Objective To explore the differences in co-altered patterns of gray matter and white matter in patients with mood disorders during depressive episodes with and without suicidal ideation， and investigate imaging biomarkers for predicting suicidal ideation. Methods A total of 48 patients with mood disorders during depressive episodes who were treated at Beijing Anding Hospital， Capital Medical University， between May 2022 and November 2023 were selected for this study. Patients were divided into a suicidal ideation（ SI） group and a non-suicidal ideation（ NSI） group using item 3 of the 17-item Hamilton Depression Scale（ HAMD-17）. Multimodal canonical correlation analysis-joint independent component analysis（ mCCA-jICA） was used to identify joint discriminative independent components and modality-specific discriminative independent components of gray matter and white matter between two groups. Support vector machine（ SVM） was used for differential diagnosis of mood disorders in patients with and without suicidal ideation during depressive episodes. Results Among 48 patients with mood disorders during depressive episodes， there were 26 in the SI group and 22 in the NSI group. There was no statistically significant difference in age， gender， disease duration， and medication use between the two groups（ all P＞0.05）. Patients in the SI group had statistically higher HAMD- 17 total scores and HAMD-17 item 3 scores than those in the NSI group［ （20.27±3.95） vs.（ 15.59±3.45）； 2.00（ 1.00，3.00） vs. 0；t/Z=4.335，-6.289； both P＜ 0.001］. Compared with the NSI group， patients in the SI group exhibited changes in gray matter volume in the left parahippocampal gyrus， left thalamus， bilateral frontal lobes， right temporal lobe， and posterior cerebellar lobe（ t=-2.96， FDR-corrected P=0.029）， and were associated with destroyed fiber integrity in the frontal radiation of the corpus callosum and the left corticospinal tract（ r=0.57，P＜0.001）， and all differences were statistically significant. In addition， statistically significant changes in gray matter volume were observed in the left occipital lobe and right temporal lobe of patients in the SI group（ t=-2.67， FDR-corrected P=0.024）. The best predictive performance for suicidal ideation was achieved by joint discriminative independent components and modality-specific discriminative independent components identified from the gray and white matter of two patient groups， yielding an area under the ROC curve of 0.760. Conclusions This study reveals the co-altered pattern of gray-white matter in patients with mood disorders accompanied by suicidal ideation during depressive episodes by integrating structural imaging and diffusion tensor imaging. Specifically， changes in gray matter volume in the left parahippocampal gyrus， left thalamus， bilateral frontal lobes， right temporal lobe， and posterior cerebellar lobe are associated with destroyed fiber integrity in the frontal radiation of the corpus callosum and the left corticospinal tract. The co-alteration of gray matter and white matter structures may be an imaging biomarker for predicting suicidal ideation.