The focus of this study was on the intracranial and orbital manifestations of COVID-19 associated with Rhino-cerebral mucormycosis. The role of MRI in the evaluation of mucormycosis has been well described previously. [9, 10]
Early signs of mucormycosis include the presence of non-enhancing turbinates, described as the black turbinate sign. [11, 12] Similar findings were also seen in our study [Figure 9].
Our findings included a significantly heterogeneous appearance within the involved sinuses, including T1 isointensity or hyperintensity, as well as T1/T2 hypointensity or isointensity [Figure 1].
Patients suffering from diabetes have already been known to be more prone to mucormycosis.  The European registry of patients with mucormycosis has also revealed a strong association between mucormycosis and diabetes as well as corticosteroid use. [13, 14] There is also evidence of an association between diabetes and SARS COV-2, including triggering of diabetic ketoacidosis. [15, 16] COVID‑19 disease itself may also predispose to invasive fungal infections, especially of the sinuses. [17, 18] By demonstrating the association of these conditions, this study gives objective clues regarding the sudden surge in mucormycosis cases. 
Prior to the COVID-19 pandemic also, mucormycosis had a higher incidence in the Indian subcontinent than in any other region of the world, probably related to the higher burden of diabetes in India.  Susceptibility to Rhino-orbito-cerebral mucormycosis in COVID-19 patients is a consequence of decrease in phagocytic activity and an increase in the accessible iron. Protons in in diabetic ketoacidosis stimulate a transferrin-mediated displacement of iron. Fungal haeme oxygenase promotes iron absorption for fungal growth. This iron dependence also explains the angioinvasive propensity of fungal pathogens. 
Our study also included a case of post-COVID Rhinocerebral mucormycosis in a patient with history of solid organ transplantation as well as one with a haemo-lymphatic malignancy.  Similar phenomena have been documented previously. [23, 24, 14] However, this has rarely been reported in a post-COVID-19 context. 
Our study predominantly utilized MRI for the evaluation of rhino-cerebral mucormycosis. MRI not only possesses excellent soft-tissue resolution but is also free from ionizing radiation. MRI is more likely to be safe in the evaluation of these patients because patients with mucormycosis often need administration of nephrotoxic drugs.  CT was reviewed only in cases where it was needed to retrospectively evaluate the bony structures, and to help distinguish between susceptibility due to hemorrhage versus susceptibility due to fungal elements. But it was only rarely needed in our study. Our study also revealed a large number of cases in which there was spread of the disease without underlying bone erosion. This is because the intrinsic pathway of spread is along blood vessels or nerves and does not necessarily require bone erosion. Skull base osteomyelitis is also well delineated on MRI [Figure 10].
There are several unanswered questions regarding the pathways of spread of mucormycosis. Mucormycosis initially involves the nasal mucosa. This is followed by spread to the paranasal sinuses (typically the ethmoid and maxillary sinuses), then the orbit, and ultimately the intracranial fossa.  Ethmoid sinuses were found to be most frequently involved in our study. This is in keeping with findings described in a recent study by S Sharma et al .
The frequency of involvement of structures may also help in determining the temporal evolution of the disease process. The direct pathways of invasion of various structures observed in our study, also help to advance the understanding of the pathogenesis of mucormycosis.
According to our study, the dominant pathway of spread of mucormycosis is across the ethmoidal air cells and the lamina papyracea into the orbit and then into the medial and inferior aspect of the extraconal fat. Subsequently, the medial and inferior rectus muscles are involved. However, the inferior aspect of the extraconal fat of the orbit may also be invaded across the orbital floor.
The superior rectus muscle is typically involved via the spread into the superior extraconal space. This is also mostly through the lamina papyracea and occasionally via the frontal bone from the frontal sinus through the superior wall of the orbit. Invasion of orbital apex following intraconal invasion may result in orbital apex syndrome [Figure 11]. Optic neuritis may ensue [Figure 12]. Invasion of the superior orbital fissure which contains cranial nerves III, IV, and VI, and branches of V1 and V2, may result in ophthalmoplegia, diplopia, paraesthesia, and loss of sensations in the corresponding territories of the cornea and face. All cases of cavernous sinus thrombosis in our study had associated orbital apex involvement (Figure 6). This is in keeping with common pathways of spread into the cavernous sinus. [25, 27] However, there were two cases in which there was posterior intraconal space and optic nerve involvement without any extraconal fat or extraconal muscle involvement. These cases had enhancing soft tissue along the inferior orbital fissure and pterygopalatine fossa, which was indicative of the spread of the disease from the ipsilateral pterygopalatine fossa and along the inferior orbital fissure into the orbit. 
There were also two cases in our study that showed direct invasion of the left orbital apex from the sphenoid sinus.
The major reservoir of mucormycosis is considered to be the pterygopalatine fossa. 
Pathways of intracranial spread most frequently included, the orbital apex  with cavernous sinus involvement spreading into the adjacent middle cranial fossa, medial to the anterior temporal lobe with pachymeningeal enhancement extra-axial and intraparenchymal abscesses in this region [Figure 8].
Another pathway of spread is across the cribriform plate and the basifrontal region (anterior cranial fossa extension),  causing subdural abscesses [Figure 14, Figure 15] and adjacent vasculitic infarcts as well as intraparenchymal abscesses.
All cases of intraparenchymal and extra-axial hemorrhages in our study were located in the basifrontal region [Figure 9]. Intracranial hemorrhage is probably due to mycotic aneurysms and angioinvasive disease.  We observed mycotic aneurysms in the cavernous ICA [Figure 13], as well as in the posterior cerebral artery.
We also observed contiguous involvement of the infraorbital nerve in 30/58 (51.7%) patients. This was found to cause widening of the infraorbital foramen. The disease spread was then found to be along the inferior orbital fissure progressing to the pterygo-palatine fossa. A few cases also showed extension across the foramen rotundum into the middle cranial fossa.
Intracranial extension into the posterior cranial fossa was also demonstrated with vasculitic infarcts in the left superior cerebellar artery territory. There was also a case of a right middle cerebellar peduncle abscess. This represented extension along the cisternal segment of right trigeminal nerve. Only a few reports have documented this phenomenon. [31, 6] This indicates a pathway of perineural spread of the disease directly into the posterior cranial fossa [Figure 16].
Diffusion restriction is often seen and helps in evaluating the extent of involvement of various structures including the optic nerve, orbital apex, and the intracranial spread [Figure 17]. 
Fungal abscesses may present with only peripherally restricted diffusion [Figure 8] or heterogeneous diffusion restriction within the abscesses. This variant is seen in our study and also has been described before by Rangarajan et al. and Gaviani et al. [32, 33]
Infarcts included vasculitic or embolic infarcts as well as watershed territory infarcts in cases of complete ICA thrombosis [Figure 5].
In previous studies on non-COVID associated mucormycosis, the most commonly documented findings in the intracranial extension of mucormycosis, apart from meningeal enhancement, have been intracranial infected soft tissue/abscesses in 50% patients with intracranial extension and infarcts only in 20% cases  Therakathu et al have demonstrated 31% involvement of intracranial structures out of which, 38% showed cerebritis or abscesses, and 30% cases showed infarcts.  In comparison, our study, where 31 cases showed intracranial involvement, demonstrated infarcts in 55% of cases and internal carotid artery thrombosis in 13.8% cases. This suggests a possibility of higher occurrence of arterial thrombotic events including carotid artery involvement and infarcts in COVID-19 associated mucormycosis.
We have only considered acute infarcts with diffusion restriction on DWI in our current study.
Given, the association of COVID-19 with various thrombotic phenomena and the findings of our study, the possibility of a higher frequency of thrombotic complications in COVID19 associated mucormycosis vis-à-vis the proportion of these complications in patients with mucormycosis alone, merits consideration.
All cases of infratemporal fossa involvement in our study also had ipsilateral pterygomaxillary fissure involvement. This corroborates the theory of spread into infratemporal fossa via the pterygomaxillary fissure.
Panophthalmitis [Figure 17] and optic neuritis [Figure 12] are devastating complications of Rhinocerebral mucormycosis and may result in permanent blindness. Their imaging findings have been described in Table 2. Early detection and prompt management of these complications may help to preserve vision.
However, in some cases with panophthalmitis, orbital exenteration may be needed. Here too, MRI plays a crucial role in the decision-making process.