A self-developed aberration-free line scanning confocal Raman imager (AFLSCRI) with a spectral resolution of 0.12 nm and a spatial resolution of 2 μm is utilized to diagnose colorectal cancer. The tissues were categorized into four subgroups (typical tissue, lipid-rich tissue, fat-rich tissue, and collagen-rich tissue) and were successfully distinguished with our Raman imaging results. Compared to traditional point-scanning Raman spectroscopy, this imager offers a much faster speed with high spectral resolution while maintaining a similar spatial resolution. The Raman spectroscopy results of the same sample of colorectal cancer remain stable and unaffected even measured after six months. The molecular composition of the tissues was analyzed, and potential biomarkers such as carotenoids and protein structures were identified for four different types of colorectal tissues. When combined with machine learning algorithms, an accuracy of 92.8% was achieved in identifying 14 pairs of normal/cancer samples. These results highlight the great potential of the AFLSCRI in label-free, rapid, and non-invasive medical diagnosis.