IFOI can have devastating effects if there is delayed diagnosis and initiation of appropriate treatment. Misdiagnosis is common, with incorrect diagnoses ranging from bacterial OC and sinusitis to inflammatory conditions, such as non-specific orbital inflammation [4]. IFOI typically occurs via contiguous spread from the paranasal sinuses and increases mortality compared to orbit-sparing disease [8, 9]. Additionally, the superior orbital fissure and optic canal open directly into the middle cranial fossa and represent avenues of further intracranial extension [9]. There are various mimics of fungal orbital infections and radiological analysis, typically with MRI, remains pivotal in supplementing the clinical assessment.

Firstly, patient demographics and clinical features may assist in differentiating these two entities. Clinical features of invasive fungal rhinosinusitis and bacterial rhinosinusitis with orbital complications have been investigated by Piromchai et al., and findings from our study are largely consistent with prior literature (Table 1) [10,11,12]. Patients with IFOI were typically older and immunocompromised (eg: haematological malignancy, immunosuppressive medications) or co-morbid with diabetes mellitus. Both OC and IFOI had a male predominance and typically reported orbital pain, periorbital erythema/oedema and proptosis, however, patients with IFOI presented with a more subacute onset of ocular symptoms. Visual compromise, evidenced by reduced visual acuity and/or optic neuropathy, was more common in IFOI compared to OC. Prior studies exploring OC and rhino-orbital-cerebral-mucormycosis (ROCM) on CT suggested that ROCM was more likely to demonstrate EOM limitation, but eyelid swelling was less common [5]. In patients with these risk-factors, naso-endoscopy may help identify an invasive fungal infection, however, the typical grey-black discolouration of nasal mucosa, ulceration and necrotic tissue may occasionally be absent [11, 13]. A higher white-cell count and C-reactive protein was observed in OC, and may be attributable to the high incidence of immunocompromised patients with IFOI, and thus may not reflect disease activity or severity.

Typical CT and MRI features of invasive sinonasal fungal infections have been described [13,14,15,16,17,18,19]. CT may demonstrate mucosal thickening, either sclerotic or erosive bony changes, fat plane effacement in the pterygoid and infratemporal fossa, and a diffuse contrast-enhancing infiltrative mass invading the orbit. Venous or arterial thrombosis may also occur [2]. Son et al. has previously explored differentiating factors between ROCM and OC on CT, reporting that thickening of the sinus mucosa was more frequent in patients with ROCM compared to OC. However, this study included OC without paranasal sinus disease, whilst our study excluded non-sinogenic OC [5]. MRI may be more sensitive than CT in detecting acute invasive fungal sinusitis and can more readily demonstrate optic nerve and intracranial extension [2, 6, 7].

Within our series, the ethmoid sinus was most commonly involved in both IFOI and OC, consistent with prior studies [5, 12, 20, 21]. In a study of acute invasive fungal rhino-sinusitis, Huang et al. observed higher incidences of frontal, ethmoid and sphenoid sinus involvement in orbital disease, with infection primarily involving the medial and inferior orbit [21]. In IFOI, contrast-enhancement may imply active inflammation, whilst LoCE signifies tissues devitalisation and necrosis, corresponding to higher fungal load with areas of coagulative necrosis [22]. Early MRI features may include oedema in paranasal sinuses and orbital fat, with focal areas of LoCE. LoCE due to fungal invasion and tissue devitalisation, described as the ‘black turbinate sign’, has been noted as an early feature (Fig. 2) [23, 24]. In this study, LoCE in the paranasal sinuses with orbital extension was a specific feature of IFOI, absent in OC. In one case of IFOI, there was LoCE in soft tissues of the right sphenoid sinus and orbital apex contiguous with the pterygopalatine fossa and skull base (Fig. 1). Additionally, LoCE of the EOM at the orbital apex was suggestive of fungal invasion of the EOM, a feature previously described [25].

IFOI and OC had varying patterns of soft-tissue involvement. Firstly, abnormal soft tissue in nasolacrimal ducts, extraconal fat, and pterygopalatine fossa indicates imminent orbital and cerebral invasive fungal infection, and may be demonstrated by loss of normal T1 hyperintensity in the peri-sinus fat planes [17]. Fungal elements may be observed as low to intermediate T1 signal, and are generally T2 hypointense, although intermediate or high signals have occasionally been noted [14,15,16, 26]. Meanwhile, OC typically demonstrates T2 hyperintensity [12, 26]. Orbital involvement in IFOI was characterised by an infiltrative mass invading the orbital apex, contiguous with affected paranasal sinuses (Table 2). Inflammatory stranding in the pterygopalatine fossa, retro-antral and masticator space were indicative of an invasive fungal infection [17, 18].

A range of radiological features were observed depicting intracranial and/or extra-orbital/extra-sinonasal extension in OC and IFOI (Table 2). Intracranial extension in IFOI was represented by dural thickening and cavernous sinus expansion. Meanwhile, in OC there was dural thickening and/or contrast-enhancement, cerebritis and frontal lobe abscesses (demonstrating diffusion restriction on DWI). These findings are typical for each disease. The authors draw attention to the case depicted in Fig. 7, whereby MRI demonstrated abnormalities suggestive of an IFOI with superimposed bacterial OC. Firstly, LoCE in the left ethmoid region and permeative bony de-ossification of the frontal sinus, along with microbiological confirmation of intraoperative specimens, were consistent with an invasive orbital aspergillosis. However, this case also demonstrated superolateral orbital fat cellulitis, a lacrimal gland abscess (uncommon in IFOI) and a left frontal lobe abscess, as indicated by a peripherally contrast-enhancing lesion (Fig. 7). In its early stages, bacterial abscesses are seen as T1 hypointense and T2 hyperintense areas with minimal or heterogeneous contrast-enhancement [19]. Typically, fungal abscesses demonstrate a hypointense core with a surrounding iso- to mildly hyper-intense rim, with peripheral contrast-enhancement [2]. Almost all fungal abscesses show non-enhancing intra-cavitory projections directed centrally from the wall, and there may be high T2 signal in the core with a surrounding hypointense rim [27]. These features were not observed in the intra-orbital collection. The outer margin of the pyogenic abscess wall will be smooth or lobulated, in contrast to fungal lesions which will have crenated wall in more than half of the abscesses [19, 27]. A distinct feature which may differentiate between bacterial from fungal abscesses is the ‘dual rim’ sign on DWI, however DWI was not performed in this case [28]. Furthermore, pyogenic abscesses show diffusion restriction whereas fungal lesions may have variable diffusibility (i.e. variable DWI and ADC) [29]. These differences in radiological appearance may be attributed to the different composition of fungal abscesses. Fungal abscesses are typically multiple and can involve the basal ganglia, whilst bacterial abscesses are often solitary lesions sparing the basal ganglia [19, 30].

The expedient diagnosis of IFOI can be a complex challenge, due to variable patient risk factor, co-morbidities, clinical course and response to initial treatment, guiding further investigation and commencement of empirical treatment. Ultimately, the presence of the aforementioned radiological features are indicative of an invasive fungal infection and may help to differentiate IFOI from sinogenic OC. The presence of these extensive and/or specific radiological features in high-risk patient demographics (e.g.: highly immunocompromised, patients with poorly controlled diabetes mellitus) should prompt endoscopic evaluation and/or early surgical management to obtain intraoperative microbiological confirmation of an invasive fungal pathology [25, 31,32,33,34]. The most frequently affected sites, and hence the most sensitive for biopsy are the middle turbinate, nasal septum and floor of the nasal cavity [1, 13, 35].

Patients with IFOI demonstrated significantly worse prognosis, including visual compromise and mortality. In many cases, IFOI led to devastating and life-threatening outcomes, with many patients proceeding to enucleation. Meanwhile, prognosis in OC was more favourable, with majority of patients achieving complete resolution following an appropriate course of IV/oral antibiotics. The utility of pre-intervention MRI features in predicting visual and mortality outcomes in acute invasive fungal rhinosinusitis has been investigated [20]. In Idowu et al.’s study, orbital apex and cerebral artery (i.e. arteritis/arterial occlusion) involvement were risk factors for poor visual outcomes. Visual compromise in IFOI may be due to a combination of direct optic nerve invasion, ischemic and/or compressive optic neuropathy [36, 37]. Meanwhile, increased mortality was associated with facial soft tissue and nasolacrimal apparatus involvement.

There are several study limitations pertaining to the retrospective nature and small sample size, with limitations on statistical power. Additionally, at our tertiary institution, MRI scans were typically reserved for OC whereby there was diagnostic uncertainty and/or suspicion of severe complications, such as suspected intracranial extension. This may be reflected in the higher incidence of superior orbital collections and involvement of the superior muscle complex, supero-medial and supero-lateral orbital fat involvement in cases of OC. Thus, the inclusion criteria may depict cases of greater disease severity [38]. However, the authors note that cases of severe sinogenic OC encountered are more likely to reflect realistic clinical scenarios whereby expedient differentiation from IFOI is necessary and more challenging.