I have never had a Scan Dual II, but I have used a Coolscan III and now an LS2000 (Super Coolscan 2000) which are perhaps even a bit older than the Scan Dual II.
The LS2000 is similar to the Scan Dual's specs and I get wonderful results from the LS2000 so in theory I'd have thought the Scan Dual would be good enough; it was certainly well regarded enough in the day.
Have you projected the slides? This may help you see better whether the slides are underexposed. It sounds obvious but Kodachrome is notoriously fickle about underexposure and 1 stop is often more than enough to either make or break a shot. Mostly you've got 1/2 a stop of play at very most in my experience.
I use VueScan and that works well for me for Kodachrome. The IT8 target will help you get accurate colours but not necessarily help with the muddiness of the scan.
I found this review of the SDIII here:
http://www.imaging-resource.com/SCAN/DSEIII/DSEIIIA.HTM - in it they try scanning a very contrasty slide of a train -- there may be some tips of what you can do to boost the shadow detail or reduce the grain in the shadows, I think the SD II and III were largely similar.
If this is the first time you are using Kodachrome (and isn't it just so horrid that this stuff is no longer made!) and scanning, it might just be your exposure needs a bit of work on. You might get away with 1/2 a stop under on projection, but that can make a big difference to a scanner with Kodachrome.
Here's an example:
This:
http://www.flickr.com/photos/lilserenity/4507422736/
Is a shot that is 1/3rd a stop underexposed. It was a high contrast scene anyway, but shooting this accidentally at an effective ISO 80 (I use a Leica MR meter, and to some extent you have to interpret the needle's proximity to other stop markings if it's close to them...) -- the result is a scan that is passable but the shadows have immediately blocked up on a scan. Projected this is passable and there is some shadow detail. However projected, there's a 150W light bulb, and in the scanner 3 LEDs in my case!
This:
http://www.flickr.com/photos/lilserenity/4504101150
Is a properly exposed Coolscan LS2000 scan (left) and 9900F flatbed (right) -- seeing as the Scan Dual II isn't too dissimilar in raw specs to the LS2000 you should get good results with absolutely nailed exposures.
Also I've not IT8 calibrated. Maybe I should, but I got on how the slides are projected and then tweak the scan to that look.
So my consideration here would be the exposure of those slides,
Vicky