In 2020, rectal cancer caused over 339,000 deaths globally, and 732,000 new cases were reported. Historically, Stage II and III tumors, also known as locally advanced rectal cancers (LARC), have been treated with surgical resection, radiation, and chemotherapy. However, advances in neoadjuvant (preoperative) treatment now enable up to 35% of patients to achieve complete tumor death, or complete response, with radiation and chemotherapy alone. In these individuals, surgical resection has shown no benefit and carries the significant risks of major complications, prolonged recovery, and reduced quality of life. Unfortunately, standard clinical testing and radiographic and endoscopic imaging modalities poorly differentiate post-treatment scars from the residual tumor. Confounded by post-treatment fibrotic reaction and edema, the poor performance of current technology makes it extremely difficult to identify complete responders before surgery. Due to this technological gap, surgical resection remains the standard of care (SOC) for all patients outside of specialized tertiary care centers. With improved imaging modalities, widespread adoption of nonoperative management would reduce treatment morbidity for thousands of rectal cancer patients annually. One promising modality, photoacoustic imaging, uses hemoglobin as an endogenous contrast agent to map tissue vascular networks. For clinical use, we have developed and tested a new co-registered acoustic resolution photoacoustic microscopy and ultrasound (AR- PAM/US) endoscopy prototype system, together with a deep learning neural network classifier. Initial testing demonstrated a unique marker of complete tumor response – specifically, recovery of normal mucosal vascular architecture within the treated tumor bed. We hypothesize that our co-registered AR-PAM/US system and the neural net classifiers can assist surgeons to examine the residual tumor microvessel network and assess rectal cancer patients’ pathologic complete response after neoadjuvant treatment. We also hypothesize that serial AR- PAM/US studies will perform significantly better than SOC methods in predicting complete response at treatment conclusion and during post-treatment surveillance. We propose to advance and optimize our prototype AR-PAM/US system, probe, and software and to optimize AR-PAM neural network classifiers to accurately differentiate complete responders from those with residual cancer. We will prospectively assess the ability of co-registered AR-PAM/US technology to improve SOC imaging in a cohort of LARC patients on post-treatment risk management and surgery recommendation. We will also monitor a group of LARC patients to determine if the co-registered AR-PAM/US technology can assess changes in tumor vascular and blood oxygen saturation and identify rectal cancer response, both during the course of treatment as well as in post-treatment surveillance. If successful, this technology will directly reduce the number of unnecessa...