RESECT-AI: Reliable Estimations for Safer Endoscopic Cancer Therapy with Artificial Intelligence

结直肠癌（CRC）在英国所有癌症中发病率第4高，在全球范围内发病率第3高。每年报告140万新病例和70万相关死亡。据估计，到2030年，发病率和死亡人数将增加60%。目前的全球经济影响估计为1000亿英镑。在英国有23万人患有CRC，每年报告41 k新发病例和16 k相关死亡。检测和诊断CRC的金标准是结肠镜检查。在结肠镜检查过程中，大肠的内表面用柔性管上的摄像机检查，称为内窥镜。内窥镜医生可以识别和诊断可能需要治疗的病变。在英国进行了200万次结肠镜检查，在全球范围内进行了6600万次结肠镜检查。这是一个非常常见的临床程序，然而，还有许多未满足的临床需求。RESECT-AI项目将使用机器学习来支持内窥镜手术中的治疗。通过内窥镜图像确定癌症的浸润深度，将患者分为内窥镜切除术或结直肠手术，检测残留组织，预测切除后出血风险。RESECT-AI将降低医疗保健支付者的成本，改善患者预后，改善患者体验。

Colorectal cancer (CRC) has the 4th highest incidence of all cancers in the UK and the 3rd highest worldwide. 1.4m new cases and 0.7m related deaths are reported each year. The incidence and number of deaths is estimated to rise by 60% by 2030\. The current global economic impact is estimated to be £100Bn. There are 230k people living in the UK with CRC, with 41k new incidents and 16k related deaths reported yearly.The gold standard in detection and diagnosis of CRC is colonoscopy. In a colonoscopy procedure the inner surface of the large intestine is inspected with a camera on a flexible tube, called the endoscope. The endoscopist can identify and diagnose lesions, which might require treatment. There are 2m colonoscopies conducted in the UK and 66m worldwide. It is a very common clinical procedure, however, there are a number of unmet clinical needs.The RESECT-AI project will use machine learning to support therapy in endoscopy procedures. The depth of invasion of cancer will be determined from endoscopic images to stratify patients to endoscopic resection or colorectal surgery, the residual tissue will be detected, and the risk of bleeding will be predicted post-resection.RESECT-AI will reduce cost to healthcare payers, improve patient outcomes and improve patient experience.