Computer-Guided Surgery for Gastric Carcinoma

Lymphadenectomy offers the only hope for cure when lymph nodes are involved. In gastric cancer, three approaches have been pursued to preoperatively predict node status in individual patients, modern radiological imaging techniques, sentinel node and technique that uses a computerized database of information to convert a large amount of information and experience to a treatment decision for an individual patient. The aim of this study was to evaluate accuracy in preoperative prediction of lymph node status in selected patients with the help of computer analysis for stage-appropriate surgery. With the help of computer programs Win Estimate and Microsoft Access, we constructed an artificial neural network that calculated a statistical prediction of nodal status in an observed patient with preoperatively gathered data. In 110 patients who have undergone R0 resection with D2 lymphadenectomy, the differences between the individual results generated by artificial neural network calculation and the actual data were compared. The accuracy of computerized predictions of N0 stage for study group is 91%, sensitivity 94% and specificity 87%. The results of accuracy of computerized preoperative prediction of N2 stage are 88%, with sensitivity 94% and specificity 88%. Preoperative analyses of patient data and tumour characteristics offers a rational approach to individualizing tumour therapy where the extent of lymph node dissection is tailored to the type, site, and stage of the tumour, thereby minimizing the disadvantages associated with the extensive operative procedure