Published Results Regarding Performance of Textural Parameters to Characterize Tumors in 18F-FDG PET
| Cancer | Textural index | Resampling scheme (R) | Lesion segmentation | Results |
| Cervical | ||||
| n = 14, pretreatment 18F-FDG PET (2) | Energy; contrast (CM); homogeneity; entropy | Threshold = 0.4 × SUVmax and optimization (CT dimension) | Postradiotherapy outcome (R vs. NR): AUC(Texture) > AUC(MV) > AUC(SUV)* | |
| n = 20, pre + during + posttreatment 18F-FDG PET (7) | Run indices; Zone indices | R = 256 | Threshold = 0.4 × SUVmax and optimization (CT dimension) | Correlation with the changes in tumor heterogeneity during therapy: HGRE, SRHGE, LRHGE, HGZE, SZHGE, LZHGE, SUVmax, SUVmean |
| Head and neck | ||||
| n = 9, pretreatment 18F-FDG PET (2) | Energy; contrast (CM); homogeneity; entropy | R unspecified | Manually segmented | Postradiotherapy outcome (R vs NR): AUC(MV) > AUC(homogeneity) > AUC(contrast) > AUC(SUV) > AUC(entropy) > AUC(energy)* |
| Esophageal | ||||
| n = 41, pretreatment 18F-FDG PET (3) | Homogeneity; correlation; energy; entropy; contrast (CM); dissimilarity; coarseness; contrast (NGLDM); busyness; Run indices; Zone indices | R = 64 | Fuzzy Locally Adaptive Bayesian (26) | Therapy-response prediction (R vs. NR): AUC(texture) > AUC(SUV)* Entropy: NR > R‡ GLNUz: NR > R‡ Homogeneity: NR > R‡ ZLNU: R > NR‡ |
| n = 20, posttreatment 18F-FDG PET (5) | Contrast (CM); correlation | R = 64 | Threshold > 2.5 SUV units | Prediction of pathologic response (R vs. NR): AUC(contrast) > AUC(correlation) > AUC(SUV) > AUC(MV)* Contrast: NR>R‡; correlation: R > NR‡ |
| n = 40, presurgery 18F-FDG PET (8) | Energy; entropy | R unspecified | Threshold > 2.5 SUV units | Correlation with T and N stage: r(entropy) > r(energy) > r(SUVmax)† Prediction of advanced stage ([Ia-IIb] vs. [IIIa-IV]): entropy only |
| NSCLC | ||||
| n = 53, pretreatment 18F-FDG PET (6) | Coarseness; contrast (NGLDM); busyness | R unspecified | Manually segmented | Therapy-response prediction (R vs NR): AUC(contrast) > AUC(coarseness) > AUC(busyness) > AUC(SUV) > AUC(MV)* Coarseness: NR > R‡; contrast: R > NR‡; busyness: R > NR‡ |
| n = 27, pretreatment 18F-FDG PET (4) | Energy; contrast(CM); homogeneity; entropy | R unspecified | Manually segmented | Correlation with locoregional recurrence: r(SUV) > r(texture)† Correlation with local recurrence: r(texture) > r(SUV)† |
| Several types of cancer | ||||
| n = 30, pre- and posttreatment 18F-FDG PET (9) | Homogeneity; entropy; energy; run indices | R unspecified | Automated Random Walk Image Segmentation | Prediction of patient outcome or variation in uptake region characteristics: entropy, SRE |
↵* AUC (X) > AUC (Y): classification obtained with index X was better than with index Y, based on AUC.
↵† Index: R > NR, index was higher in responders (R) than nonresponders (NR).
↵‡ r (X) > r (Y): Spearman rank correlation coefficient (r) with index X was higher than r for index Y.
Run indices: textural index calculated from GLRLM (SRE, LRE, low gray-level run emphasis [LGRE], HGRE, short-run low gray-level emphasis [SRLGE], short-run high gray-level emphasis [SRHGE], long-run low gray-level emphasis [LRLGE], long-run high gray-level emphasis [LRHGE], GLNUr, RLNU, run percentage [RP]). Zone indices: textural index calculated from gray level zone length matrix (short-zone emphasis [SZE], long-zone emphasis [LZE], low gray-level zone emphasis [LGZE], high gray-level zone emphasis [HGZE], short-zone low gray-level emphasis [SZLGE], SZHGE, LZLGE, LZHGE, gray-level nonuniformity for zone [GLNUz], zone length nonuniformity [ZLNU], zone percentage [ZP]).