Establish a deep learning model based on multi-parameter magnetic resonance imaging to predict the efficacy of neoadjuvant therapy for locally advanced rectal cancer.This study intends to combine DCE with conventional MRI images for DL, establish a multi-parameter MRI model for predicting the efficacy of CRT, and compare it with the DL and non-artificial quantitative MRI diagnostic model constructed by conventional MRI to evaluate the role of DL in MRI predicting CRT. And this study also tries to build a DL platform to assess the efficacy of LARC neoadjuvant radiotherapy and chemotherapy, accurately assess patients' complete respose (pCR) after CRT, and provide an important basis for guiding clinical decision-making.

Inclusion Criteria:

• Clinical suspicion or colonoscopic pathology of rectal cancer
• Age over 18 years
• Informed consent and signed informed consent form

Exclusion Criteria:

• Poor magnetic resonance image quality, such as severe artifacts
• Previous treatment for rectal cancer
• History or combination of other malignant tumours
• Not Locally Advanced Rectal Cancer (LARC)
• Not received neoadjuvant therapy or not completed neoadjuvant therapy
• No surgery
• Time interval between MRI and surgery was more than 2 weeks
• Patients were lost to follow-up and voluntarily withdrew from the study due to adverse reactions or other reasons

Patients with locally advanced rectal cancer (LARC) treated with neoadjuvant therapy and radical surgery