PROSTATIC INTRAEPITHELIAL NEOPLASIA - DEVELOPMENT OF A BAYESIAN BELIEF NETWORK FOR DIAGNOSIS AND GRADING

The diagnosis and grading of prostatic intraepithelial neoplasia (PIN) are affected by uncertainties that arise from the fact that almost al l our knowledge of PIN histopathology is not expressed in numeric form but rather in descriptive linguistic terms, words and concepts. A Bay esian belief network (BBN) teas used to reduce the problem of uncertai nty in diagnostic clue assessment while considering the dependencies b etween elements in the reasoning sequence. A shallow network was devel oped with an open-tree topology, with a root node containing the diagn ostic alternatives and seven first-level descendant nodes for the diag nostic clues. One of these nodes was based on tissue architecture and the others on cell features. The results obtained with prototypes of r elative likelihood ratios showed that beliefs for the diagnostic alter natives are very high. The network can grade and differentiate PIN les ions from other prostate lesions with certainty. A number of diagnosti c clues greater than seven did not significantly improve network perfo rmance, whereas a reduced number of clues resulted in decreased belief s. A BBN for PIN diagnosis and grading offers a descriptive classifier that is readily implemented and allows the use of linguistic, fuzzy v ariables. A BBN allows the accumulation of evidence presented by diagn ostic clues, each offering only weak evidence.