The present study examined visuospatial and visuoconstructional abilities across multiple dementia subtypes, highlighting distinct neuropsychological profiles that may inform differential diagnosis and targeted interventions [10,11,12,25]. CBD participants exhibited severe copy and memory impairments, consistent with prior findings linking peri-Rolandic and parietal cortical atrophy, basal ganglia dysfunction, and tau pathology to disrupted visuomotor integration and constructional apraxia [27,33,34]. These deficits likely reflect impaired connections between parietal and motor regions, resulting in disorganized spatial representations, while recognition performance remained relatively preserved, indicating that memory systems were less affected [10,12].

Participants belonging to the FTD-SV group showed copy and memory deficits, which may arise from semantic processing difficulties affecting object recognition rather than pure spatial impairment [22,23,24,25]. FTD-BV patients showed disorganized or fragmented copies alongside recognition deficits, consistent with executive dysfunction stemming from frontal white matter tract abnormalities, which disrupt planning and organization rather than visuospatial perception per se [22,26]. In FTD-NFV, the predominant cognitive deficits are effortful, agrammatic speech production, phonemic paraphasias, impaired syntax, and reduced fluency, with relatively preserved visuospatial recognition and memory, at least early in the disease course [35].

Participants diagnosed with PDD demonstrated deficits in most scores but relatively preserved recognition, aligning with the dual-syndrome hypothesis in which posterior cortical involvement selectively impairs visuospatial processing, while medial temporal memory networks remain largely intact [19,20,21]. These visuoconstructional deficits may also be exacerbated by dopaminergic and cholinergic dysfunction affecting visual attention and discrimination, with variability influenced by disease stage [19,21].

LBD participants showed pronounced memory deficits but comparatively intact copying, reflecting posterior cortical and occipital hypometabolism that primarily disrupts memory consolidation and attentional processes [13,14,15,18,36]. Recognition performance in LBD was unexpectedly similar to controls, potentially influenced by disease stage and compensatory mechanisms such as visual hallucinations, highlighting the heterogeneity of visuospatial and memory deficits in this group [14,15].

The ADD group displayed prominent recognition deficits with relatively preserved copy performance, consistent with medial temporal lobe and hippocampal pathology disrupting memory encoding and retrieval, while early-stage posterior cortical atrophy may have limited effects on figure copying [4,7,8,12]. These findings reinforce the notion that visuospatial deficits in AD progress gradually, becoming more pronounced as posterior cortical regions degenerate, leading to higher-order visual impairments such as simultanagnosia and environmental agnosia [4,6,8].

In CVD, cognitive findings typically include visuospatial impairments, executive dysfunction, slowed processing speed, and difficulties with attention and visuomotor integration, reflecting ischemic injury to subcortical and cortical networks involved in visual-spatial and attentional processing [37]. When CVD coexists with AD, cognitive profiles often reflect a combination of visual–spatial and executive deficits attributable to profound memory and visuoperceptual dysfunction, leading to a broader and more severe pattern of visual and cognitive impairment than seen in either condition alone [7,12,37]. This result mirrors the profile of the ADD group, where prominent deficits in memory and recognition were also evident.

In examining key points and discrepancies, we found that FTD-SV patients scored lower than expected concerning the BCFT total score. This may reflect sample characteristics, age, or disease severity. Conversely, LBD patients scored higher than some of the literature reports [2,4,9,15,24,27,33], potentially due to a milder disease stage or sample selection. On the copy subscale, healthy controls performed best (16), followed by PDD (13), a cluster of FTD-BV, FTD-NFV, ADD, CVD/AD, and LBD showed intermediate performance (12), FTD-SV scored lower than these groups (10), and CBD had the lowest performance overall (8.2). This pattern aligns with the literature. According to review studies [5,10,11,12,17,18], figure copy tasks indicate significant associations with the posterior parieto–temporo–occipital cortex, especially in dementia, underscoring the importance of these regions in encoding and organizing complex visual stimuli into a coherent representation. On the memory subscale, FTD-NFV showed moderate impairment (8), CVD/AD, LBD, CBD, and FTD-SV clustered at low performance levels (3–3.6), and ADD demonstrated the lowest memory scores overall (2.5). Both lesion and functional imaging studies [4,7,8,14,18,25] confirm that effective delayed recall engages networks rooted in temporal and parietal cortices, consistent with the profound memory impairments seen in ADD and related syndromes. For the recognition subscale, FTD-NFV showed mild impairment (0.7), PDD and LBD had moderate deficits (0.5), ADD performed worse (0.4), and FTD-SV, FTD-BV, and CBD showed the lowest recognition performance (0.3). Recognition taps familiarity and recollection processes that are supported by medial temporal structures (including hippocampal and parahippocampal regions) and their interactions with parietal and frontal cortices; these processes are distinct yet overlapping with those in delayed recall (memory subscale), with recognition relying more on preserved encoding (copy subscale) and storage mechanisms [6,8,12,14,18,25].

Overall, these findings largely align with the international literature. Posterior cortical and medial temporal pathology (as in ADD and CBD) is associated with the most pronounced deficits in figure copying, delayed memory, and recognition [4,6,8,10,18,33,34], while frontal and linguistic variants of FTD show relatively preserved total and copy scores but memory and recognition impairments [8,12,24,25,35]. LBD shows notable memory deficits with relatively less severe copy and recognition deficits compared to typical ADD [3,14,15,16,17,18], while CBD manifests pronounced copy and memory difficulties with heterogeneous memory effects reflecting distributed cortical dysfunction [6,27,33,34].

In summary, the distinct cognitive profiles revealed by the copy, memory, and recognition subscales can provide valuable support for differential diagnosis among dementia subtypes. For instance, pronounced deficits in figure copying and delayed memory are indicative of CBD, ADD, CVD/AD, LBD and FTD-SV or posterior cortical involvement, whereas relatively preserved visuospatial construction and recognition combined with intermediate memory impairment point toward frontal or linguistic variants of FTD. LBD and PDD can be distinguished by the presence of prominent recognition deficits with less severe memory impairment, while CBD is characterized by severe constructional difficulties and heterogeneous memory and recognition performance.

The strengths of this study include the large sample size and the use of validated neuropsychological tools. However, certain limitations must be acknowledged. The most significant stems from the considerable disparity in group sizes among the different dementia subtypes. While the total sample size is a strength, this overall number masks a critical imbalance. Our cohort includes a very large group of participants with ADD and a substantial control group, which provides robust data for these groups. However, several key diagnostic groups, crucial for the differential diagnosis focus of our paper, are represented by much smaller sample sizes (specifically the groups for LBD, n = 23; CBD, n = 22; NFV, n = 11) containing fewer than 30 participants. In addition, there is a mismatch between the gender ratio and the age structure of the groups being compared, as seen in Table 1. Counterbalancing of these elements followed the recommendations for multinomial logistic regression designs, as implemented in our statistical approach. Furthermore, the reliance on secondary data introduces constraints, particularly regarding the control of confounding variables such as medication effects and disease progression. Additionally, the study sample consisted exclusively of Greek participants, necessitating further research to confirm the generalizability of these findings to other populations.

Future research should explore longitudinal changes in BCFT performance to establish its predictive validity for cognitive decline, and combine BCFT results with neuroimaging biomarkers to further elucidate the neural substrates underlying these visuospatial and memory deficits [38]. Additionally, larger samples of less prevalent dementias, such as CBD and LBD, will help refine the sensitivity and specificity of BCFT measures across neurodegenerative disorders.