摘要 |
Methods of determining a probability of a suspect cancer nodule being malignant are provided. In one embodiment, the method begins with tabulating histogram data of malignant and benign nodules as a function of biomarker values for a specified patient population suspect of having a specific type of cancer. Next, the tabulated histogram data is separated into a plurality of biomarker bins where the bins are ranges of biomarker values, and malignancy probability fractions are calculated for each biomarker bin by dividing a number of true positives in each marker bin by a summed total of all true and false positives in each bin. Finally, a suspect nodule in a patient is scanned, a biomarker value for the suspect nodule determined, and a malignancy probability for the suspect nodule determined by reference to the tabulated histogram data and the malignancy probability fractions. Other embodiments are also disclosed. |