Introduction

Visual sensory deprivation (e.g. Charles Bonnet syndrome) and several brain disorders, including strokes in visual areas, may induce hallucinations [1,2,3,4,5]. Their underlying mechanisms remain unclear due to several factors. Firstly, these phenomena are particularly complex and likely entail various dysfunctions of specific brain regions and their connections (hodotopic dysfunctions) [6,7,8]. Secondly, the transient and dynamic nature of visual hallucinations makes them difficult to explore, many of the available studies being conducted at a distance from symptom occurrence [5, 9]. Finally, the dogma of localization in clinical neurology focuses on perceptual deficits, negative phenomena that are easier to qualify and quantify, and much less on positive phenomena, which are more subjective and more difficult to assess.

We report two cases with visual hallucinosis and hypoperfusion of the right occipito-temporal cortex, without infarction of the visual cortex, and we propose a practical and structured algorithm for the evaluation of visual hallucinations related to neurological conditions. This algorithm allows an organized exploration of the hallucinatory phenomena and the localization of the involved brain networks. It includes the classification of the hallucinations, the context of their appearance, their chronology, the assessment of insight and of the attributes of the hallucinated images (see Table 1). Since it explores dynamic, often transient, perceptual phenomena, guided analysis is particularly useful when performed while hallucinations are actually occurring and much less so when performed at a distance.

Table 1 Algorithm for semiological analysis of visual hallucinations
Full size table

Methods

Case 1

An 85-year-old woman was admitted following transient left hemianopia, manifesting as a “dark veil on the left” that lasted about 10 h. Her medical history included high blood pressure, atrial fibrillation, breast cancer at age 51 and contralateral recurrence at age 78, pulmonary embolism, and bilateral cataract surgery 4 years ago. Her neurological exam was normal. MRI revealed a punctiform area of restricted diffusion in the right ventro-lateral thalamus (motor relay region) and subocclusion of the P2 segment of the right posterior cerebral artery (Fig. 1A, B).

Fig. 1
figure 1

Patient 1 (left): A hyperintensity in the right ventro-lateral thalamus on the diffusion weighted imaging (DWI) sequence corresponding to a punctiform acute ischemic lesion and B significant stenosis of the P2 segment of the right posterior cerebral artery on the time-of-flight (TOF) sequence (white arrow). Patient 2 (right): C, D multiple arterial irregularities (white arrows) on the TOF sequence suggestive of reversible cerebral vasoconstriction syndrome

Full size image

The following day, while free of other neurological symptoms, she experienced isolated simple and complex visual hallucinations, more often in darkness or dim light. Insight was preserved. A repeat brain MRI revealed hypoperfusion in the right occipito-temporal cortex without new infarction (Fig. 2A, B). EEG during hallucinations, ophthalmological examination and automated perimetry were normal. The hallucinations gradually decreased in frequency ceasing 3 days later. Repeat brain MRI several days later showed resolution of the P2 segment subocclusion.

Fig. 2
figure 2

Patient 1 (left) and Patient 2 (right): A fluid attenuated inversion recovery (FLAIR) and B the corresponding perfusion-weighted imaging (PWI) maps (Olea Sphere®, Olea Medical) showing hypoperfusion in the right mesial occipito-temporal cortex in both patients (black arrows)

Full size image

Patient's hallucinations were assessed shortly after their first appearance, when they were manifest, and the evolution of the phenomenology was followed daily until its resolution. Simple hallucinations consisted of colored splashes and branching shapes (dendropsias). Her complex hallucinations were clear wall writings, people and plants smaller than real (micropsia), clear and brightly colored. All had neutral emotional impact. Through guided analysis, patient noticed that all hallucinations manifested in her left visual field. The writings were illegible and appeared static. The people and the plants showed depth and lateral shifting without biological movement, resulting in an ellipsoid trajectory, and the transition between the hallucinations and the background was blurred. Noteworthy, lateral component of the translational movement was present during steady fixation and was also influenced by the eye movements (e.g. changes in the perceived trajectory or speed).

Case 2

A 55-year-old woman with history of untreated essential tremor was admitted following a third episode of thunderclap headache. Neurological exam was normal. MRI showed multifocal arterial irregularities suggestive of RCVS (Fig. 1C, D). No precipitating factors were identified and nimodipine was started.

Several hours later, on the evening of the admission day, patient reported isolated simple and complex visual hallucinations with preserved insight. A repeat MRI showed hypoperfusion of the right occipito-temporal cortex without parenchymal lesions (Fig. 2A, B). EEG during hallucinations, ophthalmological exam and automated perimetry were normal. Hallucinations gradually decreased in frequency, ceasing after 2 days. Repeat MRI at one-month follow-up showed resolution of the segmental arterial constrictions, confirming the RCVS, and absent parenchymal lesions.

Patient's hallucinations were assessed soon after their first appearance, when they were manifest, and their progress was monitored daily until they resolved. Simple visual hallucinations included a bright equilateral triangle sliding from the left towards the center of the visual field, replaced several hours later by complex hallucinations of people, animals and plants. Her complex hallucinations were normal-sized, clear, vividly colored, well integrated into background, “moving” and emotionally neutral, except for one reported as a “scary head with holes in its face, large red cheeks and small black eyes”. Through guided visual analysis, patient noted that all hallucinations manifested in the left visual field and that hallucinations of people, animals and plants had in-depth and lateral shifting without biological movement resulting in an ellipsoid trajectory, and a blurred transition with the background. As with the first patient, the lateral component of translational motion was present during steady fixation and was further impacted by eye movements (e.g. changes in the perceived trajectory or speed).

Discussion

The occurrence of visual hallucinations during cerebral hypoperfusion in the ventral visual stream without infarction or underlying neurodegenerative disorder suggests dysfunctions in the visual networks [10, 11]. This is an interesting, and to our knowledge, a previously unreported phenomenon.

While in the first patient, simple and complex hallucinations were intermingled in the course of the hallucinatory phenomena, in the second patient there was a progression from simple to complex hallucinations. Insight was preserved and the hallucinations were spontaneously perceived as real and natural. To emphasize the veracity of the perceived images, patients focused on aspects like the high resolution of certain parts of the image (e.g., animal fur, leafs veins) while disregarding others (e.g., motion patterns, legibility of writing) and the guided analysis performed during the phenomenology pointed out peculiarities of certain attributes (see Table 1).

In our patients, hypoperfusion could have induced an aura-like, cortical spreading depression phenomenon similar to the one described in the penumbra region of ischemic stroke [12,13,14]. Moreover, hypoperfusion likely had an unequal impact on different components of the mesial occipito-temporal cortex inducing variable degrees of dysfunction in the visual hubs. Indeed, the V1 region is more sensitive to hypoxia due to higher cellular density and increased metabolic demand [15]. Therefore, abnormalities of the primary visual cortex may alter the output to higher-order visual areas allowing their spontaneous activation and elaboration of hallucinatory percepts [16]. Hallucinations content reflects the functional specializations of the involved brain regions [16,17,18], and since the regions encoding faces and objects have a right-hemisphere lateralization, this might explain why both reported patients showed hypoperfusion of the right occipitotemporal cortex. It is possible that the text hallucinations in Case 1 originated in areas of the right occipitotemporal cortex involved in the processing of font and handwriting [19, 20], and that guided analysis revealed their lack of semantic content through recruitment of the visual word form area and language-related areas of the left hemisphere [21].

The change in the perception of the hallucinations following the guided analysis could be due to several reasons. One possibility is that the hallucinations are intrinsically defective and that the analysis helped the patients become aware of the image flaws. Another possibility is that the hallucinations are inherently real when the patient focuses on them but their perception is altered by the guided redeployment of attention towards a global picture integrating the hallucinations into the environment. Indeed, this last hypothesis is endorsed by the fact that most of the flaws detected in the hallucinated images concerned attributes that interact or are impacted by the visual context, such as the type of movement, the relative size and the interface with the background. In physiological conditions, the detection of image boundaries is paramount in figure–background segregation [22], the segregated images being routed to the lateral occipital cortex and the background to more dorsal occipital areas [23]. Conversely, the detection of boundaries in the already segregated, hallucinated images might prompt their integration into the background and help identify flawed attributes. As altered communication between the canonical attentional control networks seems to be involved in the hallucinatory phenomena regardless of their origin [24, 25], guided analysis may therefore optimize the attentional deployment and, therefore, modulate the visual information processing and hallucination–background integration.

Conclusion

Hallucinated images, although spontaneously reported as “completely natural”, presented structural and kinetic anomalies revealing an altered, non-realistic rendering. Noteworthy, the impaired awareness of peculiarities in the attributes of the hallucinated images questions the accuracy of post hoc inquiries and points to the crucial role of detailed, structured semiological analysis in the acute phase, while visual hallucinations are actually occurring. This should be considered by future studies exploring positive perceptual phenomena.