Quantification of freshly-excised human lymph node tissue using high-frequency ultrasound
Résumé
During lymphadenectomy, 5-20 lymph nodes are excised from cancer patients. These lymph nodes can contain clinically-important metastatic regions that can be missed because conventional histopathology methods do not allow nodes to be examined over their entire volume. In this study, novel quantitative ultrasound (QUS) methods were applied to more than 250 lymph nodes. Individual lymph nodes were scanned in 3D using a 26-MHz ultrasound transducer before histology processing. QUS estimates associated with tissue microstructure were obtained from 3D regions-of-interest. QUS estimates are hypothesized to be different in noncancerous compared to metastatic tissue in lymph nodes. Four QUS estimates were obtained from backscattered spectra and the remaining nine were derived from envelope statistics. Following ultrasound scanning, a 3D histology volume depicting metastatic regions was obtained by performing serial sectioning at 50-µm intervals. Cancer detection based on QUS estimates was performed using linear-discriminant analyses, and areas under ROC curves (AUCs) were estimated. The AUC for the linear combination of four QUS estimates was 0.87 for a dataset of 95 breast-cancer nodes. Similarly, using only two QUS estimates, an AUC of 0.95 was obtained for a dataset of 164 gastrointestinal-cancer nodes. These results suggest that QUS can help guide pathologists towards suspicious regions.
Origine : Fichiers éditeurs autorisés sur une archive ouverte
Loading...