Construction method of gastrointestinal stromal tumor malignant potential classification model based on support vector machine

Abstract

The invention belongs to the fields such as oncology, iconography and machine learning, and relates to a construction method of a gastrointestinal stromal tumor malignant potential classification model based on a support vector machine. The construction method according to the invention comprises the steps of data acquisition, namely performing thin-layer CT image acquisition in a belly reinforcing period; extracting and screening a characteristic; obtaining a punishment parameter and a kernel function parameter through cross validation; establishing a classification model through the parameters; and checking model classification performance. The classification model which is constructed according to the method and is based on the support vector machine can accurately classify the gastrointestinal stromal tumor to two kinds, namely a low malignant potential kind and a high malignant potential kind. Furthermore the classification model is based on an existing image resource and does not cause an additional cost of a patient, thereby facilitating clinical popularization.

Description

technical field [0001] The invention belongs to the fields of oncology, imaging, machine learning and the like, and relates to a method for constructing a support vector machine-based malignant potential classification model of gastrointestinal stromal tumors. Background technique [0002] Gastrointestinal stromal tumors (Gastrointestinal Stromal Tumors, GIST) are a type of tumor originating from the stromal tissue of the gastrointestinal tract, accounting for most of the gastrointestinal mesenchymal tumors. Gastrointestinal stromal tumors have malignant potential, but the classification of their malignant potential has always been a difficult problem for clinicians. [0003] Currently, there are two main classification methods for GIST: modified NIH and AFIP. Both the modified NIH and AFIP include the three parameters of tumor maximum diameter, mitotic count and tumor site, and the modified NIH also includes the parameter of tumor rupture. In terms of clinical operation, ...

AI Technical Summary

Problems solved by technology

Therefore, although the existing GIST classification methods can provide a certain reference value for clinicians' diagnosis and treatment, their application value needs to be improved.

Images