Proton therapy uses the physical selectivity to produce conformal dose distributions to target tumors with sparing unnecessary dose to healthy organs. To utilize this selectivity efficiently, an image-guided function plays an essential role to places the tumor position precisely and detects the change of patient condition during daily treatment. Recently some proton therapy system vendors have provided CT-image guidance for daily patient positioning due to trend of adaptive radiotherapy. At Fukui Prefectural Hospital Proton Therapy center, we installed a CT scanner in the treatment room and have performed CT image guidance as clinical practice since 2014. We have performed retrospective analyses utilizing full sets of accumulated daily CT-images to study the daily positional variation of the target tumors as well as surrounding organs, and have validated the treatment by evaluating daily dose variation. In this report, first, based on above retrospective analyses, we discuss the importance of not only CT-based registration but also daily dose monitoring of targets for a more reliable and safer proton adaptive treatment. Second, we present the study of AI-based technologies for CT-image guidance proton therapy to achieve more reliable and the cost-effective treatment as a future prospective.