The Value of Whole-Tumor Histogram and Texture Analysis Using Intravoxel Incoherent Motion in Differentiating...

Area of Science:

• Radiology and Medical Imaging
• Oncology
• Biophysics

Background:

• Gastric cancer diagnosis and staging are critical for treatment planning.
• Accurate preoperative differentiation of gastric cancer subtypes is challenging.
• Intravoxel Incoherent Motion (IVIM) MRI offers insights into tissue microstructural and microcirculatory properties.

Purpose of the Study:

• To evaluate the efficacy of whole-tumor histogram and texture analyses of IVIM parameters in differentiating pathologic characteristics of locally advanced gastric cancer.
• To assess the ability of IVIM-derived metrics to distinguish between different tumor grades and Lauren classifications.
• To determine the diagnostic performance of IVIM analysis for preoperative subtyping of gastric cancer.

Main Methods:

• Retrospective analysis of 80 locally advanced gastric cancer patients who underwent MRI with IVIM sequences before surgery.
• Whole-tumor segmentation and extraction of histogram and texture metrics from IVIM and apparent diffusion coefficient (ADC) values.
• Statistical comparison of metrics between different tumor differentiation grades and Lauren classifications using univariate and multivariate analyses.
• Receiver Operating Characteristic (ROC) curve analysis to evaluate diagnostic performance and interobserver agreement.

Main Results:

• IVIM parameters, including ADCmedian, Dslowmedian, and Dslowentropy, showed high diagnostic performance for tumor differentiation (AUCs up to 0.948).
• The perfusion fraction (FP P95th) was the optimal metric for distinguishing diffuse-type from intestinal/mixed gastric cancer (AUC=0.896).
• Dslowmedian demonstrated moderate performance in differentiating signet-ring cell carcinomas (AUC=0.738).
• Good to excellent interobserver agreement was observed for most IVIM parameter measurements.

Conclusions:

• Whole-tumor histogram and texture analyses of IVIM parameters provide a non-invasive method for preoperative differentiation of gastric cancer subtypes.
• IVIM-derived metrics, particularly FP P95th, are valuable for classifying gastric cancer subtypes, outperforming the mono-exponential model for Lauren classification.
• These findings support the use of IVIM MRI in the preoperative assessment of locally advanced gastric cancer.