I don't recommend testdisk because it's not intended for GUID/GPT partition tables,. What you need is a tool to convert your mistakenly MBR partition table (which testdisk can recover under normal circumstances) back to GUID/GPT, using the GPT Fdisk tools that I mentioned. Not sure if []roi[] agrees but please read up a bit more about testdisk before you go even deeper (see what I did ther

- there's a function in testdisk to do a 'deeper search' if quick search fails)
TestDisk is perfect for discovery (analysis), and as the last screenshot reveals, it has discovered what the partition map should look like; which is clearly quite different from the current one.
What's a little strange is the dual HPFS - NTFS partitions, but I suspect that's some latent echo of a previous config + we'd anyway look at contiguous sectors + disk size (which implies we should ignore the 1st HPFS partition & considering his current map also shows the 2nd values + that the Windows partition is working confirms this).
So IMO the valid partition maps values are the 1st, 2nd and 4th one.
To recreate the partition map I'd like you avoid TestDisk, and rather recommend either pdisk or gpt; the article I attached in an earlier post details the pdisk process using the information obtained from TestDisk's analysis.
With regards to the need to do a deeper search; I'm not so sure we need that, as the quick search appears accurate.
What would help is if @grantza could describe what were the partition sizes before the problem, for example: how big was the Mac and Windows partitions: the analysis shows this to be 200Mb for EFI (which btw is the standard size), 284Gb for the Mac partition, and 36Gb for the Windows partition; all totaling to 320Gb.
If this is correct then the next step I'd recommend is recreating the correct partition map with pdisk (and only if that fails, revert to gpt)