Amblyopia is traditionally defined as a congenital anomaly in the spatial–temporal processing of visual information (Noorden, 1977; Ciuffreda et al., 1978b; Hamasaki and Flynn, 1981; Asper et al., 2000a,b) occasioned by abnormal visual experience early in life associated with strabismus (misalignment of the visual axes of the two eyes), anisometropia (refractive differences between the eye), and/or visual deprivation (Webber and Wood, 2005). Studies such as those of noted that the saccadic reaction times to visual stimuli of strabismic amblyopic eyes were significantly longer than those by the dominant eye (Ciuffreda et al., 1978a; Hamasaki and Flynn, 1981). Strabismic amblyopia (SA) eyes were also known to show eye movement problems such as unstable fixation, and pursuit difficulty (Asper et al., 2000a; Kanonidou et al., 2014). More recently, studies have described similar temporal processing and eye movement deficits associated with anisometropic amblyopia (Niechwiej-Szwedo et al., 2010, 2011a,b). Thus, given that eye movements and attention are intrinsically linked (Wurtz et al., 1982), and visual attention to a particular stimulus or location is known to selectively amplify the activity of sensory neurons (Goldberg and Wurtz, 1972), it is not surprising that visual attention and perception are often anomalous through either strabismic and anisometropic amblyopic eyes making treatment for amblyopia difficult to achieve in older children and adults. Indeed a systematic review and meta-analysis of recent research into dichoptic training, perceptual learning and video gaming has noted that “some participants may benefit from the new treatments and aid some recovery of visual acuity through the amblyopic eye but not binocular stereo acuity” (Murphy et al., 2015; Tsirlin et al., 2015), suggesting that aspects of dynamic attentional function through the amblyopic eye continue to be defective.

Visual attention deficits have been reported in many previous studies of amblyopia (Asper et al., 2000a,b; Simmers et al., 2003; Ho et al., 2006; Popple and Levi, 2008), with evidence of temporal processing inadequacies and inability of the amblyopic eye to initiate eye movements and shift attention as rapidly as the dominant non-amblyopic eye (Goldberg and Wurtz, 1972; Noorden, 1977). Sharma et al. (2000) also reported that SA patients undercounted features presented to the amblyopic eye, although the attentional cueing effects were the same as for controls while Thiel and Sireteanu (2009) noted line bisection task deficits in SA patients similar to that seen in neuropsychological cases of hemineglect. Tsirlin et al. (2018) also utilized a conjunction visual search task that required top-down attentional processes to provide evidence of functional deficits in visual attention in amblyopes (Tsirlin et al., 2018). However, details of aspects of the neural mechanisms underlying visual attention deficits in amblyopia are still largely unknown.

Perceptual learning treatment for visual attention has been reported to be effective in treating amblyopia by improving visual acuity or restoring stereoscopic function monocularly for a short time (Huang et al., 2014; Tsirlin et al., 2015). However, when training is discontinued and binocular vision is available, the slower activation of voluntarily directed eye movements (and hence attention) through the strabismic or anisometropic amblyopic eye is likely to favor the dominant, faster to activate eye and reinduce amblyopia as an adaptive measure to prevent double vision, and temporal mismatch of images (Harrad et al., 1996), and potentially most importantly splitting of ocularly driven attention.

Such a potential scenario is supported by our previous brain imaging study that demonstrated that activation of the visual attention network driven by the strabismic amblyopic eye in performance on goal-directed and stimulus-driven visual attention tasks is reduced compared to that of the non-amblyopic eye (Wang et al., 2015, 2017). However, functional connectivity (FC) of the visual attention network in strabismic amblyopic patients in the resting state has not been addressed to date suggesting the need for investigation of FC of the cortical network structure in visual attention-related areas of strabismic amblyopes. On the other hand, there are a number of reports of abnormal connections in extra striate visual spatial network cortices and between hemispheres differences in anisometropic amblyopia during resting state functional magnetic resonance imaging (fMRI) (Ding et al., 2013; Joly and Franko, 2014; Wang et al., 2014).

Visual attention is primarily driven by the dorsal cortical visual stream, which includes the primary visual cortex (V1), middle temporal area (MT), intraparietal sulcus (IPS), and frontal eye field (FEF) (Gilbert and Li, 2013; Liu et al., 2014; Patel et al., 2015) and frontal cortical areas with the three longitudinal fasciculi filling a significantly larger volume in the right hemisphere (Corbetta and Shulman, 2002). V1 is the initial area in the cortical visual processing system, and feeds to MT, although MT in developing primate also receives other subcortical fast visual-related signals via Pulvinar with projection of signals to IPS and FEF (Mundinano et al., 2019; Kwan et al., 2021). IPS is known to be critically involved in contralateral spatial selective attention (Gillebert et al., 2011) and FEF contribute to fixation, saccade, pursuit, and vergence movements and cognitive processes such as attentional orienting, visual awareness, and perceptual modulation (Vernet et al., 2014; Matsubayashi et al., 2018).

Thus, in the present study, we chose to evaluate groups of strabismic amblyopic (SA) and anisometropic amblyopic (AA) patients and compared them with healthy controls (HCs). Resting-state functional magnetic resonance imaging (rs-fMRI) and high-resolution structural MRI scans were performed with each subject. The blood oxygen level-dependent (BOLD) and structure-related signals in visual attention-related regions of interest (ROIs), including the gray matter volume (GMV), amplitude of low frequency fluctuations (ALFFs), regional homogeneity (ReHo), and FC among the six ROIs. We expected to find functional and structural deficits in the ipsilateral visual attention network in the two amblyopic groups in the resting state given that eye movements and shifts in attention are predominantly driven by the ipsilateral hemisphere (Troost et al., 1972). Such information should progress understanding of visual attention network deficits in those with amblyopia and highlight the hemispheric limitations on neural control of initiation of eye movements and shifts in attention.

This study has evaluated BOLD activation during resting state conditions in the visual attention network ROI including V1, IPS, and FEF in groups with SA and AA and HC demonstrating that activation intensities and resting FC among are deficient in both SA and AA participants compared to HC. We also investigated the structural changes in visual attention-related brain areas and found that the GMV in the V1 was anomalous in SA patients but not in AA patients. Furthermore, zALFF and zReHo measures for the right FEF of SA and right IPS of AA, showed positive correlation with LogMAR cVA of amblyopic eye indicating that poor acuity in the amblyopic eyes correlated with high activation in one area only in each subtype of amblyopia.

In our earlier studies we have shown that the function of both the FEF and IPS areas in SA patients is markedly decreased for goal-driven and stimulus-driven attention tasks, compared with that in HCs (Wang et al., 2017). Under resting-state conditions, we have now found that the left FEF areas that are involved in driving eye movements and hence visual attention to salient stimuli in both SA and AA patients, also showed lower activation intensity and homogeneity than those in HCs, but that the right FEF and IPS showed higher activation intensity and homogeneity than those in HCs presumably because all the amblyopic subjects had experienced abnormal visual experience through the amblyopic eye since infancy and the right brain areas associated with the visual attention network is volumetrically and functionally greater in the right cerebral hemispheres, and is more dominant than that in the left hemisphere (Heilman et al., 1983; Shulman et al., 2010; Thiebaut de Schotten et al., 2011). Therefore, high levels of activation in the FEF and IPS in the right hemisphere could be due to compensatory increases in resting-state situations that enhance the visual attention function of amblyopia patients. The decreased activation intensity and homogeneity in the left FEF was also possibly due to suppression of the attention network by the dominant right cerebral hemisphere and the fact that both the SA and AA groups included 13/20 left eye amblyopes and it is known that eye movements and shifts in attention are predominantly driven by the ipsilateral hemisphere (Troost et al., 1972).

Currently there are a number of studies suggesting that the functional activation in the V1 are abnormal in patients with amblyopia (Jia et al., 2010; Li et al., 2013; Restani and Caleo, 2016), with visual attention modulation and goal-driven attention tasks in SA patients is reduced in V1 (Hou et al., 2016; Wang et al., 2017). There are also suggestions that V1 areas and the corpus callosum connecting them does not mature normally under abnormal visual conditions such as amblyopia (Restani and Caleo, 2016). Similarly in this resting-state study, functional activation in both V1 areas was also defective in SA patients while our VBM results suggests that the GMV in the V1 was increased in SA patients but not AA patients. The AA group results accorded with earlier AA research (Liang et al., 2019), and a higher GMV in SA group often represents unmatured development of the cerebral cortex (Barlow, 1975; Giedd et al., 1999; Sowell et al., 2004). Furthermore, this finding could just reflect that V1 area in SA patients is even more monocular with less binocular connections than in AA and HC patients and hence did not show normal decrease in synapse number early in life (Zhang et al., 2005; Jenks and Shepherd, 2020).

Functional connectivity values among visual attention areas were dramatically perturbed in both SA and AA patients compared with health controls. Thompson et al. (2019) have speculated that the higher-level processing areas of the brain including the striate and extra striate areas are involved in the neural suppression of amblyopia in the visual cortex. We propose that visual attention network deficits associated with eye movement and saccade initiation are the key reason for functional suppression in the visual cortex. In this study, the increased connectivity of the SA and AA groups between the left/right V1 areas and the right FEF may be a monocular indirect compensatory effect for the difference in the temporal activation of eye movements and hence of involuntary bottom-up attention in the ROI in the visual attention network. FC between the right FEF and right IPS was also dramatically increased in both groups of amblyopic patients remembering that the right visual attention network is stronger than that in the left hemisphere (Shulman et al., 2010; Posner, 2012), especially with regard to salience information. Thus, it is possible that the increased connectivity between the right FEF and right IPS is due to the usual suppression of the amblyopic eyes (13 left, 7 right in both groups in this study) within the visual attention network.

In the zALFF and zReHo comparisons, the SA and AA groups showed a high degree of similarity in both the FEF and IPS areas. This finding indicates that the impairment of the visual attention network was similar in the two kinds of amblyopia. However, both V1 areas were exceptions in both groups. As the primary visual cortex, the V1 area receives visual input from both eyes and is the original site of breakdown of spatial frequency processing and binocular function in amblyopia (Crewther and Crewther, 1990; Kiorpes and McKee, 1999; Kiorpes et al., 2003). The different manifestations in V1 in resting condition is likely to be due to a lifetime of abnormal visual experience with the ocular misalignment of a strabismus generating unmatched spatial and temporal visual input from the amblyopic eye, and with the different magnifications in anisometropia generating unclear visual input of larger magnifications from the more amblyopic eye (Levartovsky et al., 1998). A previous study also found that SA resulted in more serious visual deficits and greater suppression from the non-amblyopic eyes than in anisometropic amblyopia (Agrawal et al., 2006; Jia et al., 2010) and SA is usually characterized by greater oculomotor abnormalities than AA. Therefore, it is likely that the SA group showed greater functional and structural anomalies, than did the AA group.

The FEF area and visual attention network also play a key role in eye movement control (Wang et al., 2015; Mahon et al., 2018) so at least for the SA eyes that have long been reported to show eye movement problems such as unstable fixation, saccade delay, and pursuit difficulty (Helveston and Von Noorden, 1967; Asper et al., 2000a; Kanonidou et al., 2014). It is likely that strabismus is associated with considerable eye movement deficits in the development stage (Schor, 1975). This would be expected to more directly affect the development of V1 areas compared with other ROIs because of the close relationship to visual signal processing. This may explain why we found that the GMV of V1 was greater in the SA group than in the AA and HC groups and may also explain the greater eye movement deficits of SA eyes in adult patients.

In correlation analysis, clinical visual acuity of strabismic amblyopic eye, measured as LogMAR cVA showed positive correlation with activation of the right FEF in SA possibly because the right FEF is considered the main brain area to control eye movements and visual attention in normal binocular humans (Silvanto et al., 2006; Thiebaut de Schotten et al., 2011). There is no correlation between the poor cVA of amblyopic eye and V1 structure and function in SA patients. The function of right IPS in AA also showed a positive correlation with the reduced LogMAR cVA of anisometropic amblyopic eye, probably because the less perturbed eye movements function of AA patients did not correlate with FEF disfunction, but the unclear visual signal resulted in abnormal visual transmission in parietal lobe (Felleman and Van Essen, 1991; Friston and Buchel, 2000), and caused the correlation between function of IPS and cVA of amblyopic eye in AA patients.

The major limitations of this study were the fact that the laterality of the amblyopic eyes was not controlled in this study, with 13 Left amblyopic eye and 7 right dominant eyes in both SA and AA groups. This predominance of LE amblyopes with perturbed ocular motility and ability to drive attention would also be expected to result in less activation of the left ipsilateral visual attention network and enhanced dominant right eyes (13/20 for each group) activation of the right hemisphere visual attention network. This laterality bias would also have further confounded the right hemisphere dominance of visual attention. Hence the second limitation of this study was our focus on the key visual attention-related brain areas, rather than function and FC in other brain areas such as visual areas per se. Our choice of ROI was based on prior research relating to the vast oculomotor literature and the association between impaired eye movements, saccadic reaction times and visual attention-related functions in strabismic and anisometropic amblyopes.

Regarding laterality statistical measures were utilized to minimize its effects. Comparison of left SA patients and right SA patients indicated that there was no statistical difference in activation value as measured with zALFF and zReHo in all the six ROIs. The comparison results of Left AA and right AA patients are similar, and the comparison results of left non-dominant eye HC and right non-dominant eye HC area also similar (Supplementary Table 1) and so, the left and right amblyopic eye patients were considered as one group and left and right non-dominated eye HC also. Furthermore the left or right amblyopic eyes were coded as 1 or 2 and made concomitant variables in the statistical process to reduce the effect of the problem while bearing in mind that each eye projects to both hemispheres (Leguire et al., 1990; Zhang et al., 2005).

Future studies of resting state FC would benefit greatly from a more rigorous considerations of the importance of ocular laterality and a more in-depth testing of temporal ocular motility for each participant prior to comparison of GMV and activation densities in strabismic and anisometropic amblyopia. Lastly inclusion of other ROIs along the visual attention networks in particular frontal cortical areas, as sites of top-down visually driven cognition and middle temporal (MT) which is involved in activation of bottom-up visual attention and eye movements.

This study has demonstrated that resting-state visual attention network function was impaired in SA and AA patients though probably more so in the SA participants with worse clinical VA and likely worse ocular motility and least activation of both V1 areas. The right FEF and right IPS areas of both the SA and AA groups showed higher activation than those of the HC group but lower activation than those of the HC group in the left FEF area. Both the V1–right FEF and right IPS–right FEF functional pathways showed higher connectivity in the SA and AA groups than in the HC group suggesting that the 13/20 right eye dominant participants also usually showed enhanced monocular activation of their ipsilateral hemisphere attention networks. GM increased in both V1 areas in the SA group compared with the HC group suggesting less refinement of binocular synapses and connections across the corpus callosum.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Human Ethics Committees of La Trobe University, Melbourne, VIC, Australia and Southwest Hospital, Chongqing, China. The patients/participants provided their written informed consent to participate in this study.

HW: data analysis and manuscript writing. ML and JW: data collection. SC: research guiding and manuscript writing. ZY and TY: data analysis and research guiding. DC: research guiding. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.