The gold standard for cancer diagnosis is invasive biopsy and subsequent pathologic diagnosis to identify the presence and grade of dysplasia. Unfortunately, most medical practitioners lack expertise to distinguish true precancerous lesions and early cancers from the more common benign confounders. While the potential of optical imaging to delineate areas containing high-grade dysplasia has been demonstrated, previous implementations have lacked the ability to guide provider actions in real time. However, recent advances in optical sensor technology, digital light projectors, and deep learning frameworks provide an opportunity to incorporate real-time interactive clinician guidance with multimodal optical imaging in ways that were not previously possible. The goal of my dissertation research is to develop a multimodal optical imaging system that provides active guidance to clinicians to improve the early detection and complete surgical removal of cancerous lesions. I demonstrated the promise of this approach for two different clinical applications, including: 1) developing a Multimodal Mobile Colposcope (MMC) for interactive biopsy guidance in cervical precancers; and 2) Developing a low-cost Active Biopsy Guidance System (ABGS) for interactive biopsy guidance in oral precancers.