Bipolar disorder (BD) is a chronic deteriorating brain condition featuring energy and mood fluctuations that affects approximately 2.4 % population worldwide (Merikangas et al., 2011). BD is characterized by recurrent occurrence of manic or hypomanic episodes, depressive episodes or a combination of manic and depressive symptoms (Grande et al., 2016). Among these, depression represents the most prevalent (Judd et al., 2002, 2003) and disabling symptomatic manifestation and consistently lasts much longer (Tondo et al., 2017). The overall occurrence of BD throughout a person's lifetime is between 3 and 5 % (Angst et al., 2002; Bauer and Pfennig, 2005). Globally, BD holds the 6th position in terms of disability-adjusted life years when compared to other mental disorders (Whiteford et al., 2013).

Around 40–60 % of individuals with BD present neurocognitive impairments, indicating a strong propensity of heterogeneity (Solé et al., 2017). Additionally, cognitive deficits can span multiple domains such as processing speed, working memory, executive functions, attention, and social cognition (Lee et al., 2014; Sheffield et al., 2018). Cognitive abnormalities were consistently observed throughout all stages of BD and reached the peak during acute episodes (Kurtz and Gerraty, 2009), which may be used to predict negative psychosocial outcomes and unfavorable employment outcomes (Baune and Malhi, 2015; Tse et al., 2014). Although neuroimaging research has demonstrated that cognitive impairment may arise from disrupted neuroplasticity and altered functional or structural connectivity within cognition-related neural circuits (Carlson et al., 2006), the precise underlying neurobiological mechanisms remain unclear.

Anomalous brain activities have long been reported in BD in task-based functional magnetic resonance imaging (fMRI) (King et al., 2018; Xiao et al., 2021). For example, untreated young subjects with BD exhibited functional changes in brain regions involved in cognitive processing and the integration of cognitive and emotional processes during a continuous performance test (Li et al., 2022). Another study has used an emotional face n-back task to distinguish BD from Major Depressive Disorder (MDD) (Bertocci et al., 2012). Recently, the resting-state fMRI (rs-fMRI) technique gains more advantages due to the simplicity in signal acquisition, minimal patient effort required (Smitha et al., 2017), and potential to identify novel biomarkers (Vargas et al., 2013). However, the current advances obtained from rs-fMRI are rather controversial, and the most consistent findings were functional and structural abnormalities within the frontal-limbic circuits shown by regional homogeneity (ReHO) (Liu et al., 2012), amplitude of low-frequency fluctuations (ALFF) (Chrobak et al., 2021; Meda et al., 2015) and fractional ALFF (fALFF) (Marino et al., 2021; Yang et al., 2019). Additionally, some authors found increased functional connectivity (FC) within the subcortical (striatal-thalamic) prefrontal networks and limbic system (Anticevic et al., 2015), while others identified reduced FC (Anand et al., 2009; Zhang et al., 2022). Similar impairment was also documented in other different regions including the parietal regions (Liu et al., 2012), fusiform gyrus (Ongür et al., 2010), cerebellum (Sankar et al., 2023) and occipital gyrus (Sun et al., 2022).

Current rs-fMRI studies involved in BD could not specify the phases of BD, or concentrated solely on specific mood state of BD. In some studies, the authors tried to differentiate BD from MDD (Qiu et al., 2021; Zhang et al., 2017), as well as schizophrenia (Lui et al., 2015; Zhang et al., 2021). In several studies, the authors focused on specific phases of BD, such as euthymia and depression (Chrobak et al., 2021; Meda et al., 2015; Romeo et al., 2022; Zhou et al., 2022). Only a few rs-fMRI research studied all three bipolar phases. For instance, a recent study has used dynamical degree centrality (dDC) to explore shared and distinct patterns across BD in depressive, manic, and euthymia states, which mainly found differences in the left inferior parietal lobule/middle occipital gyrus and right precuneus/posterior cingulate cortex (Sun et al., 2022).

Neuroimaging studies have generated a wealth of knowledge about intermediate phenotypes in BD and its subgroups with limitations in understanding specific genetic mechanisms. BD is characterized by high genetic heritability (Kieseppä et al., 2004), and genome-wide association studies have identified multiple risk genetic loci including ion-channel-encoding genes (e.g. CACNB2 and KCNB1) (Mullins et al., 2021) and synaptic components (e.g. RIMS1 and ANK3) (Stahl et al., 2019). Imaging transcriptomics offers the advantage of studying the relationship between gene expression and brain imaging alterations with a relatively small sample number. Furthermore, many genetic neuroimaging association studies have identified genes related to brain imaging phenotypes in psychiatric patients, such as MDD (Zhu et al., 2023). However, to our knowledge, no previous transcription-neuroimaging association research has ever probed the three types of BD to elucidate their heterogeneity.

Until now, there is currently a lack of fMRI studies focusing on all three bipolar phases, highlighting the need to concentrate on the subgroups of patients, which is necessary to separate trait- and state-related neural abnormalities. In the present study, fALFF and FC analyses were applied to investigate the distinct alterations in brain activity of three stages of BD patients and healthy controls (HCs). Additionally, the association between the altered fALFF values and clinical features was also examined. Then a transcription-neuroimaging association analysis was performed to determine different clinical stages of BD and the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were conducted to elucidate the functionally enriched genes and their associations with specific disorders. We hypothesized that there were potential correlations between fALFF changes and gene expression in BD.