基于MRI的人工智能在直肠癌中的应用进展

Application progress of MRI based artificial intelligence in rectal cancer

直肠高分辨率MRI具有极高的软组织分辨率,能够清晰显示直肠肠壁、直肠系膜筋膜、腹膜反折及邻近器官的受侵状况,是评估直肠癌(rectal cancer,RC)的首选影像学方法。然而,常规MRI的语义特征仍然不足以辅助临床医生做出诊疗决策。因此,在RC患者的诊疗随访过程中,需要新型无创影像学标志物来定量描述肿瘤特征,指导临床制订治疗策略,实现个体化诊疗。随着人工智能(artificial intelligence,AI)在医学中的发展及广泛应用,其为基于高分辨MRI的RC评估提供了客观的参考依据,能更好地辅助临床医生做出精准的诊疗决策。本文总结了近年来AI在RC病灶分割、T分期评估、淋巴结转移预测、新辅助治疗后疗效评估、预后预测中的应用,并进行了总结与展望,旨在帮助读者更好地了解基于MRI的AI在RC方面的应用进展,并为将来的研究提供一定参考方向。

High resolution MRI of rectum is the preferred imaging method for evaluating rectal cancer(RC)because of its high soft tissue resolution and its ability to clearly display the rectal wall,mesocenteric fascia,peritoneal reflow and invasion of adjacent organs.However,the semantic features of conventional MRI are still insufficient to assist clinicians in making diagnosis and treatment decisions.Therefore,in the treatment and follow-up process of RC patients,new non-invasive imaging markers are needed to quantitatively describe tumor characteristics,guide clinical development of treatment strategies,and realize individualized diagnosis and treatment.With the development and wide application of artificial intelligence in medicine,it provides an objective reference basis for colorectal cancer evaluation based on high-resolution MRI,which can better assist clinicians to make accurate diagnosis and treatment decisions.This paper summarizes the application of AI in RC lesion segmentation,T stage evaluation,lymph node metastasis prediction,efficacy evaluation after neoadjuvant therapy,and prognosis prediction in recent years,and makes a summary and prospect,aiming to help readers better understand the application progress of MRI-based AI in RC,and provide some reference direction for future research.