Molecular communication via diffusion (MCvD) is considered as one of the most feasible communication paradigms for nanonetworks, especially for bio-nanonetworks which are usually in water-rich biological environments. Two effects that deteriorates the signal in MCvD are noise and inter-symbol interference (ISI). The expected channel impulse response of MCvD has a long and slow attenuating tail due to molecular diffusion which causes ISI and further limits the slow data rate of MCvD. The extent that ISI and noise are suppressed in an MCvD system determines its effectiveness, especially at a high data rate. Although ISI-suppression approaches have been investigated, most of them are addressed as non-essential parts in other topics, such as signal detection or modulation. Furthermore, most of the state-of-the-art ISI-suppression approaches are performed by subtracting the estimated ISI from the total signal. In this work, we investigate ISI-suppression from a new perspective of filters to filter ISI out without any ISI estimation. The principles for a good design of ISI-suppression filters in MCvD are investigated. Based on the principles, an ISI-suppression filter with good anti-noise capability and an associated signal detection scheme is proposed for MCvD scenarios with both ISI and noise. We compare the proposed scheme with the state-of-the-art ISI-suppression approaches. The result manifests that the proposed ISI-suppression scheme could recover signals deteriorated severely by both ISI and noise, which could not be effectively detected by the state-of-the-art ISI-suppression approaches.