Artifi cial Intelligence-Based Prediction of Breast Cancer Recurrence Using Preoperative Contrast-Enhanced Computed Tomography and Clinical Information

機関リポジトリ 機関リポジトリ (HANDLE) オープンアクセス

書誌事項

公開日
2026-03
資源種別
departmental bulletin paper
公開者
新潟大学医学部保健学科

この論文をさがす

説明

Predicting the recurrence of breast cancer before surgery is crucial for guiding treatment and follow-up protocols. This study aims to develop an artifi cial intelligence (AI) model that predicts postoperative recurrence of breast cancer using only preoperatively available data. These data include contrast-enhanced computed tomography (CECT) images, which are used to detect distant metastases and can be acquired in a short period, as well as clinical information, including needle biopsy results essential for defi nitive diagnosis. This retrospective study included 132 patients with invasive ductal carcinoma. We developed AI models using clinical information alone, CECT images alone, and a combination of both. Classifi cation performance was evaluated using the area under the receiver operating characteristic curve (AUC) with leave-one-patient-out cross-validation. The AI model using the combined data achieved the highest AUC. This fi nding suggests the potential for developing a recurrence prediction method using CECT images and clinical information, which can be collected preoperatively without additional cost or invasiveness. This would make it possible to predict postoperative recurrence before surgery and could help patients and physicians make decisions about postoperative treatment plans.

収録刊行物

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ