Image scale is a good *initial* guide. With a 350mm scope you have:
ASI533MC: 206.3*3.7/350 = 2.22"/px undersampled
ASI183MC: 206.3*2.4/350 = 1.414"/px well sampled
ASI294MC: 206.3*4.63/350 = 2.72"/px undersampled
ASI2600MC 206.3*3.7/350 = 2.22"/px undersampled
Undersampling is not fatal--you get square stars, but you can recover by dithering (during acquisition) and drizzling (in processing). You need to take a lot of exposures for good results though.
Other criteria: Field of view (Go to astronomy.tools (imaging mode) and put in your camera and scope F/L, and see how various DSO will frame). The ASI533 and ASI183 are roughly the same size (but one is square and the other rectangular), the ASI294 is larger, and the ASI2600 is largest.
Additionally, the ASI533 and ASI2600 are latest technology with no amp glow (amp glow can be calibrated out, but no amp glow allows scalable darks).
The ASI183 is a 12-bit camera, the ASI533 and ASI294 are 14-bit, and the ASI2600 is 16-bit. The more bits the better the resolution, *but* you can compensate by shooting a *lot* of images with lower-bit cameras. If your total integration time is in multiple hours, it probably doesn't matter (much), if it is smaller then the higher bits have a distinct advantage.
Hope I didn't add to the confusion. If you want a recommendation: The ASI183 is the perfect match for a 350mm scope, but if you have bigger scopes as well, you will like it less. The ASI2600 is a standout in this group, but a pricey one. If you can't afford this, then it becomes a tradeoff between the extra under-sampling of the ASI294 (a minus) and the smaller FOV of the ASI533 (a minus). As the 350mm rates to give me an adequate FOV anyway, I might go with the ASI533, but check out the calculator above and make up your own mind.