
4. Tight anatomical coverage: Define scan range from the superior margin of the frontal sinus to the floor of the maxillary sinus (inferior to hard palate) and no further. Every additional centimetre of scan length increases DLP proportionally. Avoid including the cranial vault unless a concurrent intracranial indication exists — if both brain and sinus evaluation are required, these are best separated into two protocols with individual dose accounting.
5. Paediatric-specific optimisation and clinical justification review: Children are at substantially greater lifetime radiation risk per unit dose than adults. In paediatric patients, apply the paediatric DRL values from AAPM Report 200 and the Image Gently Alliance, use weight- or size-based AEC calibration, and — critically — consider whether imaging is genuinely necessary. A structured referral pathway that reserves CT for surgical planning (and uses clinical assessment or MRI for initial diagnosis in children) can dramatically reduce population dose in this vulnerable cohort.^[12]

Top 10 pathologies on low dose CT paranasal sinuses

Pitfalls for radiographers

Primary scanning pitfall: inadequate chin tilt and FOV positioning error

The most significant and most frequently encountered scanning error in CT paranasal sinuses is failing to achieve adequate chin extension, combined with either dropping the superior FOV boundary too early (cutting out the frontal sinuses) or failing to extend the inferior FOV to include the full hard palate and lower alveolar margin. These errors do not merely reduce image quality — they render the scan diagnostically incomplete for FESS planning, as the frontal recess anatomy and the inferior maxillary sinus floor are precisely the regions where critical surgical landmarks must be visualised.

The positioning error can be insidious: the scout radiograph may appear acceptable to an eye not specifically trained to evaluate sinus scouts, yet close inspection reveals that the frontal sinus roofs are clipped at the uppermost scan start, or that the hard palate is absent from the lowermost coronal reconstruction. In FESS patients, the consequences include the surgeon operating without knowledge of the frontal recess configuration (specifically, the relationship of the agger nasi cell to the frontal ostium) or the inferior extent of maxillary sinus disease — increasing operative risk.

Pitfalls for radiologists

Primary interpretation pitfall: mucosal fluid obscuring subtle bone erosion

The most consequential interpretation error in CT paranasal sinuses is allowing the global impression of “sinusitis” to anchor attention to the fluid and mucosal findings while failing to systematically examine the underlying bony walls for evidence of subtle erosion or dehiscence. Thickened, retaining fluid pools within the maxillary or sphenoid sinuses — particularly when the sinus is near-opacified — produce a visual mass effect that suppresses the radiologist’s attention to the bony margins, exactly where pathological destruction may be occurring.

This pitfall is amplified by a common cognitive heuristic: inflammatory sinusitis is far more frequent than sinonasal malignancy. The base rate favours benign disease so heavily that the reviewing radiologist’s pattern-recognition system may categorise an opacified sinus as “chronic sinusitis” before fully analysing the bone windows. Bony erosion — irregular, permeative, and not the smooth remodelling seen with mucocele or the extrinsic pressure erosion seen with inverted papilloma — demands a completely different diagnostic pathway, including immediate escalation to contrast MRI and oncology referral.

Pitfalls for non-radiology physicians

General practitioners, ENT surgeons, and emergency physicians frequently review sinus CT images before or independently of a formal radiology report — particularly in out-of-hours settings or high-volume surgical planning clinics. The following pitfalls document the most dangerous interpretive errors seen in clinical practice among non-radiologist clinicians.

Pitfall comparison summary

AI and automation in CT paranasal sinuses

Artificial intelligence applications in sinonasal CT imaging have moved from research to clinical deployment at a pace that mirrors the broader adoption of AI in head and neck radiology. The primary value propositions are automated Lund-Mackay scoring, structured report generation, anatomical variant detection, and the identification of high-risk presurgical findings that demand explicit documentation.

Automated Lund-Mackay scoring

The Lund-Mackay (LM) scoring system, while widely adopted, suffers from significant interobserver variability — studies have reported interrater reliability coefficients (Cohen’s kappa) as low as 0.48–0.62 even among experienced radiologists, largely due to disagreement on the status of the OMU (0 = clear, 1 = partially occluded, 2 = totally occluded) and the distinction between aeration and opacification in partially filled sinuses.^[13] Deep learning algorithms trained on large annotated sinus CT datasets have achieved automated LM scoring with agreement rates of up to 0.78–0.85 ICC against expert consensus — substantially exceeding typical human interobserver reliability — and several of these systems have received CE marking in Europe for clinical assistance roles.^[14]

Anatomical variant detection and FESS safety AI

Perhaps the most clinically impactful AI application in sinus CT is the automated detection and flagging of high-risk anatomical variants that must be communicated to the operating surgeon. AI systems capable of automatically classifying Keros type, identifying Onodi cells, measuring anterior skull base height, flagging dehiscent lamina papyracea, and detecting internal carotid artery exposure within the sphenoid sinus have been developed and validated in single- and multi-centre studies.^[15] These tools reduce the risk that a presurgical report omits a critical safety finding through oversight, fatigue, or time pressure.

AI-driven sinonasal pathology detection

Convolutional neural networks (CNNs) trained on annotated sinus CT datasets have demonstrated sensitivity of 88–94% for detecting fungal mycetomas, outperforming the visual assessment of non-specialist radiologists in studies from tertiary rhinology centres.^[16] False-positive rates for mycetoma versus inspissated secretion discrimination remain a challenge — a problem that AI systems partially address through automated HU profiling and texture analysis that mimics (and scales) the manual HU measurement workflow recommended in expert practice. Regulatory-cleared AI tools for sinonasal pathology detection should be viewed as a safety net for high-volume reporting environments rather than a replacement for specialist radiological interpretation.

Structured and templated reporting automation

Natural language processing (NLP) and structured reporting platforms are increasingly embedded within PACS and reporting workflows to ensure that sinus CT reports include all clinically mandated elements. Studies have shown that in a majority of routine sinus CT reports, key variables including Keros type, OMU status, lamina papyracea integrity, and the presence of anatomical variants are either absent or inconsistently documented.^[17] AI-driven structured reporting tools prompt the radiologist through a defined checklist, ensuring completeness and generating standardised, machine-readable reports that can be used for surgical planning software integration.

Conclusion

Low dose CT paranasal sinuses is one of the highest-value imaging studies in clinical radiology precisely because so much hinges on the quality of a study that can be acquired in under ten seconds. The 100 kVp, 80–120 mA protocol with bone-kernel reconstruction and coronal MPR output meets all diagnostic requirements for presurgical FESS planning, inflammatory disease staging, and sinonasal mass characterisation — while delivering an effective dose of 0.10–0.20 mSv that is among the lowest of any CT examination in the body.

The anatomy of the paranasal sinuses is deceptively complex: the OMU, the ethmoid labyrinth, the Keros classification of the olfactory fossa, the variants of Onodi cells and concha bullosa, and the adjacent orbital and intracranial danger zones are all structures that must be actively sought and systematically reported to give the operating ENT surgeon the information needed to perform FESS safely and effectively. HU measurement remains a non-negotiable step in characterising intrasinus content — a practice that separates the fungal mycetoma (+80 to +220 HU) from inspissated secretion, the blood (+45 to +80 HU) from simple effusion, and the mucocele from the bony-eroding malignancy.

The three-tier pitfall framework presented in this article — encompassing radiographer scanning errors, radiologist interpretation errors, and non-radiologist clinical errors — reflects the multidisciplinary nature of sinonasal CT reporting and the reality that diagnostic error can arise at any point in the imaging chain. The most serious pitfalls across all three groups share a common mechanism: premature cognitive closure on a benign inflammatory diagnosis without systematic bone-window assessment and without explicit documentation of the variant anatomy and danger zones that determine surgical risk.

AI-assisted scoring, anatomical variant detection, and structured reporting are already demonstrably improving the completeness and reproducibility of sinus CT reports in early-adopter centres, and their integration into routine clinical workflow represents a direct opportunity to reduce preventable operative complications. The combination of optimal low dose acquisition, disciplined systematic reporting, and AI-augmented safety checklists positions the modern radiology department to deliver sinus CT interpretation that meets the highest standards of evidence-based practice.

References

1. Lund, V. J., & Mackay, I. S. (1993). Staging in rhinosinusitis. Rhinology, 31(4), 183–184. PMID: 8140385
2. Blackwell, D. L., Lucas, J. W., & Clarke, T. C. (2014). Summary health statistics for U.S. adults: National Health Interview Survey, 2012. Vital & Health Statistics, Series 10, (260), 1–161. https://www.cdc.gov/nchs/data/series/sr_10/sr10_260.pdf
3. Fokkens, W. J., Lund, V. J., Hopkins, C., Hellings, P. W., Kern, R., Reitsma, S., … & Agius, A. M. (2020). European Position Paper on Rhinosinusitis and Nasal Polyps 2020. Rhinology, 58(Suppl S29), 1–464. https://doi.org/10.4193/Rhin20.600
4. Kim, Y. M., Jung, S. H., Oh, J. H., Kim, S. H., Lee, S. Y., Rhee, C. S., & Min, Y. G. (2019). Ultra-low-dose CT of the paranasal sinuses: comparison of image quality and radiation exposure with conventional CT. European Radiology, 29(11), 5800–5808. https://doi.org/10.1007/s00330-019-06138-5
5. Som, P. M., & Curtin, H. D. (2011). Head and Neck Imaging (5th ed.). Mosby Elsevier.
6. Keros, P. (1962). Über die praktische Bedeutung der Niveauunterschiede der Lamina cribrosa des Ethmoids. Zeitschrift für Laryngologie, Rhinologie, Otologie, 41, 809–813.
7. Wang, J., Bidari, S., Inoue, K., Yang, H., & Rhoton, A. (2010). Extensions of the sphenoid sinus: a new classification. Neurosurgery, 66(4), 797–816. https://doi.org/10.1227/01.NEU.0000367619.24413.5F
8. deShazo, R. D., Chapin, K., & Swain, R. E. (1997). Fungal sinusitis. New England Journal of Medicine, 337(4), 254–259. https://doi.org/10.1056/NEJM199707243370407
9. Greffier, J., Hamard, A., Pereira, F., Barrau, C., Pasquier, H., Beregi, J. P., & Frandon, J. (2020). Image quality and dose reduction opportunity of deep learning image reconstruction algorithm for CT: a phantom study. European Radiology, 30(7), 3951–3959. https://doi.org/10.1007/s00330-020-06724-w
10. European Commission. (2014). Radiation Protection No. 185: European Guidelines on Diagnostic Reference Levels for Paediatric Imaging. Publications Office of the European Union. https://doi.org/10.2833/7908
11. Brody, A. S., Frush, D. P., Huda, W., & Brent, R. L. (2007). Radiation risk to children from computed tomography. Pediatrics, 120(3), 677–682. https://doi.org/10.1542/peds.2007-0910
12. Image Gently Alliance. (2022). Sinus CT optimisation in children: Image Gently clinical recommendations. https://www.imagegently.org
13. Bhatt, N. R., Bhatt, J. M., & Hopkins, C. (2017). Variation in reporting practice and use of the Lund-Mackay staging system in the United Kingdom: are we comparing like with like? The Laryngoscope, 127(11), 2470–2474. https://doi.org/10.1002/lary.26731
14. Lee, D. J., Hamghalam, M., Wang, L., Lin, H. M., Colak, E., Mamdani, M., Simpson, A. L., & Lee, J. M. (2025). The use of a convolutional neural network to automate radiologic scoring of computed tomography of paranasal sinuses. BioMedical Engineering Online, 24(1), 49. https://doi.org/10.1186/s12938-025-01376-7
15. Chang, E. H., & Alandejani, T. (2021). Artificial intelligence and sinonasal imaging: current concepts and future directions. Current Otorhinolaryngology Reports, 9(4), 246–253. https://doi.org/10.1007/s40136-021-00368-y
16. Gu, X., Ouyang, C., Zhao, Y., & Chen, X. (2022). Deep learning for the detection of maxillary sinus fungal balls on computed tomography: a retrospective study. European Archives of Oto-Rhino-Laryngology, 279(2), 773–779. https://doi.org/10.1007/s00405-021-06910-0
17. Becker, S. S., & O’Malley, B. B. (2010). Evaluation of sinus computed tomography scans: a collaborative approach between radiology and otolaryngology. Current Opinion in Otolaryngology & Head and Neck Surgery, 18(1), 22–28. https://doi.org/10.1097/MOO.0b013e32833463e6
18. Subbiah, N. K., Bakshi, S. S., Arumugam, S., & Ghoshal, J. A. (2025). Clinical and radiological significance of anatomical variations in paranasal sinuses: a retrospective CT-based study. Cureus, 17(4), e82506. https://doi.org/10.7759/cureus.82506
19. Stammberger, H., & Kennedy, D. W. (1995). Paranasal sinuses: anatomic terminology and nomenclature. Annals of Otology, Rhinology & Laryngology, 104(10 Suppl 167), 7–16. https://doi.org/10.1177/00034894950104S203
20. Zinreich, S. J., Kennedy, D. W., Rosenbaum, A. E., Gayler, B. W., Kumar, A. J., & Stammberger, H. (1987). Paranasal sinuses: CT imaging requirements for endoscopic surgery. Radiology, 163(3), 769–775. https://doi.org/10.1148/radiology.163.3.3575731
21. Khandelwal, A., Sheth, S., Jain, R., & Bhatt, N. (2020). Radiation dose from sinus CT examinations: review of current state and evidence for dose reduction. Head & Neck, 42(8), 2068–2079. https://doi.org/10.1002/hed.26152
22. Greschus, S., Albert, F., & Eichhorn, K. W. G. (2017). Imaging of the paranasal sinuses and anterior skull base. HNO, 65(6), 490–503. https://doi.org/10.1007/s00106-017-0341-0
23. Aribandi, M., McCoy, V. A., & Bazan, C. (2007). Imaging features of invasive and noninvasive fungal sinusitis: a review. Radiographics, 27(5), 1283–1296. https://doi.org/10.1148/rg.275065189
24. Kim, J. S., & Lee, J. H. (2023). Current applications of deep learning in sinonasal imaging. Radiology: Artificial Intelligence, 5(2), e220202. https://doi.org/10.1148/ryai.220202
25. Hwang, S. H., Kim, S. T., Park, J. H., Cho, J. H., Han, K. T., & Kang, J. M. (2018). Preoperative assessment of CT findings in patients with chronic rhinosinusitis. Clinical and Experimental Otorhinolaryngology, 11(3), 176–181. https://doi.org/10.21053/ceo.2017.01614
26. Javer, A. R., Gheriani, H., Mechor, B., Flamer, D., Genoway, K., & Yunker, W. K. (2019). Effect of intraoperative corticosteroid injection into the sinus ostia on postoperative sinus patency. American Journal of Rhinology & Allergy, 23(1), 60–63. https://doi.org/10.2500/ajra.2009.23.3263
27. Rosenfeld, R. M., Piccirillo, J. F., Chandrasekhar, S. S., Brook, I., Ashok Kumar, K., Kramper, M., … & Corrigan, M. D. (2015). Clinical practice guideline (update): adult sinusitis. Otolaryngology–Head and Neck Surgery, 152(2 Suppl), S1–S39. https://doi.org/10.1177/0194599815572097
28. International Commission on Radiological Protection. (2007). ICRP Publication 103: Recommendations of the International Commission on Radiological Protection. Elsevier. https://doi.org/10.1016/j.icrp.2007.10.003
29. Rotter, G., Müller, D., Röhrig, F., Malzahn, U., Duerr, H. R., & Kamperidis, V. (2021). Photon-counting CT for sinonasal imaging: a feasibility study at ultra-low radiation dose. European Radiology Experimental, 5(1), 38. https://doi.org/10.1186/s41747-021-00234-0