Supratentorial glioma grading in children by using apparent diffusion coefficient map: application of histogram analysis based on segmentation.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
OBJECTIVE: Pediatric low-grade gliomas (pLGGs) and pediatric high-grade gliomas (pHGGs) reportedly exhibit variations in outcomes and treatment. This study aimed to investigate whether histogram analysis (HA) of diffusion-weighted imaging (DWI) imaging using a manually segmented three-dimensional (3D) volume of interest (VOI), excluding cystic or necrotic portions, could demonstrate a higher correlation with the tumor characteristics than an entire tumor VOI, therefore could improve the preoperative evaluation of supratentorial pediatric gliomas. MATERIALS AND METHODS: In a current prospective study, 31 pediatric patients with pathologically-confirmed gliomas who underwent baseline DWI were enrolled. The patients were classified by histopathology according to tumoral grade: 15 had pLGGs and 16 had pHGGs. Tumoral VOIs were calculated and transferred to apparent diffusion coefficient (ADC) maps. Histogram analysis (HA) was performed to determine mean, maximum (max), minimum (min), kurtosis, skewness, entropy, standard deviation (SD), mean of positive pixels (MPP), and uniformity of positive pixel (UPP) values for ADC. Comparisons of the values of each ADC parameter within the entire tumoral VOIs (VOI 1) and the tumoral VOIs excluding cystic or necrotic portions (VOI 2) were performed. RESULTS: In VOI 1 placement, ADCmin and rADCmin were significantly lower in pHGG compared with pLGG with p = 0.004 and 0.002, respectively. In VOI 2 placement, ADCmean, ADCmedian, ADCmin, ADC MPP, rADCmean, rADCmedian, rADCmin and rADC MPP were significantly lower in pHGG compared with pLGG with p < 0.001. The HA parameters and the ratio of HA parameters of peritumoral edema were not statistically significant difference between LGG and HGG groups with p < 0.05. In VOI 1 placement, the ratio of ADCmin had higher diagnostic ability (AUC = 0.838) compared with ADCmin (AUC = 0.800). In VOI 2 placement, ADCmin had highest diagnostic ability with AUC = 0.904. The optimal cut-off for ADCmin was 587x10-6 mm2/s with Se of 93.3%, Sp of 81.2% and Youden index of 0.746. Other parameters such as ADCmean, ADCmedian, ADC MPP, rADCmin, rADC mean, rADCmedian and rADC MPP also showed higher AUC values compared with AUC values in VOI 1 placement. CONCLUSION: The HA parameters showed value in grading pediatric gliomas. Excluding cystic and necrotic portions from the entire tumor in measuring HA parameters had higher diagnostic value than the entire tumor VOI method. By using this segmentation, ADCmin had the highest performance in the prediction of histological glioma grading, followed by ADCmean, ADCmedian, ADC MPP, rADCmin, rADC mean, rADCmedian and rADC MPP.