Unreferenced statistics are of limited value. The utilization percentage is time dependent since while a packet is moving on the network utilization is 100% and while idle 0%. A utilization graph over a time period is more meaningful.
Broadcasts are not too bad. You aren't running with a lot of overhead. One question is how the nodes are set for DNS and Gateways. Specific IP generates lower overhead. Cache time for hosts also can reduce overhead.
Collisions are expected with ethernet since it is a broadcast medium. The number of nodes on the LAN helps determine the frequency of collisions, with fewer nodes resulting in a lower rate with everything else being equal. The pattern of use, illustrated by a utilization time graph, is useful in looking at collisions as well.
The CRC errors are typically wiring related - bit hits due to bad connections or noise. They aren't high enough to indicate a chronic problem.
The question is why are you wondering about the network health? Are you experiencing an issue from the user perspective - slow performance for example - or are you just looking at results from a new tool?