3D Modeling for Detecting Locally Advanced Rectal Cancer With Positive Circumferential Resection Margin

N/ANot Yet RecruitingNCT07183124

Taichung Veterans General Hospital1,500 enrolled

Overview

This retrospective study aims to develop an AI-assisted 3D modeling system to improve staging accuracy for stage II-III locally advanced rectal cancer (LARC). High-quality CT images from Taichung Veterans General Hospital will be used to reconstruct tumor boundaries and spatial relationships. The AI model will be trained and validated against MRI and pathology results to predict circumferential resection margin (CRM) status. Outcomes include sensitivity, specificity, accuracy, and agreement with standard imaging. This system seeks to support precise tumor staging and inform future clinical decision-making.

Study Type

OBSERVATIONAL

Enrollment

1,500

Conditions

General Surgery Oncology Medical Informatics

Interventions

AI-Assisted 3D Imaging Model for Tumor and CRM Assessmen

This study uses an AI-assisted 3D imaging model to analyze existing CT and MRI images of stage II-III locally advanced rectal cancer patients. The system reconstructs tumor boundaries and spatial relationships, predicts circumferential resection margin (CRM) status, and supports staging assessment. No interventions are performed on participants, and all data are collected retrospectively from routine clinical care.

Eligibility

Sex: ALLMin age: 18 Years

Medical Language ↔ Plain English

Inclusion Criteria: * Diagnosed with rectal cancer, clinical stage II-III, with no distant metastasis (M0) * Age over 18 years, with adequate physical status classified as American Society of Anesthesiologists (ASA) I-III, capable of receiving treatment and surgery * No history of other malignancies or major diseases affecting study assessment within the past three years. * Complete medical records, including available CT and MRI imaging. Exclusion Criteria: * Patients with clinical stage I or IV rectal cancer. * Age under 18 years, or physical status not meeting American Society of Anesthesiologists (ASA) I-III criteria, unable to undergo surgery or related treatment. * Presence of other major diseases or malignancies affecting tumor assessment (e.g., diagnosis of another malignancy within the past three years, uncontrolled cardiovascular disease). * Incomplete medical records or imaging data, including missing required CT or MRI images.

Outcomes

Primary Outcomes

Sensitivity and specificity of the AI-assisted 3D imaging model for predicting circumferential resection margin (CRM) negativity

Model predictions are compared with pathology results (gold standard) to assess diagnostic accuracy.

Time frame: Day 1 (At the time of retrospective imaging analysis)

Secondary Outcomes

Accuracy and agreement of AI model predictions with MRI interpretations

Agreement between AI model, MRI, and pathology results will be analyzed using Kappa statistics to evaluate consistency and reliability.