Application Of Diffusion Weighted Imaging In N Staging Of Gastric Cancer

PART ONEOptimization of b value in Diffusion Weighted Imaging for Detection of Gastric Regional Lymph NodesObjectives:The aims of this study are to discuss b value optimization of diffusion weighted imaging (DWI) for T staging evaluation of gastric cancer and evaluate inter-observer agreement of lymph node detection in DWI with different b values.Methods:Twenty healthy volunteers ranging from 20 to 65 years old were included. The conventional MRI and DWI were performed by 1.5T MR system (Signa HDe, GE Healthcare, Medical Systems) using a phased-array eight-channel coil. Single shot spin echo planar imaging combined with array spatial sensitivity encoding technique were applied in DWI at b values of 0 and 50,400,600,800,1000,1200,1500s/mm2. Two radiologists evaluated the location, long diameter, short diameter, clearness of border, ADC value of the lymph node, the standard deviation of noise, SILNs and SIfat. Among them, ADC value of the lymph node, SILNs and SIfat were measured and recorded by radiologists independently. SNR and CNR were calculated, too. Comparisons of clearness of border, SNR, CNR and mean ADC value of the regional lymph nodes in different b value groups were tested by analysis of variance (AVOVA). Inter-observer differences in measurements of SILns and lymph node ADC value were assessed by t test and reproducibility was calculated with Bland-Altman plots.Results:68 regional lymph nodes were included. There were significant differences for clearness of border, SNR, CNR and mean ADC value among different b value groups (p= 0.000). The scores of border clearness in b<1000s/mm2 groups were significant higher than b>1000s/mm2 groups and b=600s/mm2 group got the highest score. SNR, CNR and ADC value declined gradually with the increasing of b values. There were no significant differences of SILNs and ADC value between two observers in the whole sample (p> 0.05), but SILNs in 1200,1500s/mm2 groups and ADC value in b=1500s/mm2 group showed significant differences between two observers. The mean differences and limits of agreement of SILNs and ADC value for inter-observer agreement were-1.8(15.8,-19.3),-0.1×10-4(2.3×10-4,-2.4×10-4)mm2/s,respectively.Conclusions:To find a compromise between accuracy of ADC value and image quality of DWI in detecting gastric regional lymph node, b=600 and 800 s/mm2 were recommended.PART TWODiffusion-Weighted Imaging of Gastric Regional Lymph Nodes in Normal PopulationObjectives:To analyze the differences of number, size, shape, SI and ADC value of gastric lymph nodes among different age groups and regional groups in DWI, as well as establish the DWI baseline of normal gastric regional lymph nodes.Methods:Sixty healthy volunteers ranging from 20 to 65 years old were included and divided into 20 to 40,40 to 65 age group, respectively. The conventional MRI and DWI were performed by 1.5T MR system (Signa HDe, GE, USA) using a phased-array eight-channel coil. SS SE-EPI combined with ASSET were applied in DWI at b values of 0 and 800s/mm2. Two radiologists evaluated long diameter, short diameter, ADC value of the lymph nodes, the standard deviation of noise, SILNs and SIfat. At the same time, short/long diameter ratio, SNR and CNR were calculated. The differences of all outcome measures between two age groups were assessed by independent-sample t test. Comparisons of short diameter, short/long diameter ratio, SNR, CNR and ADC value of the regional lymph nodes among different regional groups in the whole sample and different age groups were tested by AVOVA.Results:Short diameter and ADC value showed significant differences among different regional groups in 20 to 40 age group (p= 0.000,0.004). In 20 to 40 age group, short diameter of No.7,8 and 12 groups were significant bigger than that of No.4 group, especially in No.8, as well as ADC value of No.4d was higher than that of No.4sa,4sb and 10 groups. Besides short diameter, SNR and CNR also showed significant differences among different regional groups in 40 to 65 age group (p<0.01). In 40 to 65 age group, short diameter of No.8 was significant bigger than No.2,4d and 9 to 11 groups, as well as SNR and CNR of No.2 were significant lower than other groups, but SNR and CNR of No.9 was on the opposite. Short/long diameter ratio of 20 to 40 age group was smaller than that of 40 to 65 age group (p= 0.018) and ADC value of 40 to 65 age group was bigger (p= 0.019).There was no significant difference of short diameter, SNR and CNR between two age groups, but to the whole sample, they showed significant differences among different regional groups (p= 0.000,0.003,0.003).Conclusions:The DWI (b=800s/mm2) criteria for normal gastric regional lymph node are defined. Adoption of these criteria may improve the accuracy of DWI for N staging of gastric cancer.PART THREEValue of Diffusion-Weighted Imaging in the Differentiation between Benign and Metastatic Lymph Nodes in Patients with Gastric CancerObjectives:To evaluate the diagnostic value of diffusion-weighted imaging for differentiating benign from metastatic gastric regional lymph nodes and compare it with diagnostic accuracy of morphological criteria.Methods:27 patients were underwent with upper abdominal enhanced CT, conventional MRI and DWI scan. D2 lymphadenectomies were then performed in all patients. SS SE-EPI combined with ASSET were applied in DWI at b values of 0 and 600,800s/mm2. All lymph nodes (long diameter ≥ 5mm) found in DWI were exactly node-by-node matched with the specimens. Using the results of histology as the golden standard, the detection rate of gastric regional lymph nodes in CT and DWI were evaluated. The differences of all outcome measures between benign and metastatic groups were assessed by independent-sample t test. ROC curves were generated to evaluate the capabilities of all outcome measures including ADC value for distinguishing metastatic lymph nodes in the whole sample and each regional sample. The independent predicted factors for metastatic lymph nodes in DWI were selected out by FC Binary Logistic regression model and tested by ROC curve. Comparisons of all outcome measures among different regional groups in benign and metastatic groups were tested by AVOVA. Meanwhile, the differences of all outcome measures among normal group, benign group and metastatic group were tested by AVOVA, too.Results:194 DWI-specimen matched regional lymph nodes were included. The lymph node detection rate of CT was 34.3% and DWI was 64.1%. There were significant differences of enhancement pattern and short diameter between benign and metastatic groups (p= 0.006,0.000). SNR of metastatic group was higher than that of benign group in DWI,80.20 vs.59.15 (b=600s/mm2),63.46 vs.52.18 (b=800s/mm2), respectively. But the results of ADC value (×10-3mm2/s) were on the opposite,1.058 vs.1.290 (b=600s/mm2),0.992 vs.1.300 (b=800s/mm2), respectively. Short diameter in metastatic group was bigger than normal group and benign group. ADC value declined step by step from normal group, benign group to metastatic group.Diagnostic abilities of border, enhancement pattern, short/long diameter ratio and SNR were poor (AUC< 0.7).The AUC of short diameter and ADC value in DWI (b=800s/mm2) was 0.849,0.735, respectively. When a mean ADC value of 1.1×10-3mm2/s was used as a threshold value for differentiating metastatic from benign nodes, sensitivity and specificity were 81.7% and 72.4%, respectively. When a mean short diameter of 5.05mm was used as a threshold value, sensitivity and specificity were 64.8%,71.5%, respectively. Combining ADC value and short diameter, the AUC, sensitivity and specificity could reach 0.925,87.32% and 85.37%, respectively.There were significant differences for short diameter and ADC value among different regional groups in benign group (p= 0.019,0.000) but no significant differences in metastatic group. The AUC of short diameter in all regional groups were over 0.7 except in No.1 and 3 groups. The threshold of short diameter in No.8 (7.6mm) was biggest and that in No.l was smallest (3.8mm). The AUC of ADC value in all regional groups were over 0.7 except in No.11 group. The threshold of ADC value in No.11 (0.965×10-3mm2/s) was lowest and that in No.2 was highest (1.290×10-3mm2/s).Conclusions:Diffusion-weighted imaging is useful in discriminating metastatic nodes in patients with gastric cancer. Combining with short diameter can further improve differential diagnostic ability and provide more valuable imaging information for preoperative N staging of gastric cancer.