Recent studies have identified microvascular invasion (MVI) as the most vital independent biomarker associated with early tumor recurrence. With advancements in medical technology, several computational methods have been developed to predict preoperative MVI using diverse medical images. These existing methods rely on human experience, attribute selection or clinical trial testing, which is often time-consuming and labor-intensive. Leveraging the advantages of deep learning, this study presents a novel end-to-end algorithm for predicting MVI prior to surgery. We devised a series of data preprocessing strategies to fully extract multi-view features from the data while preserving peritumoral information. Notably, a new multi-branch deep fused feature algorithm based on ResNet (DFFResNet) is introduced, which combines Magnetic Resonance Images (MRI) from different sequences to enhance information complementarity and integration. We conducted prediction experiments on a dataset from the Radiology Department of the First Hospital of Lanzhou University, comprising 117 individuals and seven MRI sequences. The model was trained on 80% of the data using 10-fold cross-validation, and the remaining 20% were used for testing. This evaluation was processed in two cases: CROI, containing samples with a complete region of interest (ROI), and PROI, containing samples with a partial ROI region. The robustness results from repeated experiments at both image and patient levels demonstrate the superior performance and improved generalization of the proposed method compared to alternative models. Our approach yields highly competitive prediction results even when the ROI region outline is incomplete, offering a novel and effective multi-sequence fused strategy for predicting preoperative MVI.

Spatiotemporal raster (STR) data employ an array of grids to represent temporally varying and spatially distributed information, commonly utilized for recording environmental variables and socioeconomic indices. To reveal the geographic patterns embedded in STR data, the clustering by fast search and finding of density peaks (CFSFDP) algorithm is considered effective and suitable. However, this algorithm encounters limitations in identifying cluster centers, handling large data volumes, and measuring the coupled spatial-temporal-attribute distance when applied to STR data. To overcome these challenges, we propose an improved method named spatial temporal-adaptive density peak tree clustering (ST-ADPTC). This method leverages adaptive density peak tree segmentation to identify cluster centers and optimizes memory usage through the k-nearest neighbors (kNN) technique. By constructing a neighborhood that incorporates both spatiotemporal and thematic attribute similarities, ST-ADPTC computes the local density of STR data, facilitating the discovery of time-varying clusters. Based on the proposed method, we develop an open-source Python package (Geo_ADPTC). Experiments conducted using benchmarking datasets illustrate improvements in cluster identification and memory reduction. Additionally, a case study of sea surface temperature data demonstrates the feasibility and effectiveness of exploring spatial and temporal distribution patterns using the proposed method.